TABLE OF CONTENTS

1............ Introduction. 5

1.1......... Policy context 5

1.2......... Procedural aspects. 6

1.2.1...... Consultation of stakeholders. 6

1.2.2...... Impact Assessment Steering Group and Impact Assessment Board (IAB) 7

2............ Problems related to the corporate structure of large, complex, and trading-intensive EU banking groups  8

2.1......... Problem driver: The unrestricted co-existence of core banking functions and trading activities within large banks. 10

2.1.1...... Problem 1: Impediments to resolution and supervision. 11

2.1.2...... Problem 2: Distorted incentives for banks. 12

2.2......... Manifestation of problems during the financial crisis. 17

2.3......... Reform efforts to date and complementarity of bank structural reform.. 18

2.4......... How would the situation evolve without EU action (baseline scenario) 22

2.5......... The EU's right to act and justification for acting. 24

3............ Objectives of reforming bank structures. 25

3.1......... General objectives. 25

3.2......... Specific or microeconomic objectives. 25

3.3......... Operational objectives. 25

4............ Reform options. 26

4.1......... Activities to be separated. 26

4.2......... Strength of separation. 29

4.3......... Preventing the risks emanating from shadow banking. 31

4.4......... Structural reform options. 33

4.5......... Issues related to implementation. 34

4.5.1...... Institutional scope. 34

4.5.2...... Role of supervisors. 36

4.5.3...... Timeline for implementation. 36

5............ Impact and comparison of reform options. 37

5.1......... Comparison criteria. 38

5.2......... Social versus private benefits and costs. 38

5.2.1...... Social benefits. 39

5.2.2...... Social costs. 40

5.2.3...... The impact on economic growth of reducing implicit public subsidies. 41

5.3......... Assessment of reform options based on subsidiarisation according to stricter rules. 42

5.3.1...... Social benefits. 43

5.3.2...... Social costs. 51

5.3.3...... Conclusion. 52

5.4......... Assessment of reform options based on ownership separation. 52

5.4.1...... Social benefits. 53

5.4.2...... Social costs. 55

5.4.3...... Conclusion. 56

5.5......... Assessment of retained reform options. 58

5.5.1...... Effectiveness. 58

5.5.2...... Efficiency. 63

5.5.3...... Coherence. 63

5.5.4...... Conclusion. 64

5.6......... Retained options aimed at increasing the transparency of shadow banking. 64

5.7......... Complementary aspects. 65

5.7.1...... Quantitative assessment of some benefits and costs. 65

5.7.2...... International aspects. 67

5.7.3...... Institutional scope. 68

5.8......... Impact on stakeholders. 70

5.8.1...... Impact on bank customers (investors, borrowers, etc.) and market liquidity. 71

5.8.2...... Impact on bank employment 75

5.8.3...... Impact on bank shareholders and unsecured creditors. 76

5.8.4...... Impact on regulators and supervisors. 76

5.8.5...... Impact on EU banking industry. 77

6............ Monitoring and Evaluation. 80

LIST OF ANNEXES

A1. Overview of structural reforms and reform proposals

A2. Summary of replies to the stakeholder consulta­tion

A3. Assessing the complementarity of structural separation with the current reform agenda

A4. Implicit subsidies

A4.1 Implicit subsidies: Drivers, Distortions, and Empirical Evidence

A4.2 Estimating the size and determinants of implicit state guarantee for EU banks

A5. Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements

A6. Qualitative assessment of benefits and costs of separating banking activities from deposit-taking entities

A7. Strength of separation

A8. Trading activities and functional structural separation: possible definitions and calibration of the institutional scope

A9. Summary of the main findings in literature on economies of scale and scope in the banking sector

A10. Quantitative estimation of a part of the costs and benefits of bank structural separation

A11. Impact on private costs – bank responses

A12. Economy-wide impact of structural separation

A13. Shadow banking – Securities finance transactions and transparency

A14. Glossary

1. Introduction 1.1. Policy context

Since the start of the financial crisis, the European Union (the "EU") and its Member States have engaged in a fundamental overhaul of bank regulation and supervision. This exercise has to a large extent been based on the reforms to strengthen global financial markets, agreed upon by global leaders at the G20 summits in London in April 2009 and thereafter and implemented in cooperation with the Financial Stability Board ("FSB") and the Basel Committee of Banking Supervisors ("BCBS").

In the area of banking, the EU has initiated a number of reforms to increase the resilience of banks and to reduce the impact of potential bank failures, the objectives being to create a safer, sounder, more transparent and responsible financial system that works for the economy and society as a whole (see in particular the new Capital Requirement Regulation and Directive ("CRR"/"CRDIV") as well as the proposed Bank Recovery and Resolution Directive ("BRRD").[1]

In line with these objectives several EU Member States (Germany, France, Belgium and the United Kingdom ("UK")) as well as third countries (United States ("U.S.")) have taken a step further and introduced, or are in the process of introducing, structural reforms of their respective banking sectors to address concerns related to financial institutions that are too-big-to-fail ("TBTF")[2]. Structural reform measures have also been suggested in reports published in the Netherlands. See Annex A1 for an overview and summary of the main legislative initiatives.

Also international institutions such as the FSB, the Bank for International Settlements ("BIS"), the International Monetary Fund ("IMF"), and the Organisation for Economic Cooperation and Development ("OECD") have been working on a number of initiatives to improve the resilience of the financial sector and not long ago called for a broad and global debate on bank business models, which includes a review of bank structural measures. This review of structural measures was called for because of the continued growth of TBTF banks in relation to the size of the financial system, as well as because of the adoption or planned adoption of structural measures in several jurisdictions (e.g. separation of activities into different legal entities, intra-group exposure limits, etc.).

In this context, Commissioner Barnier announced in November 2011 the setting up of a High-level Expert Group ("HLEG") with a mandate to assess the need for structural reform of the EU banking sector and, with the objective of establishing a safe, stable and efficient banking system serving the needs of citizens, the EU economy and the Internal Market, to make relevant proposals for further action at EU level. In agreement with President Barroso, he appointed Erkki Liikanen, Governor of the Bank of Finland as chairman of the HLEG.[3]

The HLEG presented its report to the Commission in October 2012 (Liikanen (2012)). It concluded that the existing and on-going regulatory reforms do not address all the underlying problems in the EU banking sector, as these reforms do not fully correct incentives for excessive risk-taking, complexity, interconnectedness and intra-group subsidies. The HLEG stated that reforming the structure of banks is necessary to complement the existing and on-going banking reforms and recommended the mandatory separation of proprietary trading and other high-risk trading activities into a separate legal entity within the banking group. The HLEG envisaged that this separation would be mandatory only for banks where the activities to be separated amounted to a significant share of the bank’s business.[4]

Following a public consultation on the HLEG recommendations, the College of Commissioners debated bank structural reform in early 2013. President Barroso concluded the debate by noting a: “broad consensus in favour of an approach at European level, while stressing that the impact analysis would provide essential clarification.” The President called for an impact assessment to: “examine the various possible options and their implications”.[5]

On 3 July 2013, the European Parliament ("EP") adopted an own initiative report called "Reforming the structure of the EU banking sector"[6] with a large majority. The EP welcomes the Commission’s intention to bring forward a proposal for structural reform to tackle problems arising from banks being TBTF in order to provide greater resilience against potential financial crises, restore trust and confidence in banks, remove risks to public finances and deliver a change in banking culture. The EP calls on the Commission to: (i) provide for a principles-based approach to structural reform of the European banking sector and to that end stresses e.g. the need to reduce risk, complexity and interconnectedness; (ii) ensure the continuity of retail activities; (iii) ensure that trading activities reflect underlying risk and do not benefit from implicit public subsidies; and (iv) ensure that the separated entities have different sources of funding, with no undue or unnecessary shifting of capital and liquidity between these entities and activities. The EP also calls for further measures to strengthen bank governance and promote competition.

1.2. Procedural aspects 1.2.1. Consultation of stakeholders

During the process of considering structural reform of banks, stakeholders have been consulted on a number of occasions. The HLEG met with a variety of stakeholders during its mandate (e.g. different types of banks, bank investors, large corporates, SMEs, retail client associations, supervisors and European and international regulators) and held a public consultation targeting banks, corporate customers and retail clients and their associations.[7] The HLEG received 83 responses, the large majority of which were from banks and other financial institutions, followed by retail customers and their associations and, lastly, corporate customers.  

The Commission also held a public stakeholder consultation after receiving the HLEG final report in October 2012.[8] Out of the 89 replies received, almost half came from the banking industry.

Structural bank reform has also been subject to discussions with Member States in the Financial Services Committee on 14 November, in the Economic and Financial Committee on 23 November and at the ECOFIN meeting on 4 December. The need for a coordinated action at EU level was clearly highlighted.  

As part of preparing this Impact Assessment, the Commission services held an additional public stakeholder consultation during the course of spring 2013 based on a consultation paper. Amongst others, the consultation invited banks to model the impact of different types of structural reforms.[9] Annex A2 summarises the replies.  

The Commission services received 540 replies. These responses came from the expected type of respondents: banks and other financial institutions, corporate clients, investors, public authorities, and consumer associations and individuals. The number of responses from individuals (439) and consumer associations (11) stand out.

The consultation responses highlight a clear distinction between the responses of banks, on the one hand, and consumers and non-bank financials on the other hand. The former are to an overwhelming extent against structural separation (with the exception of some cooperative banks). The latter are largely in favour. The views of other categories are more balanced. Corporate customers, while acknowledging the need to address TBTF, express opposition, based on the potential impact of such reforms on their cost of financing.

Regarding the type and strength of structural measures and what activities to separate diverging views show up again. A large portion of banks express a strong opposition to structural reform or endorse only the plain separation of proprietary trading from deposit taking. Consumer associations and individuals on the other hand argue that separation of proprietary trading and market making activities along the recommendations of the HLEG was the minimum effective, and expressed a preference for either separating all investment banking activities from deposit taking or prohibiting credit institutions from carrying out any investment banking activity.

1.2.2. Impact Assessment Steering Group and Impact Assessment Board (IAB)

An Inter-service Steering Group on bank structural reform was established in March 2013 with representatives from the Directorate Generals COMP, ECFIN, EMPL, ENTR, JUST, MARKT, SG, SJ, TAXUD and the JRC. The Impact Assessment Steering Group met in March 2013, April 2013 and September 2013 and supported the work on the Impact Assessment.

The draft Impact Assessment was submitted to the Impact Assessment Board (the “IAB”) of the Commission  on 19 September 2013 and discussed with the IAB on 16 October 2013.The IAB initially issued a negative opinion and provided its recommendations for improvement on 18 October 2013. The main recommendations were (i) to improve the problem description and baseline scenario, (ii) to better describe and explain the reform options, (iii) to better assess impacts and better demonstrate the effectiveness of the retained reform options, (iv) to better present stakeholder views, and (v) to add a glossary. The Commission Services resubmitted a revised Impact Assessment on 18 December 2013, alongside with a separate document explaining to the IAB how the IAB recommendations had been incorporated. The IAB subsequently issued a positive opinion on 15 January 2014, whilst providing further recommendations for improvement, asking in particular to strengthen the analysis of the justification, alternative reform options, impact, and stakeholder views with respect to the transparency measures that had only been introduced in the resubmission of the impact assessment. The IAB also recommended further strengthening the structural reform options presentation and the assessment of the impact and effectiveness of the retained reform options. The current, final version of the impact assessment has significantly expanded the analysis of the transparency reform measures and has further elaborated on the other two IAB recommendations.

2. Problems related to the corporate structure of large, complex, and trading-intensive EU banking groups

Banks play an important role in channelling funds from savers to borrowers. This intermediation role is particularly important in Europe, as reflected in the absolute and relative size of the EU banking sector compared to those in other major economies (Table 1 and Chart 1).[10]

Table 1: Size of EU, US and Japanese banking sectors (2010)

|| EU || USA || Japan

Total bank sector assets (€ trillion) || 42.9 || 8.6 || 7.1

Total bank sector assets/GDP || 349% || 78% || 174%

Top 10 bank assets (€ trillion) || 15.0 || 4.8 || 3.7

Top 10 bank assets/GDP || 122% || 44% || 91%

Notes: Top 6 banks only for Japan. Source: European Banking Federation (2011).

The EU banking system is also a highly diversified eco-system made up by around 8000 banks that operate according to different business models and different ownership structures. However, over time the market evolved to produce a few very large, complex, interconnected banking groups that offer a diversified set of services such as commercial banking, traditional investment banking, asset and wealth management services, and capital market activities such as market making, brokerage services, securitisation and proprietary trading.[11]

Chart 1: Size of selected EU banks (2012 assets in € billion and as % of national GDP)

Source: SNL Financial (total assets), Eurostat (GDP)

Several of these banking groups have weathered the crisis well, helped by extraordinary and unprecedented sector-wide state support. Without state support (which in some cases is on-going) the EU financial system would have faced a far worse banking crisis (European Commission (2011, 2012)).[12] The (contingent) taxpayer support to date that benefit the EU banking sector amounts to 40% of EU GDP (€5.1 trillion in parliamentary committed aid measures) and has undermined the solidity of several Member States' public finances.[13]

The on-going reforms including the BRRD and Single Resolution Mechanism (“SRM”) will undoubtedly ensure that the vast majority of banks will in the future be fully resolvable. The resolution of the largest and most complex banking groups may nevertheless involve specific challenges that could place a significant burden on both creditors and public the safety net. Moreover, to the extent that market expectations remain that government support may be forthcoming in a future systemic crisis, in particular for the largest and most complex banks, the latter may continue to benefit from an implicit public subsidy. Empirical analyses typically confirm that implicit subsidies exist and in most cases are significant with subsidies amounting to billions of euros annually. For example, JRC estimates that implicit public subsidies enjoyed by the largest European banks that jointly represent 60-70% of EU assets amount to approximately EUR 72-95 billion and EUR 59-82 billion in 2011 and 2012 respectively.[14]

Section 2.1 of this Impact Assessment argues that several of the problems that have materialised in the EU banking sector can be traced back to the unrestricted co-existence of core banking activities with trading activities within large and complex banking groups. This problem driver has contributed to banks growing, becoming TBTF and also too complex to fail ("TCTF") and too interconnected to fail ("TITF"). As further elaborated upon in section 2.2, the financial market activities of banks have contributed to the failure of major banks in Europe. These problems have not been fully addressed by current reforms, as outlined in section 2.3. As a result, section 2.4 highlights that several Member States have filled this gap by pursuing national structural reforms. While these national reforms share the same objectives, they differ in several respects, notably as regards the activities subject to separation and the strength of separation. Such divergences create tensions in the internal market. There is therefore a case for action at the EU level to ensure effective, efficient and coherent reforms (section 2.5).

2.1. Problem driver: The unrestricted co-existence of core banking functions and trading activities within large banks

Large European banks typically combine retail and commercial banking activities and wholesale and investment banking activities in one corporate entity[15] or in a combination of closely connected entities with limited restrictions on transactions between them.[16] The unrestricted co-existence of activities gives rise to the following two main problems, as visually summarised in the “problem tree” in Chart 2, and as discussed below:

· Problem 1: Impediments to effective resolution and supervision;

· Problem 2: Distorted incentives for banks: the implicit public safety net generates moral hazard and leads to excessive trading and balance sheet growth, misallocation of resources, distortions of competition, management and monitoring problems, and the combination of activities within a deposit taking entity can lead to conflicts of interest and flaws in bank culture and standards.

Chart 2: Corporate structure related problem driver, problems and consequences

2.1.1. Problem 1: Impediments to resolution and supervision

Intra-group links arise through, for example intragroup cross-shareholdings, trading operations whereby one group entity deals with or on behalf of another group entity, central management of short-term liquidity within the group, and guarantees and commitments provided to or received from other companies in the group. Amongst others, these economic links are put in place to promote group business activities, to enable the group to operate on an integrated basis across different legal entities, or to ensure competitive financing terms to the entire group (BCBS (2012)).

Intra-group links complicate the resolution and recovery process in the event of failure. They can also impede effective supervision and resolution efforts and increase contagion risk across the group. The financial crisis has highlighted the problems of resolving large banking groups in bad times; the sheer complexity of banks' organizational structures and business models with economic functions and business lines spanning multiple legal entities make it extremely difficult to quickly isolate the problematic and non-viable elements of a banking group. As a result, resolution has up to date been disorganized, involved entire banking groups (as opposed to only the non-viable parts) and has relied significantly on public support. More simplicity in terms of organizational and business structure could therefore lead to easier and more effective supervision and also resolution. Anticipating public intervention in resolution leads to the so-called implicit subsidy which in turns distorts incentives of banks and other relevant stakeholders.

In addition, impediments to resolution can arise from banking groups being highly connected to each other through interbank borrowing and lending and derivatives markets. This is primarily because when highly complex banks are “interconnected” it may not only be difficult for the supervisor/resolution authority to gain insight into the operations of the group but also because it may be difficult to isolate banking groups (or parts of them) that are so connected and resolve them safely and quickly without destabilizing the entire financial system. JRC also finds that banks which are more interconnected are likely to benefit from a higher implicit guarantee (see Annex A4.2).  

2.1.2. Problem 2: Distorted incentives for banks

Deposit taking-banks are by nature exposed to potentially damaging depositor runs. Therefore public safety nets in the form of deposit insurance and lender of last resort facilities (as well as bail-outs) have been introduced. Despite these safety nets, deposit-taking banking groups are currently largely unrestricted in the type of banking activities they undertake and benefit significantly from the explicit and implicit public safety nets.  The public safety nets distort incentives by encouraging banks to take excessive risks which in turn give rise to excessive trading and balance sheet growth, misallocation of resources, distortions of competition, management and monitoring problems, conflicts of interests and culture shocks as well as flaws in banking standards.

Moral hazard, excessive trading and balance sheet growth and misallocation of resources: Deposit-taking banks are by nature exposed to potentially damaging depositor runs. Therefore public safety nets in the form of deposit insurance and lender-of-last-resort facilities have been introduced.[17] However, the public safety nets also have the effect of incentivising banks to expand and take excessive risks with the funds available to them (the so-called "moral hazard" problem) because monitoring and market discipline is muted when risk is not appropriately priced. For example, insured depositors do not lose part of their investment upon a bank’s failure. As a result, they have no incentives (let alone the ability) to monitor the actions of the banks to which they lend. More generally, explicit and implicit public safety nets reduce disciplining incentives of depositors and/or bank creditors and lower a bank’s funding cost. This allows banks to expand and increase their debt issuance and hence to leverage up more quickly, in particular if they are not restricted to relationship-oriented banking activities. High leverage in combination with limited liability incentivises excessive risk-taking by banks even further given the asymmetric payoffs to bank managers and shareholders as upside gains are being privatised, whilst downside losses are being socialised.[18] Even a bank manager that is entirely unaware of the existence of the implicit public subsidy will take advantage of it by assuming high levels of artificially cheap debt and by assuming risky positions that are expected to benefit his shareholders.

Residential real estate bubbles and crashes illustrate that wholesale and investment banking activities do not necessarily have to be more risky than retail and commercial banking activities. However, it is difficult to separate the real estate bubbles from the role of financial innovation, including securitisation of mortgages and derivatives to structure these products for distribution to investors. Moreover, the recent real estate crises, unlike most in history, imperilled sovereigns because sovereigns have been obliged to bail out not only traditional deposit banks, but also banking groups and activities that they should not have had to bail out. Targeting the safety net to those core banking activities that deserve subsidisation and protection because they address a market failure reduces the scope of the public safety net. The nature of the banking activity is what matters the most. The ability to take risks will depend on whether the activity is relationship-oriented or transaction-oriented. Whereas relationship-oriented retail and commercial banking activities are difficult to scale up quickly and easily, market-based and transaction-oriented wholesale and investment banking activities can to the contrary be scaled up more easily.[19] Genuine relationship-oriented banking activities are time-consuming to build and maintain, whereas transaction-oriented banking activities are more deal-oriented and can be replicated more easily.

In addition, JRC analysis points to the possible existence of an incentive for banks towards trading and away from lending activities as a result of the current minimum capital requirements. Such an incentives bias is found not be (fully) eliminated by the Basel III minimum capital requirements reform (see Annex A5).

Bank balance sheets in the EU, particularly those of the largest banking groups, grew significantly in the years leading up to the financial crisis (see charts 3 and 4 below).[20] Much of the balance sheet growth that took place was driven by intra-financial-sector borrowing and lending, rather than real economy lending. For the EU aggregate bank balance sheet, loans to households and non-financial corporations only make up 28% of total assets (March 2012). A significant part of banks' explicit and implicit taxpayer-subsidised pre-crisis activity consists in inter-group financial borrowing and lending.

The increasing dominance of intra-financial business is also reflected in global currency and derivatives markets developments, where large (“broker-dealer”) banks’ trading with non-financial customers (e.g. corporates, governments) has decreased substantially over time both for foreign exchange and interest rate derivatives (BIS (2013)). Recent OECD research raises important concerns about derivative trading giving rise to excessive leverage, interconnectedness, and conflicts of interests (Blundell-Wignall et al. (2012, 2013)). The notional value may not be informative about the riskiness of the derivative positions. Chart 5 Plots the “gross credit exposure” of derivatives positions for the biggest systemically important banks. First, the Gross Market Value (GMV) measures what it would cost to replace all trades at current market prices. It is typically significantly smaller than the notional value. While the notional value of global derivatives was 586 trillion USD in December 2007, the GMV at the same time was only 16 trillion USD. Even when valued at GMV, derivatives will still be important as a proportion of the balance sheet. Second, financial firms have offsetting positions that can be netted and banks expressly hedge most of their positions. The GMV minus netting is the Gross Credit Exposure (GCE). It is against the GCE that collateral is held. It amounted to 3.3 trillion USD in December 2007, against which 2.1 trillion USD was held. The final global open exposure hence amounted to 1.2 trillion USD. Changes in volatility may shift the GMV quickly and netting provides no protection against such shifts in market risks, because netting is about settlement amounts using prices at the point of close out. When the crisis hit in 2008, the GMV more than doubled from 15.8 trillion USD to 35.3 trillion USD, the GCE increased from 3.3 trillion USD to 5 trillion USD and the estimated collateral had to rise from 2,1 trillion USD to 4 trillion USD. Banks faced significant margin calls in a highly risky environment.   

Chart 3: Evolution of liabilities 1998-2012 (euro area, € billion) || Chart 4: Evolution of assets 1998-2012 (euro area, € billion)

||

Notes: Customer deposits are deposits of non-monetary financial institutions excluding general government. Source: ECB data. || Notes: Customer loans are loans to non-monetary financial institutions excluding general government. Source: ECB data.

Chart 5: The gross credit exposure (gross market value minus netting) of derivatives and collateral

Source: BIS, ISDA, OECD

Research also suggests that the deepening of financial markets in the last 10 to 15 years has fundamentally destabilised banks by introducing a trading and fee-based culture in large banking groups. As a percentage of total assets, smaller banks tend to engage more in traditional commercial banking business, resulting in a balance sheet that has more loans (chart 6) and fewer assets held for trading (chart 7) compared to larger banks. These average figures hide significant variations between banking groups, though (chart 8).  

Chart 6: Importance of loan making for EU banks (2011) || Chart 7: Importance of trading activity for EU banks (2011)

||

Source: ECB consolidated banking data. || Source: ECB consolidated banking data.

Chart 8: Assets held for trading, share of total assets (2012)

 

Source: SNL Financial

The implicit public safety net also leads to a misallocation of resources. The expectation of public support artificially increases the size of the financial sector in aggregate and hence diverts resources away from other sectors of the economy. According to academic research, the benefits of more banking activity may not always compensate for increased financial stability risks and other disadvantages. Cecchetti and Kharroubi (2012) empirically find that the enlargement of the financial system beyond a certain size is associated with reductions in real productivity growth. In part this may be due to the financial sector competing with the rest of the economy for scarce resources. Arcand et al. (2012) also find that there can be “too much” finance. When private credit reaches 80% to 100% of GDP (which is largely exceeded for several crisis-affected EU Member States such as Cyprus, Denmark, Ireland, the Netherlands, Portugal, Spain and the UK) further private credit is found to be negatively associated with GDP growth. The hypothesis is that excessively large financial systems may reduce economic growth because of the increased probability of a misallocation of resources, the increased probability of large economic crashes[21], or the endogenous feeding of speculative bubbles. Philippon (2008) observes that outstanding economic growth was achieved in the 1960s with a much smaller financial sector.[22]

Competition distortions: Implicit public subsidies distort competition in the market place and raise barriers to entry to the extent that: (i) small and medium-sized banks are less likely to benefit from such subsidies to the same extent as the large ones and are therefore being disadvantaged; and (ii) weak banks in strong Member States are more likely to enjoy a good support rating[23] and are not disciplined by the market place. These findings have been confirmed by the work undertaken by JRC (see Annex A4.2) as well as other papers (Schich and Lindh (2012) and De Grauwe and Ji (2013)).

Managing and monitoring problems: The financial crisis has highlighted the problems associated with the complexity of banks' businesses and the scope of their operations. There are increasing signs that investors shun this complexity[24] and find it difficult to monitor and therefore understand the underlying risks.[25]

Banks with a variety of activities also require more complex management and are more difficult to regulate.[26] More simplicity in terms of corporate structure could lead to easier and more effective management and regulation. The prudential regulation of banks is also difficult for investors to understand. Accordingly, investors do not fully exercise the “watch-dog” monitoring function granted to them under Basel's “pillar 3” (market discipline).[27]

Conflicts of interests, bank culture shocks and flawed standards: There are significant cultural divides within banking groups. In general, conflicts of interest are more likely to materialize when an institution provides multiple financial services. The main concern is that the bank uses the informational advantage it gains from conducting different activities to its own advantage, instead of seeking to meet the best interest of customers and investors. Whereas regulation and self-regulation aim to address such conflicts of interest, it is difficult to monitor and control the flow of information within large banks. Recent and large-scale events such as the rigging of the Libor benchmark rate – legal fines to date amount to more than USD6bn[28] – and ongoing investigations by several regulators relating to trading on global foreign exchange markets illustrate the limited effectiveness of softer forms of governance separation (e.g. Chinese walls). It is also illustrative of how banks have increased their profits to the detriment of their customers by exploiting proprietary customer information to their own benefit.

2.2. Manifestation of problems during the financial crisis

The problems highlighted in section 2.1 are at the root of the financial crisis. Capital market-based activities have contributed to the failure of major banks in Europe. The majority of the large and complex EU financial institutions that received state support in 2008 and 2009 had trading income to total revenue ratios that were relatively large. For example, having analysed a sample of large and complex EU banking groups, IMF research suggests that almost 80% of all supported banks that received official support in 2008/2009 traded significantly more than average (Chow and Surti (2011)).

Deeper markets have allowed banks to trade more and take larger trading positions with higher associated profits in the good times. However, the higher profitability comes with higher risks, which may compromise bank stability in the bad times. Analysis by the JRC allows disentangling the return on assets (ROA) by type of activity (see also Annex A5). Chart 9 below illustrates the higher profitability and volatility of trading activities.

Chart 9: Average return on assets of European banks as a percentage of total assets by type of activity

Source: SNL and JRC. See Annex A5.

The shift towards a transaction-oriented banking model and the corresponding excessive trading also contributed to boost the size of bank balance sheets. The expansion of bank balance sheets outpaced GDP growth and hence could not be funded by retail funding sources which are more tightly linked to GDP growth and increasingly pushed large banking groups towards short term wholesale funding (repo, money market funds, interbank borrowing, etc.). As a result of their trading activities and increasingly transaction-oriented banking model, banks such as RBS and Allied Irish Bank had significant exposures to structured credit and securitised assets. Also, the capital market activities of TBTF banks effectively enabled other banks to inappropriately rely upon short-term wholesale funding to finance their activities (e.g. Spanish cajas). If Lehman Brothers or any large European bank would have been less connected to deposit taking banks, the impact of their failure would have been less disruptive (see also section 5.8).

As a result, governments would have a smaller incentive to resort to bail out policies, as concerns on contagion and on the continuity of business for deposit banks concerns would be more contained.

2.3. Reform efforts to date and complementarity of bank structural reform

The EU has already initiated a number of reforms to increase the resilience of banks and to reduce the probability and impact of bank failure. These reforms include measures to strengthen banks' solvency (the capital and liquidity requirements part of the CRR/CRDIV package); measures to strengthen bank resolvability (the proposed BRRD); measures to better guarantee deposits (the revision of the Deposit Guarantee Schemes directive (the "DGS"); measures to improve transparency and address the risks of derivatives and to improve market infrastructures (European Market Infrastructure Regulation (the "EMIR") and related revisions to the Markets in Financial Instruments Directive ("MiFID")). Additionally, in order to break the negative feedback cycle between the sovereign and banking risks and to restore confidence in the euro and the banking system, the European Commission has called for further development of a Banking Union, building on the single rule book that will be applicable to all banks in the entire EU. This will include a Single Supervisory Mechanism (“SSM”) and a Single Resolution Mechanism ("SRM), which will be mandatory for members of the euro area but open to voluntary participation for all other Member States.

Despite this broad-ranging reform agenda further measures are needed to reduce the probability and impact of failure of TBTF banks. Such measures have global support, as evidenced by recent statements by G20 leaders and ministers.[29]

As regards the probability of failure, the business of credit intermediation is inherently unstable and prone to liquidity and solvency shocks. Banks are therefore required to put in place adequate shock absorbers, in the form of liquid assets that can be sold without loss to meet unexpected cash outflows ("liquidity buffers") and in the form of sufficient own funds to absorb potential losses ("solvency buffers"). The CRR/CRDIV reform package has increased the quantity and quality of such funds and will thus enable banks to absorb more losses before defaulting.

However, capital requirements are not a panacea and there are limits to what they can achieve. More specifically:

· Addressing remaining TBTF problems by means of higher capital requirements would not address the fundamental inconsistency of on the one hand "taxing" systemic risk and excessive trading with high capital requirements while at the same time allowing these activities to be performed by entities that enjoy explicit coverage of public safety nets;

· Irrespective of the changes to the market risk capital requirements that increase the amount of capital that is required, banks could still have significant incentives for engaging in trading activities given the particularly substantial profits of such activities (see Annex A5). This has induced a broad-based shift towards these activities, at the expense of traditional activities, with an increase in systemic risk being the consequence (Boot and Ratnovski (2012));

· The ratio of risk weighted assets to total assets is significantly lower for TBTF banks, which typically have an important trading book, than for other banks (see chart 10).[30] In addition, the risk-based capital requirements based on value-at risk ("VaR") model calculations can still be small compared to the size of trading assets.[31] Standard setters at both international and European level are currently critically assessing the consistency and accuracy of the risk-weighted asset approach;[32]

· Whereas nominal, non-risk weighted capital buffers could be considered to address these risks, such buffers are blunt and also distort adequate risk-taking incentives. They would have to be set at a high level to fully off-set the remaining incentives in favour of trading. Introducing such additional buffers would also further complicate the prudential framework. This complexity also stems from the increased variety and complexity of bank activities that have required much more complex capital standards (Hoenig and Morris (2011)). These complex standards are difficult to monitor and understand for banks, supervisors, and the market.[33] Structural reform may help to simplify supervision and enforcement of capital requirement regulation;

· Capital requirements also do not address several problems referred to above, such as conflicts of interest and a misalignment between a commercial banking and an investment banking culture within a single “unstructured” banking group;

Chart 10: RWA to total assets: G-SIFI banks versus Non G-SIFI banks

 

Source: Bloomberg, OECD, Blundell-Wignall et al. (2013)

While capital requirements are a major instrument in reducing the probability of failure, it would appear unwise to rely on one instrument only to address the TBTF problem. Structural bank reforms complement the reforms related to capital requirements by imposing direct constraints on specific activities, as opposed to capital requirements that depend on the riskiness of the individual entity and/or of the consolidated group. Structural reform would also be a more direct way of making sure that insured deposits are not used freely throughout integrated groups to fund transaction-oriented activities that are not customer-oriented and hence should not benefit from the implicit government support. It could also complement the systemic risk charges for systemically important banks by adding another disincentive towards banks excessively expanding their risky trading activities, thus putting a break to the main source of unsustainable bank growth in recent years.

As regards impact of failure, implementation of the BRRD will pave the way for the orderly resolution of normal EU banks and thus significantly reduce the impact of failure of such banks on public finances.[34] The resolution powers will be challenging to exercise for TBTF banks, given their particularly large, complex and integrated balance sheets and corporate structures. As a result, while the potential for eventual public support is certainly reduced, it may still not be eradicated if the powers are not in all instances fully applied. The impact of a failure of a large and complex bank may still be significant. Structural reform will increase the options available to authorities when dealing with failing banking groups. By increasing orderly resolution credibility, it will also improve market discipline and bank balance sheet dynamics ex ante.  

The resolution planning offers a vehicle to address potential impediments to resolution. However, it is built on judgement by authorities in individual cases. In the absence of a more clearly structured corporate group structure, it might be extremely difficult for a supervisory authority to exercise its discretionary judgment and impose e.g. a divestment of a part of a large and complex diversified banking group, especially if other competent authorities are not responding with similarly harsh measures in comparable cases.[35] All this may explain market perceptions of remaining implicit subsidies and call for further clarity as regards potential additional structural measures.[36]

Structural reforms could help the orderly resolution of TBTF banks. It could make the newly granted powers in BRRD more effective for TBTF banks, as resolution authorities would deal with separate, segregated and simpler balance sheets. This would make it easier to monitor and assess the different entities of a banking group and it expands the range of options at the disposal of resolution authorities. Additional measures for TBTF banks would be in line with the BRRD’s proportionality principle.

Structural reform can potentially curtail contagion by clearly mapping and controlling intra-financial sector exposures. If left uncontrolled, bail-in may give rise to undue contagion (as bail-in related losses may create losses and distress at other linked financial institutions). As a result, the credibility and effectiveness of the bail-in tool may be hampered.

Ex ante structural reform would also complement the available preventative powers of the BRRD that imply a more institution-specific reorganisation of selected banking groups and which have a narrower resolution objective only. Combining structural reform legislation with the BRRD could over time lead to a greater alignment between business lines and legal structures.

Banking Union is meant to reduce the inappropriate links between sovereigns and their banks. However, by doing so, implicit subsidies and the corresponding problems of moral hazard, aggressive balance sheet expansion, and competition distortions discussed in section 2.1.2 become even more prominent. As a result, Member States may be reluctant to mutualise (future) risks through Banking Union, in the absence of structural reform and credible orderly resolution mechanisms. Targeting the safety net to those core banking activities that deserve subsidisation and protection because they address a market failure reduces the scope of the public safety net and will be a catalyst for the willingness of EU Member States to push ahead with Banking Union.

Chapter 4 develops different forms that structural separation could take. These different reform options are then assessed and compared in chapter 5. That comparison demonstrates in both qualitative and quantitative ways that structural reform has value added in further addressing the problems of TBTF banks. While the exact impact depends on the specific design of the reform option in question (e.g. range of activities to be separated, strength of separation), in general structural reform along the lines outlined in chapter 4 would increase the private cost of engaging in excessive trading activities of primarily intra-financial nature, thus leading to a contraction of such activities as banks adjust to the new reality and hence, other things being equal, a reduction in bank size. It would thus contribute to a better deployment and allocation of resources towards the real economy. It would also facilitate the task of resolution authorities, which in times of stress would imply lower costs related to bank failures. This would benefit the EU economy, as public finances would no longer need to be imperilled to support failed banks. Targeting the safety net to those core banking activities that deserve subsidisation and protection because they address a market failure reduces the scope of the public safety net. At the same time, depending on the scope of activities to be separated and strength of separation, a degree of efficiency might in principle be lost owing to notably reduced economies of scope (see Annex A9). As stated above, the magnitude of these benefits and costs depend on the specific reform option chosen.

2.4. How would the situation evolve without EU action (baseline scenario)

As a response to these concerns, several EU Member States (Germany, France, Belgium and the UK) as well as third countries (U.S.) have introduced or are currently in the process of introducing structural reform measures applying to their respective banking sectors. The reforms in France, Germany, Belgium, the UK and the U.S. all have in common that they prescribe the separation of selected banking activities from a deposit taking entity (for banks above certain thresholds in all countries except the U.S.). Structural reform measures are also under consideration in the Netherlands (Annex A1).

Given the fundamental freedoms set out in the Treaty of the Functioning of the European Union (the "TFEU") divergent national legislation may affect capital movements and establishment decisions of market participants. Under the freedom to provide services, banks authorised in one Member State can freely provide all banking services in other Member States. Under the freedom of establishment, banks can either open a branch or a subsidiary in other Member States. The rights and obligations linked with the two differ. Branches are legally dependent parts of the credit institution. As such, they continue to be supervised as a part of the whole bank by the home Member State supervisory authority. A subsidiary is an independent legal part – and considered as any other legal entity – and becomes subject to supervision in the Member State where it is established, which is thereby considered its 'home'. National structural reforms can accordingly only apply to institutions that are headquartered in that Member State – and their branches in other Member States – as well as locally incorporated subsidiaries of banks from other Member States. Local branches of banks from other Member States are not affected.

This means that under national structural reform, banks operating in the same national market would be subject to different rules depending on whether they are subsidiaries (subject to reform) or branches (not subject to reform).[37] National reforms accordingly run the risk of becoming ineffective, if locally incorporated banks were to relocate and branch back in (for local banks subject to reform) or switch from subsidiary to branch status (for banks from another Member State).[38] However, the effects could also be more pervasive in the sense that banks, rather than relocating by legal means, could relocate in "economic" terms and thus avoid national rules by moving particular activities (by e.g. booking certain transactions in another part of the banking group located in another Member State).

Regulatory arbitrage could over time, and to the extent it became material, compound some of the underlying problems. Divergent national legislation may also undermine efforts to achieve a single rulebook applicable throughout the Internal Market. This is a general problem, as the financial crisis has highlighted that the single financial market does not work optimally if national legislation is significantly different from one country to the other. It can also create specific problems regarding supervision, notably for the future SSM, where the ECB would have to supervise banks subject to different legislation regarding bank structure, thus undermining the establishment of a single rulebook within the EU. Divergent legislation would also make the management of cross-border institutions more difficult and costly, notably in terms of ensuring compliance with divergent and possibly inconsistent rules. It would also lead to safety net distortions, as the DGS of the Member State subject to relocation would face a heavier burden. In sum, if not all Member States address TBTF banks in a roughly consistent way, not all relevant TBTF banks would be subject to reform.[39] Moreover, even those banking groups subject to national reforms would be able to circumvent the rules thanks to the Treaty freedoms, their existing cross-border network of branches and subsidiaries and their right to transfer capital and liquidity across EU borders. Conversely, those arbitrage opportunities would be closed if common rules were to be adopted at EU level. In sum, addressing TBTF banks in an effective manner requires a coordinated EU approach (chart 11).

Chart 11: Potential problems and consequences of uncoordinated national reforms

The default option for this Impact Assessment is to take no policy action as regards structural bank reform at the European level. This represents the baseline against which the incremental impact of structural bank reform options will be evaluated. That baseline includes the non-structural reform elements of the current reform agenda, notably CRR/CRD, BRRD and the first two pillars of the Banking Union (SSM, SRM). The incremental impact will be measured by means of relevant social benefits and costs of the different reform options (see Chapter 5).

2.5. The EU's right to act and justification for acting

In accordance with the principles of subsidiarity and proportionality set out in Article 5 of the TFEU, the Union shall act only if and in so far as the objectives of the proposed action cannot be sufficiently achieved by the Member States, while the content and form of Union action shall not exceed what is necessary to achieve the objectives of the Treaties.

In this case, only EU action can ensure that EU banking groups, many of which operate in several Member States, are regulated by a common framework for structural reform. If the EU does not provide a common framework, divergent national legislation have the potential to distort the Internal Market for the reasons highlighted above. Uniform rules on bank corporate structure are needed to enhance financial stability, facilitate the orderly resolution and recovery of the group, enhance cross-border provision of services and the establishment in other Member States, and prevent regulatory arbitrage.

Common rules on bank structural reform are particularly important for an effective Banking Union. Different national rules would make the task of the ECB in its capacity as a single supervisor difficult, as it would have to supervise banks subject to different, and potentially inconsistent, national rules. Similarly, the future SRM would have to resolve banks subject to potentially different national requirements regarding their organisational and operational structure. The advantage of uniform rules are particularly clear for TBTF banks in Banking Union participating Member States by (1) making supervision easier, by providing for one set of rules that would over time contributing to simpler and more transparent group structures with clearer delineation of business lines; (2) facilitating the task of the Single Resolution Mechanism; and, (3) limiting the risks to be insured by a future EU DGS subject to risk mutualisation (and hence increased implicit subsidies).

The objective of an EU initiative on structural reform would be to adopt uniform measures which have as their objective the establishment and functioning of the Internal Market in financial services. The appropriate legal basis is Article 114 of the TFEU. Article 114 leaves the choice of legal instrument open (either directive or regulation). The advantage of using a regulation is that it would be directly applicable to relevant institutions and create legal certainty as opposed to a directive which would require national implementing legislation that could well be divergent. With a regulation institutions would know when and how the rules apply and be ensured that they apply in a similar fashion to all banks across the EU. A regulation is particularly important for the SSM and SRM for the reasons elaborated upon above.

3. Objectives of reforming bank structures

The objective of bank structure reform is to address the problems and underlying problem drivers highlighted in Chapter 2. To that end, Chapter 3 distinguishes macroeconomic (Section 3.1), specific or microeconomic objectives (Section 3.2), and operational objectives (Section 3.3).

3.1. General objectives

Reducing the risk of systemic instability – reducing the risks of banks becoming or wanting to become TBTF, TCTF and TITF: A key objective of structural reform is to make banks that provide essential services to the real economy more resilient in the event of endogenous or exogenous shocks but also more resolvable in the event of a failure, thus reducing the severity of future financial crises.

Reducing Single Market fragmentation: Many of the banks that will be affected by structural reform legislation operate across borders and seek to benefit from the opportunities created by the single financial market. A common legislative framework on structural reform would prevent the fragmentation of the Internal Market and increase the effectiveness of the future SSM and SRM. 

3.2. Specific or microeconomic objectives

The objectives of structural reform should be to address the problems highlighted in chapter 2. Accordingly, the different options outlined in chapter 4 will be assessed on the extent to which they achieve the following micro-economic objectives:

(1) Facilitate bank resolution and recovery;

(2) Facilitate management, monitoring and supervision;

(3) Reduce moral hazard;

(4) Reduce conflicts of interest, improve bank culture and standards;

(5) Reduce capital and resource misallocation; and

(6) Improve competition.

3.3. Operational objectives

On the basis of the above, the operational objectives of structural reform would be to reduce the magnitude of the problems currently encountered. The following set of quantifiable operational objectives would form the base of future evaluations:

(1) Reduce the size of implicit public subsidies, i.e. reduce the artificial funding cost advantage of TBTF banks (after controlling for bank characteristics such as size, risk-taking, etc.);

(2) Reduce excessive trading by TBTF banks; increase the lending to non-financial customers as a percentage of total assets.

4. Reform options

This chapter develops the policy options that will be subject to further assessment. In particular, sections 4.1, 4.2, 4.4, and 4.5 introduce options regarding (i) the scope of activities to be separated; (ii) the strength of separation; and, (iii) the institutional scope of separation, and considers the timeline for implementation. As each of the examined options may lead to a situation whereby certain banking activities may migrate away from regulated banking groups towards “shadow banks”, the structural reform options must necessarily be accompanied by measures improving transparency and data reporting of the shadow banking activities (section 4.3).

4.1. Activities to be separated

The Single Market has brought significant benefits to EU Member States. It contributes to solid economic growth and supports employment (ECB (2012)). Integration in the markets for banking and other financial services is one key element of the Single Market. Among other benefits, financial integration has contributed to the convergence of and decline in financing costs for corporations and households and the opening up of investment and diversification opportunities across Europe.

Financial integration in Europe had progressed significantly in the years prior to the crisis, in particular in the wholesale markets. The adoption of the euro and, shortly afterwards, the Financial Services Action Plan were major milestones in this integration process. 

Financial institutions and banks in particular have adapted to this new economic reality by expanding in scale, reach and scope of activities. This has led to the emergence of large financial "one-stop shops" combining within one entity or group the provision of a diverse set of services. Moreover, in their quest for economies of scale and scope, banks have also consolidated; first within national borders, then beyond. This process has allowed banks both to provide a broader range of services to their clients, as well as serving clients operating across borders. This process was particularly pronounced in the EU, given the Single Market and enshrined treaty freedoms. This has enabled banks to respond to the increasingly sophisticated needs of their global clients. The financial crisis has clearly illustrated the impact on financial stability arising from an ever more global and integrated financial system with ever larger units of financial service providers.

As a result, the banking activities undertaken by large EU banking groups today range from retail and commercial banking (RCB) activities to wholesale and investment banking (WIB) activities. Examples of RCB activities include, amongst others, insured deposit taking, lending to households and SMEs, and the provision of payment system services. Examples of WIB include, amongst others, underwriting, market making, brokerage services, and proprietary trading.

Notwithstanding the benefits highlighted above, the financial crisis has clearly illustrated the impact on financial stability arising from an ever more global and integrated financial system with ever larger units of financial service providers. Therefore, the basic rationale of structural reform is to separate certain risky trading activities in order to facilitate the resolvability of banks and to ensure that these activities do not endanger bank activities that are regarded as critically important for the real economy. In the national structural reform efforts to date, the separation has been applied at different “locations” between and within the range of RCB and WIB activities. Depending on the strength of separation and the scope of targeted activities, separation leads to different degrees of restrictions on some banks’ ability to provide certain services.

Accordingly, concrete options to separate banking activities logically end up between, at one end of the spectrum, a narrow trading entity and a correspondingly broad deposit-taking entity and, at the other end, a broad trading entity and a correspondingly narrow deposit entity. In the first case, relatively few activities are being separated from the deposit entity (funded by guaranteed deposits in contrast to the trading entity), and it accordingly remains relatively free to provide a broad set of trading and capital market activities. In the latter case, a much broader set of activities is separated, and the deposit entity is accordingly much more constrained in the activities it can engage in.

· “Narrow” trading entity and “broad” deposit entity: A first polar case is that in which only relatively few trading activities are being separated from a broad deposit entity, namely those types of trading activities where traders are speculating on markets using the bank’s capital and borrowed money, for no purpose other than to make a profit and without any connection to trading on behalf of customers (i.e., proprietary trading). Proprietary trading is the purchase and sale of financial instruments for own account with the intent to profit from subsequent price changes. The importance of dedicated proprietary trading desks has decreased over time and currently appears to be of relatively limited importance for many large EU banking groups. Internal hedge funds are similar in spirit. This would also involve trading in physical commodities.[40] In this polar case, the set of activities to be separated from the deposit entity would roughly correspond to the French banking law and the German law (where the separation largely takes place within the group), as well as the Volcker rule in the US (where the separation amounts to a prohibition and a banning of the activity from the group altogether). See Annex A1.

· “Medium” trading entity and “medium” deposit entity: A second case is one in which market making (and possibly more) is added to the above set of activities to be separated from the deposit entity. Separating customer-related market-making, proprietary trading and selected similar activities is roughly aligned with the proposals made by the HLEG.

In general terms, market making is the purchase and sale of financial instruments (government bonds, corporate bonds, equities, derivatives, etc.) for own account at prices defined by the market maker, on the basis of a commitment to provide market liquidity on a regular and on-going basis. Market makers provide "immediacy" to clients and investors by facilitating their requests to buy and sell quickly and, arguably, in a cost-effective way for them. For example, an investor anxious to sell an asset relies on a market maker's standing ability to buy the asset for itself immediately. Likewise, an investor who wishes to buy an asset often can call on a market maker to sell the asset out of its inventory. By doing so, market makers can instil greater investor confidence in the functioning of financial markets and encourage investors to trade confidently. Market making makes up a significant part of large banking groups' trading revenues. Without market makers, customers would face higher transaction costs, and security prices would be more volatile. However, from a legal and economic point of view, market making (and the securities inventory used to facilitate customer trading) is difficult to distinguish from proprietary trading, in particular for “outsiders”. A market maker acquires a position at one price and then lays off the position over time at an uncertain average price by providing liquidity to customers. The ultimate goal is to "buy low, sell high". In order to accomplish this goal on average over many trades, with an acceptable level of risk for the expected profit, a market maker relies on its expectation of investors’ needs and the future path of market prices. Although traders involved in the actual trade are able to identify any given transaction as being of a market making or proprietary trading nature, such a distinction no longer is simple from the perspective of an outsider such as a manager, regulator, supervisor, creditor, or judge. Indeed, a market maker might legitimately choose to take a long position in an asset either in anticipation of client demand to allow the order to be fulfilled quickly or to facilitate a quick sale by a client of an illiquid asset.

While it is possible for institutions other than banks (such as funds) to take on a similar role to market makers, banks do have a natural advantage in acting as market makers because of the fact that banks have a variety of other relationships with the clients who want to make trades and the fact that acting as a market maker for a security is often a natural follow-on activity for securities underwritten by the banking group.

The most active market makers in financial markets today are high frequency traders, many of whom trade as voluntary market makers with no obligations to maintain markets. According to several academic studies, high frequency market making is a profitable enterprise and, more importantly, market quality has improved alongside the growth in algorithmic trading. These results are frequently interpreted as support for a structure where participants supply liquidity because it is a profitable and viable activity on its own (see Anand and Vankatamaran (2013) for a more in-depth analysis). Several important market makers are not taking any deposits, suggesting that market making is a viable activity on its own.

· “Broad” trading entity and “narrow” deposit entity: This case corresponds to a relatively broad range of activities being separated from the deposit entity and is one in which all wholesale and investment banking activities are to be separated. In this stylised option, trading entities would perform activities such as underwriting[41], advisory services, brokerage services, derivatives transactions, investing, sponsoring and structuring activities related to certain securitisation activity[42], in addition to proprietary trading and market making (See Annex A6 for a more elaborate description and assessment of these additional WIB activities). Separating a broad range of investment banking activities will result from the UK structural reform. The US also organises banking groups as bank holding companies that perform core banking activities and other banking activities through different affiliates.

In determining which activities should be subject to separation, the Commission services have considered: (i) the extent to which losses related to an activity would impact a bank’s balance sheet; (ii) the extent to which an activity gives rise to market or counterparty risk; (iii) the importance and potential impact of the activity on systemic risk; (iv) the customer-oriented nature and usefulness of an activity for financing the real economy, and (v) the extent to which the banking activity resolves a market failure (such as asymmetric information) in the economy. The application of these criteria leaves a narrow range of wholesale and investment banking activities that require further analysis, e.g. proprietary trading, market making, underwriting, investing, sponsoring and structuring activities related to certain securitisation activity, and derivatives transactions. The focus on wholesale and investment bank activity is consistent with the significant increase in trading and market-based activities, documented in Chapter 2.

There are certain activities, however, that are critically important for the continuity of a banking group (i.e. the taking of deposits and provision of retail payment services). Those activities would under any of the options outlined above always reside with the deposit-taking entity.

4.2. Strength of separation

When determining the strength of separation, a starting point is to consider three broad forms of separation: (i) accounting separation; (ii) subsidiarisation; and (iii) ownership separation, i.e. prohibition of certain business lines. These forms of separation display a varying degree of severity and intrusiveness in banks’ business structure. They are not mutually exclusive but build on each other. For example, subsidiarisation presupposes a degree of accounting separation. Furthermore, subsidiarisation can coexist with prohibition (ownership separation) of certain activities.

(a) Accounting separation: The lightest degree of intervention is accounting separation. This would require banks that provide integrated financial services to make separate reports for their different business units and make them publicly available. A certain degree of accounting separation already exists in the EU. Accounting separation would increase transparency, as it would lead to banks having to put more information in the public domain, thus in theory facilitating market monitoring and supervision. However, it would not significantly affect economic incentives. For example, it would not impose any restrictions on intra-group legal and economic links, and it is thus unlikely to be sufficient to address the major policy objective of this exercise. Accordingly, it would not contribute to addressing the TBTF problem. Even so, while insufficient to address the main objectives, a degree of stronger separation of accounts by business lines would, by construction, follow from – and be necessary for – more ambitious forms of separation.

(b) Subsidiarisation: A second form of separation is to require subsidiarisation, i.e. to require banking groups to separate the activities of different business units into separate legal entities ("subsidiarisation"). Subsidiarisation could also require these entities to maintain separate capital structures, i.e. for the business units to become “ring-fenced” subsidiaries with their own capital and funding and with rules on how the different subsidiaries deal with each other (e.g. limits on intra-group capital flows and arm's length pricing).

Importantly, under subsidiarisation, universal banking groups could continue to provide their clients with a diverse set of banking services. However, they would have to structure their group differently with some of the services provided in legal entities separated from the other parts of the group (“structured universal banking”).

Subsidiarisation can take many forms. Under subsidiarisation, there will still be links of legal, economic, operational and governance nature between the banking group and the functionally separate legal entities. As a result, choices need to be made as regards the degree of independence of the subsidiary. More specifically:

· As regards legal separation, one could limit oneself to requiring the setting-up of a separate legal entity to which the relevant activities could be separated for banks that fall under the institutional scope of the regulation. Given the activities under consideration, that entity would most likely be an investment firm. Or, one could consider options that would provide a stronger degree of legal separation by governing also the ownership links between this new entity and the rest of the group. For example, one could consider rules limiting certain parent-subsidiary ownership structures (e.g. whereas the group could continue engaging in a universal set of activities, the trading entity of the group could not own the group’s deposit bank); or rules prohibiting direct ownership links between deposit banks and trading entities, thus calling for a group holding company on top;

· As regards economic separation, the separate legal entity would normally have to respect the CRD/CRR requirements related to capital, liquidity, leverage, and large exposures on an individual basis. To provide a stronger degree of economic separation, further rules could be considered for the relations between the separated entities and other group entities, such that intra-group transactions could be on arm's length basis; whether certain restrictions on exposures, funding pattern and activities should be introduced to further isolate the deposit taking entity from (international and financial-system originated) shocks. In addition, one could consider whether the subsidiary would have to raise funds separately;

· As regards governance separation, one could consider the degree of independence of the board of the separated entities (e.g. degree to which directors should be independent from rest of group), as well as whether or not the separated entities should have their own risk management structures even if currently reserved for “significant” institutions by CRD. Also one could require the management body to uphold the objectives of structural reform;

· As regards operational separation, one could consider the degree to which infrastructure related to payment systems and IT and data could be shared among group entities, or whether it would also need to be separated.

The reforms pursued to date have chosen different combinations along these separation dimensions, even though there is a large degree of complementarity (see Annex A1).

(c) Ownership separation: The most intrusive degree of structural intervention is ownership separation. Under this form of separation, there would be separate ultimate ownership of assets supporting different activities. Accordingly, those services would have to be provided by different firms with different owners that are not in any way affiliated with each other.

This was the approach followed by the 1933 Glass-Steagall Act.[43] The justifications for this strong type of separation were to: (i) prevent inherent conflicts of interest; (ii) reduce the financial power of depository institutions; (iii) reduce depository institutions' ability to engage in risky securities activities; and (iv) prevent managers of depository institutions – focused on prudence – to enter markets that are focused on risk-taking.

For the reasons stated under point A) above, accounting separation will not be subjected to further assessment in this Impact Assessment.

Three degrees of separation will be subject to consideration in the remainder of this Impact Assessment, two based on different forms of subsidiarisation (given the wide range of specific subsidiarisation rules) and one based on ownership separation. A first separation option would contain a limited degree of subsidiarisation in legal and economic terms. It would require the creation of a separate legal entity, but would limit itself to the degree of economic and governance separation that currently follows automatically from such an obligation. A second “moderate separation” option would include an additional, stricter degree of legal, economic, and governance separation, whereas a final “complete separation” option would be equivalent to full ownership separation, as described above, i.e. effective prohibition of certain activities. More specifically:

(1) Subsidiarisation with intra-group links restricted according to current rules (“SUB”): in terms of legal separation, this option would require a separate legal entity. In terms of economic separation, it would restrict itself to the economic and governance requirements that currently result from this degree of legal separation. That entity should be subject to the CRD/CRR prudential requirements in terms of capital, liquidity, leverage and large exposures on an individual basis. This would result in a degree of economic and governance separation, depending on whether these requirements would be waived or not;

(2) Subsidiarisation with tighter restrictions on intra-group links (“SUB+”): in order to more effectively address intra-group funding subsidies, this option would require a more significant degree of subsidiarisation in legal and/or economic terms. In terms of legal separation, it may include rules on ownership links between separated entities within the group. This option may also require a stricter degree of economic separation, such as separate funding for the two entities. It also includes stricter economic separation, notably by considering rules on intra-group relations (e.g. requirements that intra-group transactions be on third party, commercial terms; ensuring that current large exposure restrictions are not waived and possibly apply stricter requirements; providing limits on intra-group guarantees (deposit-taking entity would not support trading entity); and, stipulating a higher degree of governance separation (e.g. limits on cross-use of board directors within the group); or

(3) Ownership separation: under this option, banking groups would not be allowed to engage in certain activities. They would accordingly have to divest or wind down any such activities that they currently engage in.

4.3. Preventing the risks emanating from shadow banking

As each of the examined options may lead to a situation whereby certain banking activities may migrate away from regulated banking groups towards shadow banks where there may be less scope for control by supervisors (whether or not located with the EU). To that end, each of these options must necessarily be accompanied by measures improving transparency and data reporting of the shadow banking activities.

The work done by the FSB has highlighted that the disorderly failure of shadow banking entities can carry systemic risk, both directly and through their interconnectedness, with the regular banking system. The FSB has also suggested that, as long as such entities remain subject to a lower level of regulation and supervision than the rest of the financial sector, reinforced banking regulation could drive a part of banking activities beyond the boundaries of traditional banking and towards shadow banking. The European Commission has recently adopted a Communication setting-out a roadmap for tackling the risks inherent in shadow banking. In line with the 2013 FSB Recommendations endorsed at the St-Petersburg G20 Summit, the measures foreseen in this roadmap include, among others, measures aimed at strengthening transparency and data availability in the shadow banking area.

Because of its size and close links to the regular banking sector, the shadow banking sector poses a systemic risk. The first factor is size. The latest studies indicate that the aggregate shadow banking assets are about half the size of the regulated banking system. Despite the fact that shadow banking assets have decreased slightly since 2008, the global figure at the end of 2012 was €53 trillion[44]. In terms of geographical distribution, the biggest share is concentrated in the United States (around €19.3 trillion) and in Europe (Eurozone with €16.3 trillion and the United Kingdom with around €6.7 trillion). The second factor which increases risks is the high level of interconnectedness between the shadow banking system and the regulated sector, particularly the regulated banking system. Any weakness that is mismanaged or the destabilisation of an important factor in the shadow banking system could trigger a wave of contagion that would affect the sectors subject to the highest prudential standards.

To prevent that banks shift parts of their activity in the less regulated shadow banking sector, it is necessary to ensure that any structural separation measure be accompanied by measures improving the transparency of shadow banking. Due to their size and close links with the banking sector securities financing transactions (SFTs) such as repurchase agreements, securities lending, other equivalent financing structures and re-hypothecation are a particularly relevant issue to address. SFTs display structural similarities with banking activities as they can lead to maturity and liquidity transformation and increased leverage, including short-term financing of longer-term assets.

As explained in detail in Annex A13, when assessing the transparency of the SFTs markets and rehypothecation, three main themes emerge: (1) the monitoring of the build-up of systemic risks related to SFT transactions in the financial system; (2) the disclosure of the information on such transactions to the investors whose assets are employed in these transactions; and (3) reducing the uncertainty about the extent to which assets have been rehypothecated. EU regulatory authorities lack the necessary data to better monitor the use of SFTs and the risks and the vulnerabilities for the stability of the financial system that they imply. At the same time the investors are not properly informed whether and to what extent the investment fund, in which they have invested or plan to invest in, has encumbered or intends to encumber investment assets by means of engaging in SFTs and other financing structures that would create additional risks for the investors. Finally, there is a lack of transparency about the extent to which clients' and counterparties' assets can be rehypothecated, or about the risks posed by rehypothecation.

Because the risks resulting from a migration from the banking sector toward the shadow banking sector are not directly related to the decision that is retained as regards the degree of the structural separation, the options aimed at increasing the transparency of the shadow banking sector will be assessed separately. The assessment of the impacts for the structural separation is not linked to the assessment of the impacts for the shadow banking transparency, and inversely. However every decision that is taken on the structural separation aspect will have an impact on the shadow banking sector, therefore Annex A13 analyses the potential response to that threat for the financial stability.

The following section presents the nine reform options for the structural separation whereas the Annex A13 presents and assesses the potential options aimed at increasing the transparency of SFTs. For each of the three main themes that have been identified, a series of policy options have been identified and assessed. The reform options on the structural separation are aimed at credit institutions whereas the options on shadow banking are aimed at each financial counterparty that performs SFT activity. This difference in scope is necessary because the entities acting in the shadow banking universe are diverse in nature. A targeted approach would not have achieved a level playing field between all actors.

These elements will notably allow supervisors to better identify interconnectedness between deposit taking entities and some shadow banking entities and will shed more light on some of their funding operations. It will also allow both market participants and supervisors to better evaluate possible risk exposures of shadow baking entities, leading to better-informed investment decisions by the former and better-targeted and timely actions by the latter. Annex A13 provides for details about the impact of introducing more transparency in securities financing transactions.

4.4. Structural reform options

The combination of the different range of trading and capital markets activities to be separated and strengths of separation (see sections 4.1 and 4.2) yields nine stylised reform options, as visualised in matrix form in Table 2. In the remainder of this Impact Assessment, we will refer to the different reform options, according to the labels in Table 2 (reform option A, reform option B, etc.). To the extent possible and for illustration purposes only, the Commission services have mapped the different national and expert group structural reform proposals and/or legislative initiatives into the matrix of stylised reform options. See also Annex A1 for a more elaborate discussion of these reform initiatives.

Table 2: Overview of options

Activities strength || Functional separation 1 (SUB) Current requirements || Functional separation 2 (SUB+) Stricter requirements || Ownership separation Ownership separation

Narrow trading entity/ broad deposit entity E.g. Proprietary trading  + exposures to HF (PT ) || Option A || Option B [≈ FR, DE baseline, BE] || Option C [≈ US Volcker]

Medium trading entity/ medium deposit entity E.g. PT + market-making (MM) || Option D || Option E [≈ HLEG; ≈ FR, DE if wider separation activated] || Option F

Broad trading entity/ narrow deposit entity E.g. all investment banking activities || Option G || Option H [≈ US BHC; ≈ UK] || Option I [≈ Glass-Steagall]

The reform options listed in the first column (A, D and G) all have in common that they require setting up separate subsidiaries for trading and deposit entities (“subsidiarisation” requirement) without strengthening the regulatory rules and requirements that are currently in place. The effectiveness of all three of these options thus depends on whether the social benefits are effectively achieved with a separation according to current rules. However, under current rules and regulations, some of the above prudential requirements can be waived in certain circumstances, notably upon approval from the domestic supervisor. Moreover, the current regulatory framework does not regulate legal, economic and governance links and interconnections between separate entities within a group to a significant extent. For example, large exposure rules for intra-group entities are optional (can be waived) and Member States have discretion on their level. Implicit public subsidies arising from the public safety net could accordingly continue to flow freely within integrated groups. Taken together, these options have a very limited effectiveness. The Commission services will therefore not further assess these reform options.

Selected structural reform initiatives have been mapped into Table 2 for illustration purposes only and according to the understanding of Commission Services of the latest proposals and (draft) laws. The conceptual mapping inevitably is a simplification of reality. National initiatives differ from the stylised options and the mapping does not necessarily do justice to the different national initiatives. See Annex A1 for more details on the respective national initiatives.

4.5. Issues related to implementation

The social benefits and costs of the options outlined in table 2 will be compared against each other and against the “no policy change” scenario in the next chapter. Irrespective of the preferred options that will arise as a result of that comparison, there are three more specific implementation issues that need to be addressed in any structural reform option. The first implementation issue concerns the scope of banks that would be subject to structural reform. The second implementation issue concerns the role of supervisors in structural reform implementation, and the third implementation issue relates to the structural reform implementation timeline.

4.5.1. Institutional scope

Irrespective of choices related to activities and strength, given the focus on TBTF banks identified in chapter 2, a structural reform initiative in principle targets large banks only. Hence, it only applies to a small subset of the more than 8000 banks incorporated in the EU. It would in particular not concern the vast majority of local and regional banks, such as small cooperative and savings banks, which put relatively large proportions of their balance sheet at risk to serve the local real economy as well as the banks that primarily focus on customer-related lending. The latter banks are more likely to be resolvable than larger and more trading intensive and interconnected banks.

In light of the focus on TBTF banks, an approach should be found to limit the scope for structural reform to banks that either are very large and/or engage in significant trading activity. To that effect, the HLEG recommended that only banks with significant trading activities should be subject to structural separation. However, the definition of trading activities underpinning the thresholds recommended by the HLEG, which is based on the aggregation of broad accounting categories and includes available for sale assets, has been criticised for being a poor proxy of risky trading activities.

Many financial institutions, as well as public authorities, that responded to the Commission’s public consultation about the options considered in this Impact Assessment also favoured the use of a risk-based threshold to determine the appropriate scope of banks for structural reform (See annex A2). Under a risk-based approach, the threshold would be based on a measure of banks’ trading risk. A risk-based threshold would accordingly have the benefit of clearly linking structural separation to existing measures of the risk associated with trading. The most logical source for such a risk-based threshold would be the risk-weighted assets for market and counterparty credit risk, as these are the prudential measure of banks’ trading books. However, such a risk-based approach has certain drawbacks. First, risk-weighted data is not always publicly available. In the absence of verifiable and consistent public data, the delineation of the appropriate threshold and hence scope becomes less transparent. Second, there are concerns about risk-based approaches that would rely on risk weighted assets given the reported inconsistencies between banks when presented identical stylised portfolios of assets. Furthermore, both European and international standard setters are currently reviewing the RWA approach. Both concerns led the HLEG to reject using such measures for threshold and initial scope delineation purposes.

Another alternative is to apply structural separation to banks that are designated as systemically important by competent authorities. This was also highlighted as an alternative by some respondents to the stakeholder consultation. This approach would have the advantage of linking separation to one of the main drivers of implicit public subsidies. It also has the operational advantage of linking it to a readily available measure of systemic risk that has been translated into European law by the CRDIV. Moreover, analysis by the Commission services (Annex A8) indicates that nearly all banks that are of global systemic importance engage in significant trading activity. However, the systemic risk measure may not be a perfect proxy for significant levels of trading activities, as the analysis carried out by the Commission services also highlight that banks beyond the current G-SII list engage in significant trading activity. So, whereas the list of G-SII may be useful as a starting point, it may not necessarily include all banking groups that may require closer scrutiny. Also, while the provisions related to systemically important institutions have been laid down in the CRDIV, the definitive list of G-SIIs has not yet been finalised. Accordingly, using this approach as a sole base for determining the banks subject to separation would not yield clarity in the short term.

For the above reasons, the Commission services on balance recommend retaining the HLEG recommendation for an accounting-based approach. It is likely to be more effective than the other two alternatives referred to immediately above. It would be transparent and would provide for a higher degree of legal certainty for banks and market participants as regards the scope of banks subject to structural reform. While the Commission services do not consider risk-weighted assets to be an appropriate base for setting the institutional scope, they consider that the systemically important institution-approach should be included explicitly in the sample of banking groups that require further examination. Thresholds should further identify banks with significant trading activities. Accordingly, in addition to banks designated as G-SIIs should, banks that are captured by the accounting-based trading activity threshold, in principle should be subject to the provisions of the reform..

The Commission services have therefore assessed and reviewed the recommendations of the HLEG with a view to consider in particular the suggested examination threshold levels and definition of trading activity to be taken into account (see Annex A8). The Commission services have assessed four options in that respect:

(1) Using the HLEG definition (Assets held for trading and available for sale);

(2) A more narrow definition that account for the current reform agenda by excluding securities held for liquidity purposes under Basel III, and/or a share of derivative assets that are likely to become cleared by Central Counterparties (CCPs);

(3) A definition focused on the gross volume of trading activities, as this is likely to focus on proprietary traders and market-makers; or

(4) A definition focused on net volumes, which is likely to only capture those that have a higher share of unbalanced risk trading (proprietary traders).

Preference for either of these approaches accordingly depends on which of the reform option outlined in Table 2 is being pursued. The assessment of the reform options carried out in Chapter 5 is accordingly done on the assumption that any structural reform would apply to a limited subset of TBTF banks only. A comparison of the alternatives for a threshold stated above will accordingly be carried out following the comparison of the options in Table 2.

4.5.2. Role of supervisors

Irrespective of the choice of threshold, a separate issue is whether supervisors should retain some discretion in reviewing the scope of institutions, the activities subject to potential separation as well as whether separation would need to materialise. As regards the institutional scope, any threshold may not always capture the right institutions. As regards the activity scope, there may be a case for allowing flexibility for supervisors in how to address particular activities in light of e.g. local market circumstances (e.g. market-making provisions by banks may be regarded as more useful where public markets are particularly illiquid). Finally, as regards the decision of whether separation would automatically materialise, specific characteristics of banks may be such that the objectives of structural reform are not put at risk or that diseconomies of scale and scope do not arise. Therefore supervisors may want to exempt them from separation requirements. There is therefore an a priori case for a degree of supervisory discretion.

However, this needs to be constrained for both practical and legal reasons. First, structural separation decisions go beyond simple prudential matters and ultimately reflect societal and economic choices as regards the reach of the public safety net. Such policy decisions accordingly need to be taken at political level. Second, supervisory discretion cannot be unlimited if the intention is to ensure consistent outcomes across the Internal Market. Third, unconstrained supervisory discretion reduces legal certainty. For all those reasons, structural separation decisions should be anchored at political level with only a degree of carefully framed powers granted to supervisors.

The ways of designing and framing supervisory powers will be assessed after the comparison of the different reform options as outlined in Table 2.

4.5.3. Timeline for implementation

The HLEG called on the Commission to consider a sufficiently long implementation and transition period. There are a number of rationales for considering the time dimension. First, structural separation, no matter how beneficial to society, will yield private costs. Imposing restrictions on the structure of a banking group will by construction limit bank flexibility to freely allocate capital and liquidity within the banking group. Moreover, improved market discipline and more credible resolution will get reflected in higher funding costs at unchanged balance sheet size, in particular for the trading entity. To the extent that bank shareholders and/or bank employees do not shoulder this cost, banks may pass the costs on to borrowers or may not provide the activity at all, which in principle could reduce economic growth depending on whether other alternative actors step in (see section 5.2 for a more in-depth discussion of the social costs and benefits of structural reform). Furthermore, the transfer of existing assets and liabilities to a potentially new separate legal entity requires time and involves one-off costs, the level of which would depend on the timeline available for implementation. The implementation will also depend on the role of supervisors in reviewing the scope of institutions as well as the activities subject to potential separation. Finally, the EU has adopted or will soon adopt several initiatives in related areas, including Basel III, BRRD and the SRM that will not become fully applicable until 2018 or 2019. Accordingly, in order to allow institutions subject to these related legislative initiatives to implement them in a coordinated manner and to gradually absorb related transition costs, it is appropriate to provide for a similar transition period to implement a future structural reform initiative. The latter should not become applicable before 2019.

5. Impact and comparison of reform options

The previous chapters list the problems with respect to the current corporate structure of the large, complex and trading-intensive EU banking groups (Chapter 2), describe the corresponding objectives of structural bank reform in trying to resolve the identified problems (Chapter 3), and outline the range of policy options under consideration to achieve the stated objectives (Chapter 4).

This chapter assesses and compares in qualitative terms the relevant reform options under consideration (the two last columns of Table 2). The chapter should allow understanding the rationale, objectives and viability of the preferred structural reform options. The benefits and costs of the different structural reform options will be compared, qualitatively and where possible quantitatively.

Stakeholder views on different reform options will be considered, bearing in mind the clear divergence of opinion between the responses of the banks on the one hand (against any separation and especially any option going beyond separating proprietary trading) and consumer associations and individuals on the other hand (in favour of full ownership separation, and strong subsidiarisation of all wholesale and investment bank activities as second best) highlighted above and beyond those generally held views, the limited degree of more detailed views on different reform options.

Section 5.1 presents the different criteria used to assess and compare the different reform options' effectiveness, efficiency, and coherence). Section 5.2 discusses the different benefits and costs used to assess and compare the different bank reform options according to the retained assessment criteria. It also stresses that the assessment needs to consider social, rather than private, costs and benefits. Section 5.3 assesses the options based on subsidiarisation according to strict rules (options B, E and H of Table 2); and section 5.4 assesses and compares the reform options related to ownership separation (options C, F and I of Table 2). The assessment in these two sections leads to most options being discarded and a limited number being retained for further assessment and comparison. Section 5.5 assesses the retained options against each other and suggests possibilities to further increase their effectiveness and efficiency. 5.6 discusses the retained options aimed at increasing the transparency of shadow banking. Section 5.7 discusses complementary aspects related to (i) the quantitative estimation of costs and benefits and its difficulties, (ii) international aspects such as territorial scope and impact on international competitiveness, and (iii) institutional scope aspects such as thresholds and the role of supervisors. Section 5.8 further discusses the qualitative impact on the different stakeholder groups of implementing the preferred reform options and responds to arguments raised against structural bank reform. 

5.1. Comparison criteria

Three key criteria are applied for the purpose of comparing the different reform options within this Impact Assessment.

(1) “Effectiveness” measures the extent to which the specific or microeconomic reform objectives are being met, i.e. the extent to which social benefits, i.e. benefits to society as a whole, are achieved, notably facilitated resolution and recovery in the bad times, facilitated management, monitoring and supervision in the good times, reduced moral hazard, reduced conflicts of interest, reduced capital and resource misallocation and reduced trading culture contamination across the banking group, and improved competition (see section 3.2);

(2) “Efficiency” assesses the social costs, i.e. costs to society as a whole, incurred when implementing the respective reform options (mainly foregone economies of scope); and

(3) “Coherence” measures the alignment of the reform options with the Commission’s overall policy objectives in general and the degree to which they are complementary to on-going banking reforms.

In the remainder of this Impact Assessment, greater social benefits should be taken to imply greater effectiveness, and vice versa. Greater social costs in turn will imply lower efficiency. Structural reform needs to effectively and efficiently target and address the root problems in the corporate structure of large, interconnected, complex, and trading-intensive banking groups, to the extent that the latter remain TBTF, even after taking into consideration other complementary regulatory measures. As regards the third “coherence” criterion, the Commission services take the view that the objectives and impact of all options considered are consistent with and supportive of the broad EU policy agenda and the relevant policy initiatives of the financial reform agenda, and thus in general fulfil the conditions for coherence (see e.g. section 2.3 and Annex A3 as regards complementarity with other policy initiatives in the field of banking). Consequently, this criterion has not been part of the initial assessment of the relevant reform options (sections 5.3 and 5.4). However, the extent to which the implementation of the retained options would rely on supervisory judgement may have an impact on the degree of coherence with broad policy objectives (e.g. the need for a single rulebook, the need to limit the potential for regulatory arbitrage, the desirability of legal certainty, etc.). The options that have been retained as a result of the initial analysis and comparison are thus subject to an assessment in terms of coherence in section 5.5.

5.2. Social versus private benefits and costs

When analysing the impact of structural reform, it is important to distinguish “private” (i.e. stakeholder-specific) benefits and costs from “social” benefits and costs (i.e. benefits and costs for society as a whole).[45] Whereas private costs and benefits focus on the impact on banks, bank shareholders, bank employees, or the EU banking sector, the latter is broader in scope and targets total or aggregate welfare more generally by incorporating the impact on all stakeholders in society, including bank customers (e.g. depositors, borrowers and consumers of financial services), bank creditors, and taxpayers (i.e. the public finances of governments).

Structural reform is expected to result in increased private funding costs for the banking group, in particular for the trading entity which is being separated from the deposit entity and hence no longer benefits to the same extent from being implicitly linked to the public safety net. Those funding costs are recurrent private costs. However, they do not necessarily result in any social cost, as the increased funding cost merely reflects the shift of bank risk and contingent liability away from the taxpayer and toward the unsecured debt bank creditors that should, in principle, bear the risk (and be properly remunerated for being exposed to it). The increased private funding cost for the bank reflects and is being offset by a decreased (implicit) public subsidy to the benefit of taxpayers. In sum, the increase in the private funding cost due to the removal of the implicit public subsidy should not be considered as a social cost.[46]

5.2.1. Social benefits

The social benefits retained in this Impact Assessment correspond to the microeconomic objectives of the structural reform: facilitate resolution and recovery in the bad times, facilitate management, monitoring and supervision in the good times, reduce moral hazard, reduce conflicts of interest, reduce capital and resource misallocation, and improved competition.

With respect to the latter, evidence suggests that competition in the internal market can be distorted as i) larger banks and ii) banks headquartered in countries with a strong sovereign rating are more likely to benefit from implicit public subsidies (see Annex A4.1 and A4.2). Structural reform can therefore contribute towards restoring a level playing field across small and large banks and across EU Member States.

Social benefits are not only static. Dynamic social benefits are also important, as reduced moral hazard also reduces aggressive balance sheet expansion. Structural reform aims to reduce the undue or artificial promotion or subsidisation of specific excessively risky activities, in particular those that are scalable and transactions-oriented, as opposed to relationship-oriented. Hence, the private cost increase may negatively affect specific banking activities, which is welcome and desirable (despite a corresponding funding cost increase) to the extent that these activities, on balance, have a negative impact on the real economy and society as a whole.

5.2.2. Social costs

The costs to society of structural reform need to be defined carefully. They will mainly consist in reduced genuine economies of scale and scope and increased supervisory costs to monitor and implement the reform.

It cannot be excluded that structural reform will give rise to the loss of genuine scale and scope economies. Combining different activities within one institution may give rise to genuine economies of scope related to risk diversification, revenue economies of scope, and cost economies of scope. There may also be genuine economies of scale as an increased scale may allow spreading fixed costs which can reduce average costs. However, economies of scale are activity-specific and are likely to be relatively small given that only large banks would be subject to structural reform and that empirical evidence on economies of scale for large banks is weak (see Annex A9 for a literature review). There could be however some genuine economies of scope. For example, as explained in chapter 4, integrated banks do have a natural advantage in acting as market makers because of the fact that banks have a variety of other relationships with the clients who want to make trades and the fact that acting as a market maker for a security is often a natural follow-on activity for securities underwritten by the banking group. This may allow integrated banks to perform this activity more efficiently than other market players, thus better serving clients and/or contributing to enhancing market liquidity. This can lead to some recurrent private costs. Such effects should be weighed against potential benefits that could flow from structural separation, as discussed elsewhere in this Impact Assessment.

Related to the notion of risk diversification and economies of scope are also the regulatory capital compliance costs. Regulatory capital compliance costs refer to the fact that banking groups are under an obligation to separate capital to comply with regulatory requirements on an individual and consolidated basis. As a result banking groups are expected to need to hold more capital in the aggregate than without separation, as diversification across activities may reduce the capital needed to meet capital requirements, which represents a loss of economies of scope. This is a recurrent private cost.

Operational costs are costs that banking groups incur from having to operate through separate subsidiaries. Operating with separate stand-alone management boards will be operationally more costly than operating through a single management board. Administrative systems may also have to be separated. These operational costs are recurrent or equilibrium costs. Operational costs as defined above refer to costs borne by banks.

In the remainder of this Impact Assessment, implementation costs will refer to costs borne by regulators and supervisors when writing up and monitoring the compliance of banks with the new regulatory framework on an on-going basis.

There may also be additional transitional or one-off costs related to the setting up of the new legal entities, related to transfers, sales or the winding down of business units, client migration, and corresponding administrative structure changes.

On the other hand, at least some economies of scale and scope may get exhausted when a bank reaches a certain level of assets. Diseconomies of scale such as complexity, conflicts of interest, increased private or systemic risk may dominate. In any case, economies of scale and scope should be evaluated after correcting for the funding advantage that arises from banks being TBTF. After such correction, the literature suggests that economies of scale and scope are exhausted at relatively low levels of bank balance sheet size.[47] Those levels are significantly lower than the balance sheet size of the vast majority of banks that would exceed the threshold levels considered for the potential application of structural separation (section 5.7.3.1).[48]

In sum, the literature suggests that genuine foregone synergies between relationship banking and trading activities are exhausted at a fraction of the balance sheet size for some affected banking groups. From a private cost perspective however, funding cost increases will be unavoidable for the trading entity at unchanged balance sheet size and in fact reflect the objective of structural reform to reduce implicit public subsidies for certain trading activities (increased market discipline, reduced competition distortions, no artificial balance sheet growth). Operational costs will be inevitable as well, but seem modest in comparison to the other costs and in comparison to the social benefits. Regulatory capital compliance costs should also not be exaggerated given the strong international push for more and better capital and the recent academic literature (Admati and Hellwig (2013)).

5.2.3. The impact on economic growth of reducing implicit public subsidies

Most, if not all, banking activities are valuable and useful to the real economy. But, just like most activities in a market economy, many should not be promoted or subsidised by the taxpayer unless there is an appropriate justification.

Subsidising an activity is a form of government intervention that is justified if it directly addresses a genuine market failure, such as when the activity resolves an information asymmetry (e.g. funding SMEs), avoids negative externalities (e.g. avoiding financial instability), provides a public good (e.g. market liquidity or financial stability) or addresses a market power concern (e.g. due to switching costs or network effects).

It has been argued above that the increased funding costs are the mirror image of reduced implicit public subsidies for the bank. The literature review on implicit public subsidies in Annex A4.1 and related own research in Annex A4.2 concludes that implicit public subsidies are in particular enjoyed by larger banks and are significant compared to their annual profitability.

Is lowering these subsidies justified and would economic growth be affected? To answer this question, it is necessary to determine to what extent the activities that benefit from these implicit public subsidies address genuine market failures, are underprovided, or are not overprovided due to excessive subsidies. Some activities may justify being subsidised but may exceed the optimal level.

At one extreme, there is lending to households and SMEs funded by deposits, where regulatory intervention in the form of explicit deposit insurance is warranted to avoid bank runs. At the other extreme, proprietary trading is a banking activity that is not customer-oriented or related to another core banking function, but only aims at generating profits for the bank. Therefore, it seems hard to justify that implicit public subsidies are channelled toward such activities by allowing deposit-taking banks to perform such activities.

In between, there may be more ambiguous cases. One example is activities related to secondary trading of sovereign debt (e.g. market-making). The potential reduction of market liquidity in secondary markets may to some extent depress primary market government bond prices. The risk of a corresponding yield increase, however limited in comparison to government bond yield levels, may not be desirable at this current juncture. Accordingly, when separating a particular activity, exemptions for sovereign bonds could be considered in light of this uncertainty. Such an exemption would be consistent with the zero risk weight currently assigned to sovereign bonds in the CRDIV. At the same time, an exemption could lead to distortions (e.g. crowding out of private debt issuance), would increase concerns with respect to the accumulation of sovereign debt risks, and the zero risk weight for sovereign bonds is currently under discussion. Given the uncertainty concerning the market response in terms of liquidity and sovereign bond yields, this impact assessment does not propose specific recommendations on this matter that involves a large degree of political judgement. See also section 5.8 for further discussion of the potential impact of structural reform proposals on market liquidity more generally.

5.3. Assessment of reform options based on subsidiarisation according to stricter rules

Options B, E and H (column 2 in Table 2) are reform options that roughly correspond to reforms currently being pursued by EU Member States or being recommended by the HLEG.

As regards strength of separation, these options would all foresee additional restrictions being imposed on the economic, legal, and operational linkages between the deposit and trading (sub-consolidated) entities to ensure the integrity and effectiveness of the separation.[49] The restrictions aim to effectively address intra-group funding subsidies. In terms of legal separation, it includes restrictions on ownership links between separated entities within the group. This would provide for a stricter degree of economic separation (e.g. separate funding). Irrespectively, this includes stricter economic separation in its own right, notably by considering rules on intra-group relations (e.g. requirements that intra-group transactions be on third party, commercial terms; ensuring that current large exposure restrictions are also imposed intra-group and possibly applying stricter requirements for large exposures of deposit entities; providing limits on intra-group guarantees (deposit-taking entity would not support trading entity), and stipulating a higher degree of governance separation (e.g. limits on cross-membership of board directors within the group).

In terms of scope of activities, the reform options differ:

· Reform option B foresees the mandatory separate subsidiarisation of proprietary trading and bank-internal hedge funds within a banking group;

· Reform option E foresees the mandatory separate subsidiarisation of a larger set of trading activities within a banking group, notably proprietary trading (including bank-internal hedge funds), market making, investing, sponsoring, and structuring activities related to “complex securitisation”[50], and structuring, arranging or execution of “complex derivative transactions”[51]. However, subject to supervisory approval, the deposit entity would under this option still be allowed to engage in amongst others underwriting[52], investing, sponsoring, and structuring activities related to “simple securitisation”,  lending to large corporates, and maintaining exposure to private equity or venture capital funds. The latter activities may be underprovided from a social point of view if they are no longer allowed to benefit from the implicit state support and are more of a relationship-oriented nature. They accordingly do not raise similar concerns with respect to introducing a trading culture within the deposit entity. Furthermore, these activities may be of greater importance to the real economy and their separation would accordingly come with a degree of cost to society. Proprietary trading, market making, and complex securitisation, and complex derivatives are more easily scalable and more transaction-oriented in nature;

· Reform option H requires the subsidiarisation of all wholesale and investment banking activities.

Stakeholder views differ on the merits of these reform options. According to the responses from consumer associations and individuals, this degree of separation could be acceptable, provided that it includes a sufficiently broad range of activities. According to those responses, option E would be the minimum acceptable reform option to be effective. Support is to some extent echoed by some institutional investors that call for separation so as to address excessive risk-taking and reduce moral hazard. Bank respondents, on the other hand, highlight the costs in terms of foregone efficiency associated notably with subsidiarising a broader activity range such as in options E and H. Bank respondents, as well as corporate respondents, also argue that market-making in their view is a socially useful activity, the provision of which may be hampered by reform along the lines of E or H. 

5.3.1. Social benefits

Options B, E and H are likely to be effective in facilitating resolvability and monitoring, reducing moral hazard, reducing conflicts of interest, reducing resource and capital and resource misallocation, and improving competition, albeit to different degrees and depending on the specific subsidiarisation rules.

5.3.1.1. Facilitate recovery and resolution

In terms of resolvability, subsidiarisation into a simpler group structure with further restrictions to separate the different entities should facilitate the assessment and allocation of losses, while making constituent entities that fail smaller and provide authorities with additional options to orderly resolve parts of or entire banking groups. It should enhance the credibility of the application of any resolution tool, and hence of the process of orderly resolution of failing banking groups

Separating proprietary trading, market making, complex securitisation activity, and complex derivative activity all facilitate recovery and resolution in the bad times for the following reasons.

Proprietary trading potentially gives rise to large open positions subject to market risk and counterparty risk (risk that the counterparty to the investment will fail to pay) as well as interconnectedness between institutions. Conversely, having to unwind less large open positions and not being interconnected with other banking groups to the same extent will facilitate the recovery and resolution of the deposit entities. The potential opaqueness, complexity and interconnectivity of proprietary trading represent important impediments to orderly and swift resolution. Following separation the balance sheet of the trading entity is also expected to shrink as the risks associated with its activities are fully priced.  This will induce better market discipline and reduce resolvability impediments.

Note that the impact of subsidiarising proprietary trading on the resolvability of the bank may currently be relatively limited as most banks claim that they do not engage in proprietary trading to any significant extent. However, evidence suggests that the extent of proprietary trading activities has gone down substantially since the start of the crisis.[53] Absent specific restrictions on proprietary trading activities there is no safeguard that would prevent banks from expanding such risky activities in the (near) future. Moreover, the limited data that does exist typically tend to measure only dedicated proprietary trading desk activity.[54]

Individual trading positions are treated the same way in resolution whether they result from client-driven market making or proprietary trading. Given the relative importance of market making and the quantity of positions needed to be resolved, and given the fact that market makers are typically interconnected with other large banking groups, a separation of market making activities could also have significant social benefits in terms of facilitating resolvability.

Complex forms of securitisation and derivatives allow for opaqueness and complexity which impede swift resolution and recovery. Complex securitisation also allows for significant growth in short-term debt between financial intermediaries and leads to financial intermediaries becoming intertwined.

The stricter the rules on subsidiarisation, the more resolution and recovery will be facilitated as the separated trading entity will have been run and organised on a more autonomous basis. This provides authorities with more options for swift and orderly resolution in a scenario in which parts of the group (or the group as a whole) need to be resolved. Moreover, the relevant trading activities would not be provided on the same scale as the separated trading entity would be subject to more effective market discipline.

All three reform options facilitate resolvability, but options E and H arguably more than option B, given the limited importance of proprietary trading and the broader activity scope.

Underwriting is not as easily scalable as pure market making. Given that underwriters typically retain a significant fraction of issued securities in their inventories and play an active market making role immediately after the issuance, resolution may be improved by separating underwriting from deposit taking, as the inventory of relatively illiquid assets will be smaller. However, underwriting does not give rise to similar interconnectedness across financial institutions, as is the case with market making. All in all, underwriting is a more relationship-oriented activity and less scalable as market making. The additional benefits in terms of banking group resolvability of moving from option E to option H are hence limited.

Box 1: Interaction and consistency of structural reform with the single point of entry and the multiple points of entry resolution strategies

The BRRD requires banks to submit recovery and resolution plans to the competent resolution authority. The FSB guidance, published in July 2012, indicates that large global banks will be forced to choose between two “resolution mechanisms”, which will dictate how they restructure themselves and ensure that crucial banking functions – payment systems, trade finance and deposit taking – can continue, no matter what happens to the larger group. These two stylised approaches are known as the “single point of entry” (SPE) resolution, in which resolution powers are applied to the top of a group by a single national resolution authority (for banks such as Goldman Sachs and JPMorgan that operate as an integrated group) and the “multiple point of entry” (MPE) resolution in which resolution tools are applied to different parts of the group by two or more resolution authorities acting in a coordinated way (for banks such as Santander and HSBC, that operate as locally capitalised subsidiaries).

Which type of resolution strategy (SPE or MPE) is better for a particular group will depend on the structure of the group, the nature of its business, and the size and location of the group’s losses. In either case, for bail-in to be the chosen resolution tool there needs to be sufficient loss absorbing capacity in the relevant legal entities.

From the perspective of banks the preferred resolution strategy will depend on their structure, jurisdiction and the attitude of the supervisors and regulators. Some banks may find it prohibitively expensive to raise local capital for the MPE strategy, and will opt for the SPE whenever possible.  However, regulators may oblige putative SPE banks to hold significantly more capital at the group level, inducing other banks to opt for the MPE method. Either way, investors stand to benefit from more public disclosure, from both regulators and the banks, of what will happen in the event of a crisis.

The obstacles to resolution will differ between resolution strategies. In SPE strategies, obstacles may arise from the location of primary loss absorption capacity; debt instruments governed by foreign law; and inadequate management information systems that do not support rapid valuation of losses. In MPE strategies, obstacles include legal, financial and operational dependencies within a group. Obstacles relevant to both strategies include immediate rights of exercise of termination clauses in contracts held by a bank’s counterparties and the exercise of cross-default clauses.

Under subsidiarisation with additional restrictions reform options (SUB+), the trading entity needs to be separately capitalised and funded and therefore will always be identified as a “point of entry”. Thus it should have sufficient loss absorbing capacity (“LAC”) to cover its likely losses in resolution and those of subsidiaries below it for which a separate resolution is not planned. Alternatively, it should be capable of being wound down without affecting the rest of the group. Debt is likely to be issued by operating subsidiaries to third parties, with the consequence that the use of resolution powers to convert debt to equity may result in a change of ownership, loss of control by the top parent or holding company and, potentially, separation from the group. The resolution strategy should address how such change of ownership and separation from the group can be implemented without disruption to the entity’s critical operations. The bank holding company can hold equity in the trading entity but only at arm’s length and constrained by large exposure limits. Thus losses at the level of the trading entity (or its subsidiaries) cannot be up streamed to the parent company, by recapitalising the trading entity with excess capital from the parent or from any of the non-ring fenced deposit taking entities. The bank holding company may still opt for an SPE approach for non-ring fenced subsidiaries.

The interaction between structural reform and resolution strategies under BRRD will need to be discussed between the supervisor and the resolution authority, subject to consultation procedure to be established.

5.3.1.2. Facilitate monitoring, management, and supervision

Subsidiarisation with further restrictions will materially improve market discipline and increase transparency in the stand-alone performance of the different parts of the group. The reason is that restricted interconnections will no longer allow shifting profits and losses within the group and will render rules related to governance more effective.

Separating proprietary trading, market making, complex securitisation, and complex derivative activities facilitates monitoring, management, and supervision.

Increased market discipline on the trading entity facilitates the latter’s supervision, even without factoring in the likely reduction in proprietary trading, market making, complex securitisation, and complex derivatives that result from enhanced market discipline.

The nature of proprietary trading hinders the ability of regulators, supervisors and bank managers to properly understand and thereby calibrate the risks taken, in particular tail risk (i.e., the risk that an investment will perform significantly worse than expected). It is equally complex to apply the correct capital treatment so that banks have sufficient resources to absorb losses if these occur. Proprietary trading can also be a high-frequency activity that may result in thousands of daily transactions. As a result, snapshots of the positions of these activities may have limited predictive value for future positions. This rapid movement in underlying positions significantly raises monitoring costs for the management of the banks and for market participants (such as bank creditors and shareholders).

Bank management and external monitoring by the market and supervisors will be facilitated, as the activities that are most scalable and hence difficult to monitor would be located in separate subsidiaries with their own separate governance. Subsidiarisation with restrictions would yield stronger benefits in this regard, given the stronger degree of governance separation. The additional restrictions applied to the respective entities will ensure that it becomes easier to assess the stand-alone performance of the different entities.

All three reform options facilitate management and monitoring, but options E and H arguably more than option B, given their broader activity scope. As the additional activities to be separated under reform option H are not as significant and scalable as market making, the additional benefits in terms of facilitated monitoring of moving from option E to option H are correspondingly small.

5.3.1.3. Reduce moral hazard

Separate funding requirements and restricted interconnections between sub-consolidated entities will impose a significant increase in market discipline on the trading entity. As a result, depositors would be better shielded from risk-taking originating from subsidiarised trading activities, also because intra-group large exposure rules would apply (can no longer be waived).

Separating proprietary trading, market making, complex securitisation, and complex derivative activities will reduce excessive risk taking.  Through subsidiarisation, these activities would not benefit (to the same extent) from the implicit public subsidies which would also help to re-align private and social interests. The resulting increased funding cost would reflect the inherent riskiness of the activity (although systemic risk may still not be adequately reflected in the institution-specific funding cost). As a result, moral hazard in the trading entity will be reduced.

Proprietary trading is an inherently risky banking activity that is by definition not customer-oriented. It has the ability to produce “tail risk” or systemic risk and is easily scalable (in comparison to more relationship-based activities such as lending). When part of a larger banking group, traders benefit from lower funding costs and as a result have the ability and incentive to take significant risks, even without having access to liquidity (through short-selling positions). Separating proprietary trading from the deposit entity allows shielding depositors from this type of risk-taking.

When facilitating client business through market making, a bank is likely to try and hedge most of its risks. Hence, genuine market making is generally considered to entail limited market risk. However, the actual exposure to risk may vary depending on the liquidity of the instruments, on changes in market volatility and on significant variation in the sizes of positions that market making clients may wish to acquire or liquidate. Moreover, there may be a mismatch between the position and the hedge (basis risk) and the hedge will need to be rebalanced over time as market moves alter risk profiles. Furthermore, market makers are still exposed to high counterparty risk and the concrete functioning of market making can vary in relation to different financial instruments and market models.

Given its importance as a share of trading revenues, market making entails significant risk and separating it from the deposit entity could significantly reduce moral hazard, excessive risk taking, and artificial balance sheet expansion.

In terms of reduction of moral hazard and excessive risk taking, the legal, economic, and governance requirements to be imposed under the stricter form of subsidiarisation of options B, E, and H would reduce risk taking incentives by forcing the trading entity to internalise the true cost of its risk taking and to more strictly separate it from the public safety net associated with insured deposits through (possibly sharpened) intra-group large exposure rules.

By separating banking activities from core deposit-related activities, the funding of these activities would become more risk-sensitive. As a result, banks would have less of an incentive to engage excessively in these activities. The extent to which funding becomes fully risk sensitive depends on the degree of separation. Accordingly, subsidiarisation with restrictions that give rise to a stronger degree of economic and legal separation will result in more effective market discipline compared to subsidiarisation as such.

Prudential regulatory requirements, aimed at promoting the stability of financial institutions currently apply at the consolidated group level (and at the individual level, but only if they are not waived). At the core of these prudential standards are the requirements for banks to hold buffers of capital and liquidity to absorb losses (in the case of capital) and to provide emergency funding (in the case of liquidity). A structural separation would entail different entities holding separate capital and liquidity buffers, thereby aligning the cost of regulation more closely with the risk. This promotes market discipline. It would also mean that the entities would have separate funding requirements, ending the risky cross-subsidy of trading activities with deposits.

All three reform options reduce moral hazard, but options E and H arguably more than option B, given their wider activity scope. Underwriting is not as easily scalable as pure market making. The scope for moral hazard reduction is still significant, but smaller than for market making. Market making as a follow-on activity of underwriting does imply that significant securities and derivatives inventories are being built up, and hence that risks are potentially significant, but hedging instruments exist and risks can in principle be monitored and managed. The scope for additional moral hazard reduction when shifting from option E to option H is accordingly limited.

5.3.1.4. Reduce conflicts of interest

Subsidiarisation with further restrictions will address conflicts of interest, to the extent that governance rules reach further than outlined in the CRDIV, and especially if there is a duty on the banking group to uphold the integrity and objectives of the separation. The restrictions allow safeguarding the separation objectives and ensure that a short-term trading culture will not continue to unduly influence the relationship-oriented deposit entity.

Separating proprietary trading, market making, complex securitisation, and complex derivative activity reduces the scope for conflicts of interest and avoids that a short-term oriented trading culture gets installed within a deposit entity.

Proprietary trading is particularly prone to conflicts of interests because the bank in its role of proprietary trader no longer is a service provider to its client, but becomes a potential competitor and hence faces interests that are no longer aligned with those of its clients. The bank can make improper use of client-related information to increase its own profits. The commercial bank department may have private information about the likely bankruptcy of a firm it has granted a loan and may buy credit protection against the default of the firm from the unsuspecting public, thereby reducing its own credit risk whilst earning a fee.

In theory, genuine market making aims at facilitating client business and hence bank interests are supposed to be aligned with customer interests. However, principal-agent problems need not to be confined to proprietary trading given that market making and proprietary trading activity are difficult to disentangle for outsiders to the actual transactions.

In general, if markets are opaque, such as is the case in over-the-counter markets, and if market makers have superior access to information, collusion and exploitation of conflicts of interests may occur. The origin of the problem is an inherent conflict of interest. Banks possess (asymmetric) information in the form of customer trade details, including the number and size of trades to be executed. And they have knowledge that their own proprietary positions could be harmed without (or could benefit with) trader intervention. The banks allegedly act on that knowledge, against their customer’s best interests and in favour of their own, as evidenced in recent banking scandals, related to front running, FX bid rigging, Libor benchmark rate setting, etc.[55]

Stricter rules of subsidiarisation, for example on governance separation, would provide further checks.

All reform options reduce conflicts of interests, but options E and H arguably more than option B, given their wider activity scope.

Separating underwriting will reduce the scope for conflicts of interests, as the interests of the bank as underwriter and as loan provider are typically not aligned.[56] Within a large and diversified banking group, the commercial bank department may have private information about the likely bankruptcy of the firm it has granted a loan and may hence encourage the underwriting department to sell bonds or issue shares to the unsuspecting public, thereby reducing its own credit risk whilst earning a fee. Banks have an incentive to hedge their risk as underwriters, guaranteeing the proceeds of the share issue, but this may potentially have an adverse impact on their clients’ share price. Alternatively, a bank’s lending division may feel pressured to provide bank loans to a firm whose shares have been issued by the bank’s underwriting division, even though such loans would not be granted absent any such in-house pressure. Some studies indicate that earnings forecasts and stock recommendations provided by an analyst working with the lead-underwriter are on average inaccurate and positively biased, and unaffiliated analysts perform better and provide higher long-run value to their customers.[57] The main concern is that the bank uses the informational advantage it gains from conducting different activities to its own advantage, thereby misleading customers and investors.

Having said that, evidence does not suggest that there are obvious conflicts of interests between underwriting and loan making.[58] In fact, it suggests that bonds underwritten by commercial banks default less often than bonds underwritten by investment banks. However, the conflicts of interest seem more severe and more likely to exist in a universal bank that has an underwriting division together with an asset management division. These studies seem to support the view that asset management divisions may feel pressured by the bank’s underwriting division to buy and hold poorly performing issues to make a customer satisfied, even though this may be unwise.

Next to internal monitoring and controlling procedures, there is outside regulation (for example with respect to insider trading and market manipulation in the context of the Market Abuse Directive/Regulation)) and the rule of law to contain the exploitation of possible conflicts of interests. In principle, the market can also respond to apparent conflicts of interests, thereby constraining their scope. The market can penalize the service provider if it exploits conflicts of interest in the form of a higher funding cost or lower demand for its services even to the point of forcing the provider into bankruptcy. The market can also promote new institutional means to contain conflicts of interest by generating a demand for information from non-conflicted specialized organizations.

However, the market is likely to be unable to contain the incentives to exploit the conflicts of interests. For the market to be able to do this it needs to have information on whether exploitation might take place. Sometimes, such information is simply not available or would require the revealing of proprietary information that would benefit a firm's competitors, thus reducing the incentives to reveal this information. Sometimes when corporate governance is poor, even the top management of the firm is not aware of the conflicts of interest and mala fide opportunistic individuals are able to capture the firm's reputational rents.

All in all, the above suggests that reform option H may be superior to option E as regards reducing the scope for conflicts of interests.

5.3.1.5. Reduce capital and resource misallocation

Depending on the separation restrictions and restrictions on intra-group pricing policies, trading activities will no longer enjoy cross-subsidies to the same extent and will de facto be distanced more from public safety net coverage (and its corresponding benefits). Hence, trading activities no longer are artificially promoted to the same extent.

Subsidiarisation rules impose limits on shifting excess capital within the wider corporate group, which reduces the incentives for the parent/trading entity to encourage the allocation of capital and human resources to trading and away from lending activity.

Separating proprietary trading, market making, complex securitisation, and complex trading activity reduces capital and resource misallocation.

The traditional raison d’être of deposit-taking banks is to be a financial intermediary between savers and investors (and thereby competing with capital markets that play a similar role). In comparison to capital markets, who intermediate more directly between savers and investors, deposit-taking banks are relatively good at: (i) monitoring and knowing their customers, i.e. resolving information asymmetries; (ii) providing insurance against idiosyncratic liquidity risks faced by households and firms; (iii) pooling risks efficiently; and (iv) performing risk-return tranching services to customers. None of these roles is fulfilled by proprietary trading. As such, bank capital and human resources are being misallocated to the extent that they are put at work in proprietary trading rather than in engaging in loan making and other core banking services.

The inherent riskiness of trading attracts and requires people who are good at taking short-term risks rather than lenders with a long-term perspective. Absent separation, a short-term return oriented culture may arise within the entire banking group, given the relatively high profitability (without adjusting for the riskiness, that is) associated with trading.

Academics have argued that market makers hamper the development of securities markets (Zingales (2012)). Large universal banks are currently accused of having protected their indispensable position in the global credit default swaps ("CDS") market through control of a trading body and information provider, which vetted whether new exchanges should be licensed. The alleged harm consists of exchanges being blocked from bringing part of the over-the-counter CDS transactions onto public exchanges, which would have resulted in lower transaction costs for their investor-customers, as well as in less financial instability as over-the-counter ("OTC") markets are more opaque and involve more counterparty risk.

To the extent that the additional restrictions reduce the implicit public subsidies and introduce effective market discipline on the trading entity, the latter will no longer be able to artificially and aggressively expand to the detriment of the deposit entity. Hence, this will result in fewer distortions away from lending and towards trading activities.

All three reform options reduce capital and resource misallocation, but options E and H arguably more than option B, given their broader activity scope. Given that underwriting is also a relationship-based activity, not easily scalable and may be underprovided when distanced from the public safety net, the justification for mandatory separation of underwriting to enhance capital allocation is weak. Reform options E and H can be deemed equivalent in this respect. 

5.3.1.6. Improve competition in the EU banking sector

To the extent that implicit public subsidies to TBTF banking groups will be reduced in the reform options that require subsidiarisation of trading entities with further restrictions, competition distortions will also be reduced and the level playing field between banks in the Internal Market (large versus small but also across Member States) restored.  

To the extent that the restrictions introduce more effective market discipline on the trading entity, competition on the merits between small, medium-sized and large banks will be promoted. Smaller and medium-sized banks will benefit as they currently do not enjoy equally high implicit subsidies (see Annex A4.1 and A4.2). Also competition among banks subject to subsidiarisation would be placed on a more equal playing field as there would be less scope for some banks benefiting more from implicit support due to a given Member States' perceived ability, willingness and incentive to intervene. Similarly banks that are less interconnected with other banks and banks that are better capitalised would also benefit as they tend to benefit less from implicit subsidies thank other banks (see Annex A.4.2). In that respect, separating proprietary trading, market making, complex securitisation, and complex derivatives (option E) or even all wholesale and investment banking activities (option H) would improve competition in the EU banking sector more than just separating proprietary trading (option B).

Options E and H will both significantly improve the competitive environment of EU banking. Arguably, given that option H is more prescriptive as to the mandatory separation of banking activities, option E will result in greater diversity in EU bank business models and hence be preferred over option H.

5.3.2. Social costs 5.3.2.1. Foregone economies of scale and scope

Economies of scope consist in risk diversification, cost economies of scope, and revenue economies of scope. All three reform options maintain the universal banking model but introduce varying degrees of structure as to how activities should be grouped together and as to which activities are allowed to be performed by a deposit entity that enjoys access to a public safety net. The reform options aim to (partially) retain the potential economies of scope, whilst eliminating the potential diseconomies of scope (i.e. achieve the social benefits).  

Given the limited importance of proprietary trading the corresponding risk diversification economies of scope are likely to be immaterial. Therefore there should not be significant risk diversification benefits lost in option B.

Genuine economies of scope related to lost risk diversification opportunities are possibly non-negligible for market making given the importance of market making for a bank’s revenues. However, given that the activities are not banned from the group altogether (no prohibition, just subsidiarisation), the risk diversification economies of scope are partially retained within the group for reform options E and H.

Risk diversification costs are particularly pronounced for reform option H as it foresees the mandatory separation of all WIB activities, including activities that are of great importance to the real economy and which deserve more to be linked to the public safety net (and hence enjoy certain subsidies), given that they address genuine market failures.[59] However, option H may score somewhat better in terms of cost economies of scope as it mandates underwriting and market making to be grouped together within the trading entity. This avoids costly duplication between underwriting and follow-on market making and secondary market making activities.

In option B, however, the main lost economy of scope may arise from the fact that separating proprietary trading from market making may give rise to important infrastructure duplication and hence lost cost economies of scope.  The lost cost economies of scope may be significant for option B, given that traders will need to make use of separate infrastructure, depending on whether they engage in proprietary trading or market making. Options E and H score better than B in that respect, as they keep trading activity together in a single trading entity and hence avoid costly infrastructure duplication.

5.3.2.2. Operational costs

All three reform options require the establishment of separate legal entities to conduct either proprietary trading related activities (option B), a broader set of activities (option E), or all wholesale and investment banking activities (option H), whereby specific additional legal, operational and economic restrictions between the respective trading and deposit entity should be introduced, monitored, and enforced.

Although these costs cannot be avoided they are likely to be relatively modest compared to the potential benefits that can be reaped from these reforms (see e.g. ICB (2011) and HM Treasury (2013)). Moreover, costs can be minimised by giving banks a sufficiently long phase-in period.

5.3.3. Conclusion

Among the three retained reform options that require separate subsidiarisation of a deposit and trading entity with additional regulatory restrictions, the Commission services consider that implementing reform option E is the best way forward in the EU context and would represent a significant improvement over the no policy change alternative.

The bulk of the benefits of the universal banking model are retained in all three reform options. However, option E yields significantly greater social benefits compared to reform option B. The separation of proprietary trading, market making, complex securitisation and complex derivative activity limits banking groups’ ability to take excessive risks through easily scalable trading activities and avoids that a short-term oriented trading culture contaminates the more traditional banking activities. It also achieves greater resolvability in the bad times and monitoring in the good times. It ensures (or, better, allows) that those and only those activities that would otherwise be underprovided benefit from the implicit state support. The separation in option B appears to yield lower social benefits due to its significantly narrower activity scope (proprietary trading is currently not performed by large EU groups to a significant extent). Whereas option E will give rise to greater social costs compared to option B, as the scope for risk diversification is also lowered correspondingly, these social costs are capped because subsidiarisation allows transferring (excess) capital from the trading entity to the deposit entity, as long as minimum regulatory requirements are met. In sum, option E yields greater net social benefits than B.

Option E is also deemed preferable by Commission services compared to reform option H, as it achieves similar benefits whilst, in principle, being less prescriptive and exhaustive as to the set of activities to be mandatorily separated. However, the mandatory separation of the entire set of wholesale and investment banking activities from banking groups in the EU may be preferable conditional on the specific circumstances of national banking systems, in particular where banks operate within a large financial system and can be heavily exposed to international financial markets. In any event, in light of the uncertainty as to the impact of the relatively broad structural reform options it may be desirable to err on the side of caution as regards the location of the fence and to leave discretion to the banks, subject to supervisory approval and constraints, which bundle of activities they would like to place in addition to the trading entity mandated activities in the trading entity. Recall that option E is also preferred over option H by the High Level Expert Group chaired by Erkki Liikanen

5.4. Assessment of reform options based on ownership separation

Options C, F and I foresee the ownership separation of respectively proprietary trading (C), proprietary trading and market making (F), and all wholesale and investment banking activities (I). The banking groups concerned would accordingly be prohibited from engaging in certain activities. These reform options are hence the most intrusive ones of all the reform options in Table 2 and would remove the separated activities from the perimeter of the regulated and supervised banking group. This section again first assesses the social benefits and costs of ownership separation and then assesses the social benefits against costs of the three options in light of the different activity scope.

The options based on ownership separation are the ones that get the strongest support from consumer associations and individual respondents to the stakeholder consultation. Together with some non-bank financial respondents, they stressed that these options, and in particular option I, would be the simplest and most effective options in the long term. Bank respondents, as well as corporates, on the contrary highlighted the significant costs that in their view arise from ownership separation compared to other reform options.

5.4.1. Social benefits

All reform options based on ownership separation would lead to social benefits. Typically these benefits would increase with the scope of activities subject to such separation, although the nature of the specific activity obviously will also matter.

5.4.1.1. Resolution

Ownership separation facilitates resolution of the banking group mainly because the risks linked to the prohibited activities no longer feature on the balance sheets of the financial institution. As a result, the remaining banking groups would become less complex and smaller in size.

All three reform options facilitate resolvability in this respect, but options F and I arguably more than option C given the limited importance of proprietary trading relative to other wholesale and investment banking activities and relative to its pre-crisis size. However, evidence suggests that the extent of proprietary trading activities has gone down substantially since the start of the crisis[60] and absent specific restrictions on proprietary trading activities there is no safeguard that would prevent banks from expanding such risky activities in the (near) future. The limited data that does exist typically tend to measure only dedicated proprietary trading desks (see also section 5.5.1.1 on implementation challenges).[61] Moreover, speculative proprietary trading can take place alongside customer-related trading and be performed elsewhere within the banking group (for example, treasury management and market making). Given the alleged currently low activity level now may be an opportune moment to address the potential future risks of proprietary trading and to prevent a renewed surge in the future within large and complex banking groups.

5.4.1.2. Monitoring, management and supervision

Monitoring, managing and supervision would be facilitated as the activities that are most scalable and complex and consequently difficult to monitor would be located outside the financial institution. Ownership separation would yield stronger benefits in this regard compared to the subsidiarisation options assessed in section 5.3.

All three reform options facilitate management and monitoring, but options I and F arguably more than option C given their broader activity scope. However, as the additional activities to be separated under reform option I (notably underwriting) are not as significant and scalable as market making the additional benefits in terms of facilitated monitoring of moving from option F to option I are correspondingly small.

In addition, as explained below, the Commission has recently issued a roadmap for tackling risks inherent in shadow banking thus mitigating the risk that the separated activities would shift to shadow banking.

5.4.1.3. Moral hazard

An important advantage of ownership separation is that, while intrusive, it is a pure “structural” reform which after implementation would significantly alleviate the need for continuous enforcement and supervision of the separation.[62] Ownership separation should effectively and significantly reduce the implicit state subsidy (expected bail out) enjoyed by financial institutions that are taking insured deposits and from that perspective remove the moral hazard linked to safety net coverage.

All three reform options reduce moral hazard, but options F and I arguably more than option C, given the broader activity scope. Again, underwriting is not as easily scalable as pure market making and, hence, the scope for moral hazard reduction will be significant, but smaller than for market making. Market making as a follow-on activity of underwriting does imply that significant securities and derivatives inventories are being built up, and hence that risks are potentially significant.  However, hedging instruments exist and risks can in principle be monitored and managed. All in all, the scope for additional moral hazard reduction when shifting from option F to option I is limited.

Option C is a superior way to address the moral hazard associated with proprietary trading compared to option B, as it protects retail banking activities more effectively from the risks stemming from this activity. However, given that most banks claim that they currently do not engage in significant proprietary trading activity, the impact of ownership separation of proprietary trading will primarily be to have a preventive effect going forward.

5.4.1.4. Conflicts of interest

Ownership separation is the cleanest and most effective way to eliminate conflicts of interest arising from engaging in certain activity combination. This would be particularly valuable for proprietary trading (option C), as it is particularly prone to conflicts of interests (see section 5.3 and Annex A6). It is also valuable for underwriting, as the interests of the bank as underwriter and as loan provider are typically not aligned. The benefit would be less pronounced for market making even though principal-agent problems may arise there as well, particularly in more opaque OTC markets.

All reform options accordingly reduce conflicts of interests but to a varying degree. Option C has the narrowest scope but nevertheless addresses the activity where conflicts of interest and potential bank culture contamination are the most significant.

5.4.1.5. Capital and resource misallocation

With the most risky trading activities no longer on its balance sheet, a financial institution is able to entirely focus on and allocate more resources to its core and traditional role of lending to the real economy and acting as an intermediary between savers and borrowers.

All three reform options reduce capital and resource misallocation, but options F and I arguably more than option C given their broader activity scope. Even so, given that underwriting does not give rise to a trading culture and may be underprovided when distanced from the public safety net, the justification for ownership separation of underwriting to enhance capital allocation is weak.

5.4.1.6. Competition

Ownership separation eliminates implicit cross-subsidies and ensures effective market discipline on the trading entity. Competition on the merits between small, medium-sized and large banks and across Member States is accordingly promoted. Smaller and medium-sized banks benefit, as they currently are not likely enjoying equally high implicit public subsidies (see Annex 4.1 and A4.2). In that respect, the ownership separation of proprietary trading, market making, and complex securitisation (option F) or even of all wholesale and investment banking activities (option I) improve competition in the EU banking sector more than just separating proprietary trading (option C). However, given that those options are far-reaching (prohibiting the concerned EU banking groups from engaging in many or all WIB activities) option C would be easier to reconcile with the significant diversity in EU bank business models.

5.4.2. Social costs

Ownership separation would in general lead to more significant losses of economies of scope compared to options based on subsidiarisation reform options. It would also trigger a migration of certain activities toward non-bank credit intermediaries (so-called "shadow banking entities").

5.4.2.1. Foregone economies of scale and scope

Ownership separation is the most intrusive form of separation as it results in the elimination of economies of scale and scope (while for example under subsidiarisation economies of scale and scope can be maintained to a significant extent). Whereas all other reform options maintain the universal banking model and merely intend to introduce more structure into the group, ownership separation reform options are no longer compatible with the universal banking model as it exists today in the EU, in particular when considering a broader set of activities, such as in F and I.

Ownership separation would eliminate the extension to the prohibited activities of the implicit subsidy resulting from the safety net. Those activities would accordingly become more costly, which would, other things being equal, lead to a reduction in the scale of those activities and by extension a reduction in the systemic risk.

As a result of ownership separation, the loss in economies of scope (and to a certain extent scale), would be greater than for other options (subsidiarisation without or with additional rules and restrictions) and would have the highest effect in terms of increased private funding and capital costs. These costs tend to increase with the strength of separation for the entities that include the most risky activities. If some of these activities perform an important role in the economy these additional costs may have further efficiency effects in other segments of the economy.

However, social costs appear relatively limited for ownership separation of proprietary trading (option C). A prohibition of proprietary trading is unlikely to lead to a significant loss of efficiency for large banks, as proprietary trading comprises a relatively small part of those banks' activities, and the social cost is in any case limited given that the activity can be and is performed by non-banks. Given the size of proprietary trading, there would be no lost economies of scale, as at very large asset levels economies of scale are likely to be exhausted (see Annex A9). One type of economies of scope loss is the loss of diversification benefits (of capital and profits). Channelling of profits and capital between the bank and the proprietary trading activities would no longer be possible and therefore diversification benefits between proprietary trading activities and banks' remaining activities would be lost. It is doubtful that such scope for diversification would be very substantial for the remaining bank, given that larger banks do not engage substantially in proprietary trading activities. Furthermore, given the very risky nature of proprietary trading and the low complementarity of proprietary trading with other traditional activities of banks, proprietary trading activities would rather lead to diseconomies, rather than economies, of scope through excessive complexity, risk taking and even systemic risk. However, there would be lost cost economies of scope as separating proprietary trading from market making activities may lead to some duplication of infrastructure.

Genuine economies of scope related to lost risk diversification opportunities are possibly non-negligible for market making, given the importance of market making for a bank’s revenues (option F).

Risk diversification costs are particularly pronounced for reform option I as it foresees the ownership separation of all WIB activities, including activities that are of great importance to the real economy and which would deserve to be linked to the public safety net (and hence enjoy certain subsidies), given that they address genuine market failures.[63] Given that the activities would be banned from the group completely, the risk diversification economies of scope would be entirely lost under options F and I.

For those reasons option C leads to substantially less social costs compared to options F and I.

5.4.2.2. Operational costs

Ownership separation is unlikely to create high operational costs for the banking group under consideration. It would actually lead to lower operational costs from the perspective of the bank, as once separated there would no longer be any costs associated with running that activity within the banking group (e.g. separate boards, risk management, monitoring, etc.). Naturally, from a societal point of view, there would still be operational costs associated with running the company carrying out the separated activities, but operational costs associated with continued coexistence within the group would no longer arise. These findings hold for all three options as a whole without any meaningful variations in importance.

5.4.3. Conclusion

Ownership separation has the potential to be the most effective set of options in terms of achieving the specific objectives of facilitating resolution and limiting moral hazard, conflicts of interest and capital and resource misallocation. However, it is also the set of options that may come with the highest social costs in terms of foregone economies of scope and it would remove the separated banking activities from the perimeter of the regulated and supervised banking group into the shadow banking sector (see Annex A13).

In general, the broader the scope of activities concerned, the stronger the potential benefits and costs. Whether the costs appear proportionate to the benefits achieved in addressing the objectives of facilitated resolution and reduced moral hazard, conflicts of interest and capital and resource misallocation differ between the options. As regards option C, the potential benefits would be pronounced, given the risks, complexity and interconnectedness associated with proprietary trading. Proprietary trading is also the activity that is most prone to conflicts of interest, given that it is the most distanced from client, real economy oriented activity. Even if banks currently engage in modest levels of proprietary trading compared to earlier years, ownership separation of proprietary trading ensures that a potential new surge will not materialise within the largest and most complex EU banking groups. Ownership separation would accordingly facilitate resolution and reduce moral hazard, conflicts of interest and misallocation of capital and resources to a significant extent within the largest and most complex EU banking groups. Proprietary trading activities do not provide substantial value added to the economy, can and is performed by non-banks, and foregone economies of scope seem limited. Hence, the social benefits associated with option C clearly outweigh the social costs. Moreover, given that option C delivers stronger benefits than option B with little additional marginal cost and with similar implementation difficulties, option C is superior to option B.

A significant positive balance between benefits and costs is more difficult to argue for options F and I. Whereas the separation of market-making (option F) would indeed yield substantial social benefits, the social benefits associated with other wholesale and investment banking activities (e.g. underwriting) are less pronounced. Ownership separation would also yield higher social costs in that regard, given that these activities can not all be performed by non-banks and that stand-alone investment banks may not be in a position to readily step in as a substitute. As highlighted above, market making and underwriting are activities that can and do contribute to the financing of the real economy, notably by providing liquidity to secondary markets, although the latter may require qualification given the financial crisis experience (see section 5.8 below as well as Annex A6). Market making and underwriting are perceived as socially useful activities. Being socially useful does not imply that they need to be artificially promoted by benefiting from being performed by the deposit entity enjoying access to the public safety net. However, it does imply that the ownership separation of investment banking activities –effectively eliminating economies of scope such as risk diversification – may be less obvious to justify for activities other than proprietary trading.

At this stage of economic knowledge and evidence and given the relatively high uncertainty, cautiousness as regards the activity scope for ownership separation seems appropriate. Ownership separation with a wide activity scope would limit the flexibility to cater for cultural and structural differences within and across national banking systems and risks reducing the current diversity in successful business models across the EU. Given the potential social costs associated with ownership separation, there is a strong imperative to be cautious as regards ownership separation of a broader set of activities at this current juncture. Ownership separation as per option F and I are likely to generate social costs, uncertainty and unintended consequences that would be disproportionate to the benefits, in particular as the preceding analysis has highlighted that a less intrusive reform option (E) delivers similar benefits at significantly lower social cost. Moreover, ownership separation of all wholesale and investment banking activities may effectively give rise to an under-provisioning of selected core and relationship-oriented banking activities (in particular those oriented at SMEs) and as such does not seem coherent with the overall policy objective of promoting growth and jobs.  

Hence, the Commission services believe that only option C with ownership separation of a relatively narrow activity scope provides a good net balance of social benefits and social costs and is worth exploring further.

5.5. Assessment of retained reform options

Table 3 summarises and collects the qualitative assessments of the different reform options. The result of the assessment and comparison of the different options result in reform option E being found superior to all other options, with reform option C being superior to other reform options, except for option E. Whereas option E yields higher social benefits than option C, it comes at a higher social cost, such that on balance option C is still worthwhile retaining.

Table 3: Overview of options

|| Reform options

|| No policy change || A || B || C || D || E || F || G || H || I

EFFECTIVENESS || || ≈ || ≈/+ || + || ≈ || ++ || ++ || ≈ || ++ || ++

Facilitate resolution || 0 || ≈ || + || + || ≈ || ++ || ++ || ≈ || ++ || ++

Facilitate management || 0 || ≈ || + || + || ≈ || ++ || ++ || ≈ || ++ || ++

Reduce moral hazard || 0 || ≈ || ≈/+ || + || ≈ || ++ || ++ || ≈ || ++ || ++

Reduce conflicts of interest || 0 || ≈ || ≈/+ || + || ≈ || +/++ || ++ || ≈ || ++ || ++

Reduce capital and resource misallocation || 0 || ≈ || ≈/+ || + || ≈ || ++ || ++ || ≈ || ++ || ++

Improve competition, retain diversity || 0 || ≈ || ≈/+ || + || ≈ || ++ || ++ || ≈ || +/++ || ++

EFFICIENCY || || ≈ || ≈/- || ≈ || ≈ || - || - - || ≈ || - - || - -

Reduced economies of scale and scope || 0 || ≈ || ≈/- || ≈ || = || - || - - || ≈ || - - || - -

Increased operational costs || 0 || ≈ || - || ≈ || ≈ || - || - || ≈ || - || -

Note:      ++: strongly positive; +: positive; - -: strongly negative; -: negative; ≈: marginal/neutral; ?: uncertain;

n.a.: not applicable.

This section accordingly compares these two options in terms of effectiveness (social benefits), efficiency (social costs) and coherence. 

5.5.1. Effectiveness

Option C foresees the complete ownership separation of proprietary trading, which accordingly would have to be divested from the banking group altogether (or wound down). This is a stronger degree of separation compared to option E, as option E leaves proprietary trading within the group. However, it is also a more narrow option in terms of activity scope, as it would not separate other wholesale and investment banking activities, such as market making, complex securitisation and complex derivatives.

Overall, the benefits of the stronger degree of separation of option C have to be weighed against its narrower activity scope. Accordingly, all risks associated with those additional activities (e.g. market-making, complex securitization and complex derivatives) that are affected under option E would be left unaddressed by option C. This is quite relevant, as existing evidence suggests that large banks currently engage in limited proprietary trading but do engage in significant market making, securitisation, and derivatives activities. Option C is therefore less effective in delivering the identified social benefits for TBTF banks compared to option E (see sections 5.3 and 5.4). Its main benefit would therefore be to prevent any future expansion of such activity and the corresponding concerns in terms of impediments to resolution, excessive risks, conflicts of interests and inappropriate implicit safety net coverage.

The effectiveness of option C is furthermore handicapped by potential difficulties related to implementation, notably as regards defining proprietary trading and objectively distinguishing it from market-making.

In order to make an informed choice between the two retained options requires taking into account (1) potential fundamental implementation problems, and (2) possibilities to improve reform effectiveness through tailored and framed supervisory action.

5.5.1.1. Implementation problems

Option C appears particularly challenging in terms of implementation, given the need to distinguish proprietary trading from market making, which is not a concern that arises when implementing option E. Nevertheless, option E is not immune to its own implementation challenges, given that it requires a subsidiarisation that upholds the integrity of the separation objectives. More specifically:

· Challenges of implementing option C: The main problems arise from the need to define and distinguish proprietary trading from market making and hedging, given that such a distinction seems to rely on the private intent of the trader. As regards the definition, France, Germany and the U.S. have proposed to adopt a restriction of proprietary trading activities which excludes market making activities. None of these countries, except for the U.S., have yet defined proprietary trading in an operational way. In the definition of proprietary trading special importance should be given to the way definitions can be operationally and legally enforced and supervised.

Distinguishing proprietary trading from market making is difficult both in theory and practice.[64] Although traders involved in the actual trade are able to identify any given transaction as being of a market making or proprietary trading nature, such an intent-driven distinction is not easy from the perspective of a manager, regulator, supervisor, creditor, or judge.[65] Indeed, a market maker might legitimately choose to take a long position in an asset either in anticipation of client demand to allow the order to be fulfilled quickly or to facilitate a quick sale by a client of an illiquid asset. Therefore, a proprietary trade and a trade pursuant to market making are indistinguishable based solely on the objective and observable features of the trade itself.

The precise definition of proprietary trading versus market making would also have an impact on the ability of banks to effectively perform market making activities. Market makers require substantial discretion in their ability to buffer unexpectedly large supply and demand imbalances in order to provide immediacy to their customers and to effectively perform their activity. As a result, a narrow definition of market making activities may constrain the ability of market makers of a banking group to service heightened demands for immediacy given the limits they would have to comply with.

Proprietary trading is also difficult to distinguish from treasury management operations.[66]  In particular, these difficulties arise when considering the hedging activities that banks can engage into to mitigate the risk of their commercial activities. If banks are no longer able to perform such hedging activities they would be deprived of engaging in efficient risk management. For this reason, the U.S., France, Germany and UK have made exemptions for risk-mitigating hedging activities. However, such exemptions may well raise implementation problems as well as monitoring/supervision problems.[67] In addition, banks may also by-pass the ban on proprietary trading through such hedging activities by masking their positions as simply hedges of permitted trades, though in reality such hedging positions are their real purpose.

In sum, it would therefore be feasible to ban dedicated proprietary trading activities that are clearly disconnected from any customer-related transaction. However, for a broader definition, genuine proprietary trading may look like market making, and genuine market making may in fact look like proprietary trading. Therefore in any broad definition there is scope for "type I and type II errors".[68]

· Challenges of implementing option E: Whereas option E at first sight seems easier to implement as it avoids distinguishing proprietary trading from market making, option E is still left with the complexity to ensure that proprietary trading does not take place alongside other permitted activities (notably treasury and liquidity management) in the deposit-taking entity. Whereas the supervisory burden is lighter than for option C (as proprietary trading and market making are treated similarly), the implementation concerns are therefore not avoided altogether.

Given the difference in separation strength, option E faces the implementation challenge to design and enforce the separation, triggering greater compliance costs. The activities would remain within the group and regulation and supervision would have to be established to govern the economic, legal, governance and operational links that would remain in order to uphold the separation objectives over time.

Option E also faces the challenge that the reform may be (perceived as) less effective, in particular in crisis times, due to informational and reputational contagion.[69] Even if the separation would be fully effective in crisis times, customers and creditors may not be convinced and may retreat from engaging with both entities if only one gets into significant difficulties.

Overall, the challenges of defining and distinguishing proprietary trading from other activities such as market-making and risk hedging are significant. However, banning dedicated proprietary trading desks is of course feasible, albeit easy to circumvent and less effective (e.g. London whale incident at JP Morgan). The main challenge associated with option E is to set up, enforce, and maintain the integrity of the separation over time.

5.5.1.2. Improving effectiveness through tailored and framed supervisory action

Option C suffers from an effectiveness gap compared to E, especially if the implementation problems considered above and a more modest ambition of merely banning dedicated proprietary trading desks are taken into account. There are nevertheless ways of raising the limited effectiveness of option C. It could for example be complemented by a procedure for separating additional trading activities either without further supervisory discretion (ex ante) or following further supervisory review (ex post). These two variations are examined in turn.

· Option C combined with ex ante separation of market-making (“C+ex ante”): The first option would be to combine a prohibition of proprietary trading with an ex ante separation of other trading activities (effectively combining options C and E). Given the analysis in section 5.3 that highlighted the benefits of option E, the logical consequence would thus be to separate notably market making, complex securitisation and complex derivatives. Such a combined reform option would resemble the current state of play in the US, where a Volcker rule prohibition is being imposed on bank holding companies in which the deposit taking commercial bank arm face activity restrictions that are largely similar to E and where the economic links of that entity with other group affiliates that perform other investment bank activities are subject to quantitative limits and qualitative restrictions. The commercial bank arm of a bank holding company can engage in dealing and underwriting (subject to caps and conditions and depending on type of securities), certain derivatives, and securities brokerage (see box 2 below);

· Option C combined with ex post separation of market-making following a framed decision by the relevant supervisor (“C+ex post”): Another option is to combine a prohibition of proprietary trading with a potential ex post separation of market making, complex securitisation and complex derivatives if certain metrics are exceeded. Under this approach, the supervisor would have the possibility to require the subsidiarisation of market making, complex securitisation and complex derivatives after careful scrutiny and analysis. To ensure a consistent application of the procedure throughout the Internal Market and create legal certainty for banks and other stakeholders, this procedure would need to be framed. It would thus rely on a framework of assessment under which the supervisor would have to review market making, complex securitisation and complex derivatives (and possibly other trading activities) for the banks under consideration and take further action should specific relevant and well-defined criteria be fulfilled. In particular, a rebuttable presumption can be created according to which if such criteria are fulfilled then subsidiarisation should result. This latter rebuttable presumption bridges the gap between ex ante and ex post subsidiarisation. A two-step procedure is thus created: (1) a mechanism is created that triggers an obligatory review by supervisors of market making, complex securitisation and complex derivatives; in combination with (2) guidance provided to supervisors to ensure that subsidiarisation according to certain rules is triggered when certain criteria are fulfilled. The criteria and corresponding thresholds and triggers need to be elaborated by the EBA and agreed with the Commission as the guardian of the Treaty and Internal Market.  

The trigger mechanism for the review by supervisors would be based on a reporting requirement imposed on relevant banks. Information to be reported could include amongst others the relative size of trading activities, the leverage of trading assets, whether the trades are being triggered by customer demand or not, trading book exposures, relative importance of capital requirements for counterparty and market risk as a percentage of total regulatory capital requirements, identification of trading account, trading units, activities and changes in activities, compensation/bonus allocations for staff in different trading and other business lines as well as the interconnectedness of the bank.

The value added of this complement compared to option C and to existing legislation is that it would set a uniform standard across the EU for if and how to review trading activities, under what circumstances subsidiarisation of market making, complex securitisation and complex derivatives should be mandatory, and how that subsidiarisation should be implemented. It would also create legal certainty for banks and other stakeholders, as they would know when and how the rules would apply and be ensured that they are applied in a similar fashion to all banks concerned across the EU. For these reasons this framework would also complement and provide a value added to the other tools supervisors already have through CRDIV and BRRD to impose structural measures on banks (see more in Section 2 and Annex A3).

With regard to activities other than market making, complex securitisation, and complex derivatives (such as underwriting, private equity, venture capital, etc.), the supervisor would have discretion whether to require separation and if so, in what form (subsidiarisation and/or additional restrictions).[70] However, the foreseen framework would also include a more general provision providing supervisors with the specific right to intervene, and require separation as appropriate, when there are concerns that activities would involve or result in a material conflict of interest, when there are concerns about distorted incentives such as, for example, moral hazard as a result of the implicit subsidy, and when there may be a serious concern that the activities of the bank would pose a systemic threat to the financial system.

The advantage of the above is that it would only apply to those banks that represent the greatest risk to the financial system because of resolvability concerns or distorted incentives. By clearly framing the supervisory powers as per above, risks about geographical arbitrage and fragmentation across the internal market are reduced.

The above complements to option C would reduce (ex post) or eliminate (ex ante) the gap in effectiveness compared to option E and would ensure that banks whose trading activities do not pose significant risks are not subject to (socially undesirable) subsidiarisation.

5.5.2. Efficiency

In terms of foregone economies of scale and scope, option E, by virtue of its broader activity scope, has a larger effect on foregone economies of scope, in particular through reduced risk diversification benefits, although risk diversification is being partially retained within regulatory limits. A similar potential loss of economies of scope would not materialise in C, given the currently limited importance of proprietary trading activities. However, the mandatory separation of proprietary trading would result in higher cost of duplication of trading infrastructure compared to E (as such infrastructure can no longer be shared between proprietary trading and market making activities in option C). As proprietary trading and market making are kept together under option E, such a duplication of infrastructure would not be required under E.

There are ways of reducing the potential efficiency cost associated with option E should that cost be considered excessive. This could be done by for example exempting from the separation requirement certain instruments where existing markets are particularly shallow and/or illiquid; i.e. reform option "E-".

In terms of operational costs, option E compared to option C and from the banking group’s perspective would result in greater on-going operational costs associated with running the separated trading entity within the group. Although, from a social point of view, the proprietary trading activities to be banned still need to be housed elsewhere, operational costs will be higher under option E. Whereas option E- will not trigger additional operational costs compared to E, option C+ will require significant regulatory energy to frame and elaborate the decision whether and when to subsidiarise certain trading activities, to take the subsequent decision after implementing the ex post process, and in following up the potentially divergent outcomes across banking groups. For the banks that require subsidiarisation, the operational costs of complying with the separation will apply. All in all, C+ will lead to the highest operational costs of all retained options and variations thereof.

5.5.3. Coherence

Reform options relying more on ex post separation subject to supervisory judgement (option C+ex post) in principle may give rise to divergent outcomes to banks sharing the same characteristics but subject to different supervisors, if the framing of such decisions is weak and therefore allows for flexibility. If so, banks can try to exploit such divergences by relocating activities (see section 2.4) to Member States where the supervisor has taken a more lenient approach. Such an effect would not be coherent with the overall ambition of the Commission following the financial crisis to reduce the potential for regulatory arbitrage. Furthermore, wide divergences of de facto application of EU rules would run counter the efforts to establish a Single rulebook.

Conversely, reform options with higher degree of ex ante separation and more limited reliance on supervisory judgement (option E-, option C+ ex ante) would not be associated with the same level of risk, given that separation will be more firmly set out in primary law and that less room will be granted for supervisors to exercise judgement.

Even so, there are ways of reducing the arbitrage risks associated with differing supervisory judgements. The process for carefully framing the guidance should aim to ensure as consistent an application as possible throughout the internal market. Furthermore, in practice the potential for wide divergences is limited by the de facto very limited number of supervisory authorities involved. The banking groups that on the basis of historical data look likely to be required to separate would in the majority of cases be headquartered in a Member State that will participate in the Single Supervisory Mechanism.[71] The remaining banks come from two other Member States (UK, SE).

Accordingly, while the problem of potential differences of application is a bigger issue for options with a larger ex post supervisory judgement element, the risk can be limited by the possibility to frame the subsidiarisation decision (in particular by creating a rebuttable presumption for subsidiarisation) and the de facto very limited number of supervisors concerned.

5.5.4. Conclusion

Because of its significantly broader activity scope, option E delivers greater social benefits compared to option C and hence may be deemed more effective, at limited additional efficiency cost. However, there are a number of variations to option C and E that can be considered as roughly equivalent.

The ex post complement to option C could reduce the effectiveness gap with option E, especially if that process was clear, transparent and predicable in terms of result (i.e., the more supervisory discretion is framed in terms of when and how it would be exercised the closer to E). It would nevertheless raise significant operational costs for supervisors to elaborate and implement the required framework to take bank-specific subsidiarisation decisions. The combination of C with E or E- , i.e. “option C+ex ante” above, would also be effective.

The “option C+ex post”, “option C+ex ante”, and E, albeit not identical either in terms of effectiveness, efficiency and coherence can be considered to be preferable to no intervention when balancing the net benefits in terms of improved financial stability (due to increased resolvability, reduced moral hazard, reduced conflict of interests, increased competition and diversity, an adequate culture geared towards better serving the interests of the real economy and spur growth, etc.) against the increased costs in terms of foregone efficiencies. However, beyond a largely qualitative economic analysis other considerations of a more political nature, such as timing of the reform, expected views and position of co-legislators etc., need to be taken into consideration before making a choice between these acceptable and justifiable options. As a result this report does not state a preference between the three options.

In conclusion, the final balance between the additional benefits in terms of improved ex ante financial stability (due to increased resolvability, reduced moral hazard etc.) versus the increased costs in terms of foregone economies of scale and scope and operational costs in this instance is more a matter of political choice than technical ranking.

5.6. Retained options aimed at increasing the transparency of shadow banking

As indicated previously, the decision to separate the banking entities entail risks that parts of the banking sector shift to the less regulated shadow banking sector. To make sure that this risk is properly monitored, a series of options have been assessed in Annex A13. A combination of options has been retained that will complement the options retained on the structural separation aspect.

Transparency towards regulators: Counterparties to SFTs will be required to report the details of such transactions to trade repositories. The reporting obligation will cover all market participants, regulated or unregulated. This reporting will lead to a substantial increase in the transparency of securities financing markets which a key sources of liquidity. This will also facilitate regulators' access to market data and avoid the need to compile unstandardized and dispersed information from different regulators. It would allow for complete and timely information to be reported (e.g. principal amount, currency, type and value of collateral, the repo rate or lending fee, counterparty, haircut, value date, maturity date), therefore making it possible for regulators to perform a well-timed comprehensive monitoring of the market developments. The periodic publication of aggregate data by TRs can be an additional benefit as it will improve the overall data available to investors but also for research projects. The FSB has already recommended the collection of frequent SFTs data with high level of granularity. Meanwhile, the ESRB has concluded in a paper that 'trade repository collecting transaction data based either on trade-by-trade data or exposure data is likely to be ideal for a comprehensive assessment of risks'.

Transparency toward fund investors: Fund managers will be required to disclose in their periodical reports (semi-annually and annually) the use they make of SFTs and other financing structures to their investors. They will also be required to notify in the pre-contractual documents such as the prospectus the limits they will apply as regards their intended use of SFTs. This new information will give insight to the investors on the transactions that the fund has been involved in over the previous reporting period. It is a means for the investors to check the performance of the fund and other indicators regarding the risks or costs. More generally, it gives the possibility to verify that the fund's investment strategy has evolved as announced in the prospectus. Investors will also have knowledge, prior to their investment, on whether SFTs form part of the investment strategy pursued by a fund. They will be able to measure the expected risk and reward profile linked to this activity. Their ability to compare the investment proposition of different investment funds will increase. In addition, investors will receive increased assurance that managers will not use to a greater extent than announced in the fund rules.

Transparency of rehypothecation: Specific transparency requirements will be put in place in order to increase the contractual and operational transparency. These include contractual agreement on rehypothecation, prior consent to rehypothecation, transfer of assets to an account of the party which intends to rehypothecate them.  This would ensure that clients or counterparties are fully aware of the potential risks involved, in particular in the event of default of the receiving counterparty. Furthermore, prior to the actual rehypothecation the financial instruments received as collateral have to be transferred to an account opened in the name of the receiving counterparty, which would also help prevent a future crisis scenario, where investors are uncertain about their rights, thus contributing to financial stability. Such rules are consistent with existing market practice in major securities markets in the EU. The FSB has developed a similar policy recommendation on sufficient disclosure to clients in relation to rehypothecation of assets.

5.7. Complementary aspects 5.7.1. Quantitative assessment of some benefits and costs

In general, it needs to be stressed that, given the inherent complexity and special nature of banking and given that many benefits and costs are dynamic in nature (often related to unobservable incentives), no quantitative model exists that can reliably, precisely and comprehensively estimate the social benefits and costs of structural reform proposals.

Moreover, quantitative analysis in this specific area is further constrained by the fact that intent-based proprietary trading cannot be easily distinguished from genuine market making activity (see also section 5.5.1.1). Accordingly, reliable data on the extent of proprietary trading and market-making is not available.

In order to support the qualitative assessment and comparison of reform options carried out above, the Commission services have attempted to quantify on a best-effort basis and to the extent possible some of the costs and benefits that could result from structural separation.

As one element, the Commission services invited EU banks in the context of the public consultation to model and estimate the impact of stylised structural reform scenarios on the group's balance sheet, profit and loss account and selected other bank variables between now and 2017, taking into account CRR/CRDIV and BRRD. A very limited number of banking groups responded. Moreover, the simulated impacts (e.g. funding costs, total costs, return on equity, ratings, etc.) differed substantially between respondents and gave rise to inconsistencies both within a given set of results as between different sets of estimated impacts (see Annex A11).

As another element, the JRC has attempted to assess some of the costs and benefits of structural separation (see Annex A10).  In relation to benefits, the aim has been twofold. First, to examine the impact on incentives on the basis of the distribution of gross losses across stakeholders. Second, to estimate the reduced contingent liability for the resolution process following the estimated behavioural response of banks to shrink or recapitalise the trading entity post reform. In relation to costs, the analysis has focused on estimating the increase in the banks’ private funding cost post-reform.

It is important to note that the costs and benefits that the JRC has been able to quantify are not comprehensive and are dependent on underlying assumptions (how to separate balance sheets, behavioural responses of banks, required rates of returns for the different funding sources under different scenarios, etc.). Moreover, important social benefits (including the reduction in the probability of systemic crisis and in contagion as well as the impact of the reform on avoiding conflicts of interest, misallocation of resources and facilitating supervision, etc.) and costs (such as economies of scope and scale, impacts on liquidity of secondary markets and legal costs) have not been quantified and modelled.  Benefits and costs are estimated in a scenario where the resolution framework is considered to apply effectively to the entire considered sample of banks even before any structural separation of these banks (bail-in is fully effective).[72]

With all these caveats in mind, the stylised JRC analysis indicates the following for the specific benefits and costs measured (see Annex A10 for details):[73]

In terms of benefits, the results of the analysis suggest that trading entities incentives are better aligned, as a higher part of their losses would be absorbed by their shareholders and creditors and would be not passed through to the rest of the financial system. In addition, the contingent liabilities faced by the resolution process for the banking sector in case of crisis will be reduced following structural separation. By way of illustration, in a crisis of similar severity to that of 2008, potential gross losses are estimated to decrease by up to €125 billion, while the share of gross losses absorbed by the trading entity would increase. The contingent liabilities for the resolution process could be reduced by up to €58 billion, as there would be lower losses due to better aligned incentives and reduced risk taking for the trading entity.

In terms of costs, it is important to note that increased funding costs (under both the subsidiarisation and ownership separation scenarios) is an intended outcome of structural reform, as it leads to a more efficient pricing of risk reflecting enhanced market discipline. By way of illustration, the trading entities subject to separation are estimated to face an increase in funding costs of up to 9 basis points (in a conservative scenario where risk premia do not decrease following reduced risk taking and increased capitalization of the trading entity), which will generate an increase in the average cost of funding for the affected group as a whole of up to 3 basis points. As banks subject to separation are estimated to hold about 55% of the assets of the EU banking system, this translates into a sector-wide funding cost increase of up to 2 basis points – equivalent to approximately up to €7 billion per annum.

As explained in section 5.2.2, the increase in the funding cost of the trading entity is an intended outcome of structural reform and should not strictly be considered a social cost. Nevertheless, the Commission services have tried to estimate the reduction in GDP growth that would follow from passing on the estimated funding cost increase following conservative assumptions (i.e. assuming that the entire funding cost increase is passed on into higher borrower rates rather than lower profitability or employee salaries) and making use of the QUEST DSGE model with a banking sector.[74] The model suggests that an increase in private funding costs of 2-3 basis points translates – under the retained conservative assumptions – into a reduction of the long term annual level of GDP of between 0.08% and 0.11%. This clearly represents a worst-case scenario, as the increase in the funding cost of the trading entity will not be fully reflected in interest rates for credit to the real economy. In addition, one should note that the negative effect on the level of GDP in the initial years of the simulations are less than in later years, i.e. in the long term. Thus, the average annual effect on GDP would be less than the long-term effect. Moreover, these figures represent a gross cost, not a net impact, as they do not incorporate any positive impact on GDP growth stemming from social benefits resulting from facilitated resolution, increased financial stability, etc. (see above). These results are fully compatible with the cost estimation methodology presented by the European Commission in the Impact Assessment of the BRRD.[75]

5.7.2. International aspects 5.7.2.1. Territorial scope of the reform

The objective behind determining the territorial scope is (1) to ensure that all group entities within the EU are covered by EU rules and (2) to reduce potential contamination effects stemming from EU subsidiaries and branches established in third countries. Accordingly, separation should apply to (i) EU banks, their subsidiaries and branches, including in third countries; and (ii) subsidiaries in the EU of banks established in third countries.

EU branches of third country banks operate under authorisation and conditions imposed by the Member State where the branch is established with the limitation that they cannot be treated more favourable than branches of EU banks. They do not enjoy the freedom to provide services or the freedom of establishment in other EU Member States. The conditions regarding structural separation vis-à-vis these branches should thus also be left for the Member State hosting the branch.

5.7.2.2. Impact on competitiveness of EU banks

Both the preferred reform options mirror rules that are either already in place in the US or are in the process of being implemented:

· Option C corresponds to the Volcker Rule of the Dodd-Frank Act, which prohibits US banks from engaging in proprietary trading, as well as curtailing their investments in certain funds (See Annex A1 for further detail); and

· Option E would be similar to the rules applying to US banking groups. While US banking groups (bank holding companies or financial holding companies) may engage in a wide range of financial activities, the part of the group that takes insured deposits (“insured depositary institutions”) still face activity restrictions that limit their focus to core banking activities (e.g. taking deposits, lending) and other incidental activities (e.g. custody and asset management). Furthermore, the US also has rules governing transfers between the different parts of a banking group, which are aimed at isolating the insured depository institution from excessive risks arising from the larger financial firm of which it is part (BHC, FHC) and to prevent the transfer of the subsidy arising from federal assistance to non-depository financial institutions. These firewalls have been strengthened as part of the DFA (See Box 2 and Annex A1 for further detail).

Implementing either option is accordingly unlikely to have material negative effects on the competitiveness of EU banks vis-à-vis US banking groups.[76]

5.7.3. Institutional scope 5.7.3.1. Thresholds

As argued above, the Commission services believe that an approach that combines an accounting-based methodology with an additional systemic risk metric is the best approach to determine the scope of banks subject to potential structural separation. The systemic risk is captured by the EU global systemically important institutions. Regarding the accounting-based approach, the second step has been to review the threshold and definition of trading activities recommended by the HLEG.

The Commission Services have in particular decided to analyse the definitions presented in Table 4, also considering limitations due to the need to rely on publicly available accounting data.

Table 4: Definitions of trading activities

1 || HLEG definition

2 || Exclusion of Available for Sale (AFS) assets under the assumption that they are mostly held for liquidity purposes

3 || Gross volumes (assets + liabilities) of securities and derivatives held for trading  (to focus on market and counterparty risk)

4 || Net volumes (assets – liabilities) of securities and derivatives held for trading  (to focus on market risk)

The Commission services do not believe that thresholds necessarily need to be set in legislation.[77] Nonetheless, the Commission services have carried out analysis with the aim of considering different absolute and relative thresholds. That analysis notably includes carrying out tests to ensure that any threshold captures banks sharing similar characteristics (“clustering analysis”).

The HLEG recommended an absolute threshold of EUR100bn of trading assets and a relative threshold of trading assets to total assets in the range of 15-25%. Whereas the thresholds suggested by HLEG (Option 1) would be effective in selecting systemic banks, it would also include a relatively wide set of banks that do not appear to be of systemic importance (Annex A8). The disproportionately large scope of selected banks finds its origin in the consideration of AFS assets for the purposes of defining trading activities. However, banks tend to largely hold AFS assets mainly for liquidity purposes. The other options therefore have excluded AFS assets.

Having excluded AFS assets from the definition of trading activities and having also modified the definition of trading activities so as to focus more on market and counterparty risks, the Commission services have first analysed what would constitute thresholds for the various definitions equivalent to those suggested by the HLEG, and then applied clustering techniques to further improve them (Annex A8). All definitions are based on publicly available balance sheet data and have been computed as averages over a 6-year 2006-2011 period. The results are shown in Table 5 below.

Table 5: Proposed thresholds based on cluster analysis (Annex A8)

Option || THRESHOLDS || SELECTED BANKS || SELECTED BANKS BY SIZE

|| Trade Activity EURbn || Share Trade Activity || Number || % of the sample || % of the sample in total assets || Large || Medium || Small  (then exempted)

1 || 80 || 20% || 52 || 21% || 75% || 19 || 16 || 17

2 || 70 || 15% || 32 || 13% || 60% || 13 || 10 || 9

3 || 70 || 10% || 36 || 15% || 65% || 13 || 16 || 7

4 || 30 || 8% || 33 || 14% || 51% || 11 || 7 || 15

Note: Large banks are defined as having total assets above EUR500bn. Small banks are defined as having total assets below EUR30bn. “Trade activity” refers to the different definitions as set out in Table 4. The sample of banks considered is 245. In definition 3 (option 3), the actual amount refers to half of the total gross volumes.

Analysing the banks selected by these definitions, it can be noticed that the second option, that exclude AFS from the HLEG definition, does not include all the EU banks that are currently considered as being of global systemic importance, and hence would be of limited effectiveness.

Option 4 - which would focus on the difference between trading assets and liabilities (net volumes) - would also be of limited effectiveness, as it would exclude an even larger number of banks with significant trading operations.

The preferred option is therefore option 3 (with a threshold based on gross volumes of trading assets and liabilities). This option captures nearly all the EU banks considered to be of global systemic importance and those with significant trading activities.

In order to avoid that small banks are selected, an additional floor threshold is applied. The Commission services suggest using the floor for “significant institutions” used in the SSM-ECB Regulation (EUR30bn of total assets). Following this adjustment, the preferred option 3 selects 29 medium and large banks representing about 65% of the assets in the sample. Importantly, this list includes all but one of the European globally systemic banks (“G-SIBs”) identified by the FSB. Moreover, the list is not sensitive to small changes in the thresholds.[78]

Option 3 captures both market and counterparty risk typically associated respectively with proprietary trading and market making, and is hence relevant for all the option retained above. In addition, banks designated as G-SIIs should also be subject to separation as well.

5.8. Impact on stakeholders

This section discusses the general impact of bank structural reform[79] on the numerous stakeholders. The assessment remains at high level, as the precise impact will depend on the design of the height and location of the fence and the institutional scope.

In general, opponents to structural reform typically raise a series of concerns on bank structural reform. Allegedly, bank structural reform:

· raises the funding costs of banks and hence lowers economic growth;

· harms market liquidity and hence lowers economic growth (when subsidiarising market making and other trading activities);

· increases financial instability as it makes banking groups less resilient to shocks;

· unduly abolishes the European universal banking model that has proven successful and resilient throughout the crisis;

· reduces bank employment;

· fragments the internal market by allowing banking groups to ring-fence their core banking activities according to national borders;

· would not have avoided important bank failures of the recent financial crisis such as Lehman Brothers, Northern Rock, or Cajas;

· is ineffective in shielding the deposit taking entity from trading risks due to reputational and information contagion;

· reduces competition in the EU banking sector, and

· is inappropriately timed given the reform agenda.

This Impact Assessment already has addressed several of these statements in Chapter 2 and elsewhere. This section presents additional arguments where appropriate to qualify and/or rebut the above concerns. The Commission services consider that some concerns are not valid. Others are, but merely stress the reform costs whilst neglecting the reform benefits that are likely to outweigh the social costs. The additional communication difficulty is that private costs are borne by only a few, but vocal and large banking groups, whereas social benefits are enjoyed by many small taxpayers, only become obvious in the medium term, and are not easily quantifiable (improved banking culture, reduced systemic crisis incidence, reduced implicit subsidies). 

5.8.1. Impact on bank customers (investors, borrowers, etc.) and market liquidity 5.8.1.1. Market liquidity and funding costs

Funding costs are likely to be unaffected for the deposit taking entities as well as for medium-sized competitors. The latter gain market share and increase their profitability whilst at the same time making the market less concentrated and more competitive.

However, at unchanged balance sheet size, the impact of the reform would unavoidably lead to some increase in funding cost increases for the trading entity. In fact, modest increases of funding costs for the trading entity is an intended consequence of structural reform and reflects primarily reduced implicit subsidies for the trading activities (not foregone synergies between relationship banking and trading activities, as they are exhausted at a small fraction of the balance sheet size of affected banking groups, see Annex A9). The funding cost would reflect the underlying riskiness of the trading activities. As a result, incentives to take excessive risks are reduced, artificial competitive advantages of TBTF banks are being reduced, thereby restoring the level playing field with deposit taking banks focused on credit intermediation to corporates and households.

A particular concern is that the increased funding costs would lead to liquidity problems in the market. Given that bond markets rely on market makers to act as willing buyers and sellers, subsidiarisation of market making (in particular ex-ante subsidiarisation under reform option E) triggers concerns that sovereign and corporate debt markets become less liquid. “Liquidity” is often left undefined, but typically the fear is that bid-ask spreads may increase, increasing the costs to trade at any scale. Investor options will be reduced, as trading entities can no longer trade as much and as easily as before. Price discovery is made more difficult. And price volatility may increase, if professional position takers no longer spot price divergences from rational levels and correct them through speculation and trading.

However, these concerns appear exaggerated for several reasons.  First, the market liquidity concern neglects the fact that structural separation merely aims to reduce the implicit subsidies that distort the proper market functioning. Indeed, market prices are distorted when contaminated with implicit public subsidies and the banking system may in fact produce excess liquidity (as is evident from its rapid and unsustainable expansion).

Second, international and past experience shows that subsidiarisation would not have a substantial effect on liquidity. The US has 80 years of on-going experience with subsidiarisation of investment banking activities (including market making, underwriting, etc.), as deposit taking affiliates within a Bank Holding Company are not allowed to do other than “core banking activities” (see Annex A1 and Box 2). There is no evidence that suggests that US bond markets are less liquid than European ones and have been constrained in their development. To function properly, markets need a large number of independent traders. Subsidiarisation of market making exposes this activity to its underlying riskiness. This would ensure fair competition across stand-alone investment banks and investment banking arms within universal banking groups. These limitations may increase the number of market participants, making markets instead more liquid.[80] In the era when Glass-Steagall was in place, which went much beyond mere subsidiarisation reform options, the US economy has on average been thriving, compared to the current juncture.

What matters for the real economy is the level of the interest rate at which corporates and sovereigns can fund themselves, which is a function of the supply and demand for these securities, and whether bid-ask spreads are reasonable to allow a normal degree of trading transactions. The liquidity premium only makes up a negligible fraction of the interest rate level and reflects the extent to which the security can be exchanged.

Bid-ask spreads on sovereign bonds of DE, FR, UK and large corporates were for example already at negligible levels before broker-dealer arms of universal banks started to sharply increase their inventories and market making activities in the early years 2000. Bid-ask spreads on Bund paper have not decreased in the run-up to the crisis, although large European banks have sharply increased their inventories. Also seen in perspective, bid-ask spreads are relatively negligible compared to the interest rate level: 10 year Spanish government bond yields have more than doubled and increased from less than 3.5% in June 2006 to more than 7.5% in July 2012 (chart 12). Bid-ask spreads in the period June 2006 to August 2013 on average are 2bp (0.02%) and spiked at 12bp (0.12%) in June 2012 (chart 13). The above suggests that the willingness and ability of (private sector) market makers to influence the interest rate level is relatively limited. If anything, their procyclical behaviour and excessive liquidity provision sows the seeds of future crises in which they want to deleverage excessively.

Chart 12: Yield to maturity of 10 year Spanish government bonds || Chart 13: Bid-Ask spreads of 10 year Spanish government bonds

||

Source: Commission Services – Bloomberg || Source: Commission Services - Bloomberg

Third, the bulk of the securities inventories of banks do not correspond to sovereign or corporate debt, but rather to securities issued by other financial firms. Only a tiny fraction of the outstanding securities in the Eurozone has been issued by non-financial non-state issuers (chart 14). A similar observation holds for OTC derivatives, of which only a tiny fraction have non-financial firms as counterparts. The bulk of OTC derivatives are intra-financial sector derivatives (see also BIS (2013)).

Chart 14 – Outstanding securities in the Euro Area according to issuer

 

Fourth, Richardson (2013) notes that the issue of liquidity is more relevant in times of crisis than in normal times when liquidity is typically not a pressing concern. Private banks have not performed a significant liquidity role during crisis period and central banks have stepped in to assume the role of Market Maker of Last Resort (in covered bond markets, government bond markets, etc.) next to Lender of Last Resort. Finally, the liquidity concern is often built on the presumption that more liquidity is always and inherently positive, irrespective of its level, which is not the case.[81] Financial economics does not have a good explanation yet. One explanation for excessive trading is overconfidence, as in Odean (1999). Recent work presents models in which trading and trading speed can be excessive (Glode et al. (2012) and Bolton et al. (2012)). In these models, advances in IT do not necessarily improve the efficiency of financial markets. French (2008) estimates that investors spend 0.67% of asset value trying (in vain on average, by definition) to beat the market.

5.8.1.2. Households, SMEs, depositors and taxpayers

Households and SMEs that are clients of a banking group that needs to subsidiarise certain capital market activities would typically demand retail-type services which can be provided by the deposit entity. Hence, the increased funding cost for the trading entity is unlikely to affect borrowing conditions for households and SMEs. In fact, some capital market activities entail significant risk. Separating such activities from the deposit entity will reduce excessive risk taking and artificial balance sheet expansion and hence may lower the funding cost for the deposit entity.

Medium-sized competitors or new entrants that are not subject to mandatory separation may gain market share from large banking groups if artificial competition distortions in favour of too-big-to fail banking groups are being reduced. Hence, whereas some banking groups may face increased costs and may no longer serve certain customers, those activities may be picked up by smaller competitors that do not face structural separation requirements. Customers are accordingly not likely to be left unserved.

The impact of the preferred policy options on households, taxpayers and depositors will be positive. Households will predominantly be clients of the deposit taking subsidiary and hence will no longer be vulnerable to the same extent to trading and capital market related risks and conflicts of interests. At the same time, given that the funding cost of the deposit taking subsidiary is not estimated to go up, there should be no or a limited impact on borrowing rates or deposit rates. Importantly and although not very visible to each of them, they benefit from the reduced likelihood of taxpayer support for large banks. They will also benefit from being less vulnerable to conflicts of interest within the bank, as the culture of the deposit taking subsidiary should be more relationship-oriented and less affected by a transaction- and deal-oriented culture.

SMEs will primarily be banking with the deposit taking subsidiary whose funding costs should not be negatively affected by the separation of trading activities, hence there should be no impact on borrowing rates or deposit rates of SMEs. SMEs will also benefit from reduced conflicts of interest within the bank. Given that banks no longer will have distorted incentives to allocate their capital towards trading activities, SMEs are also likely to benefit from bank capital being more devoted to serving the real economy.

The preferred policy options do not imply the end of the universal banking model, but implements “structured” universal banking and largely retains any potential economies of scope (e.g. one-stop shopping) that may be of particular importance to SMEs (even though many SMEs are likely to have more than one banking relationship in order to foster competitive terms). The subsidiarisation is expected to reduce diseconomies of scope, such as excessive risk taking and conflicts of interest. 

5.8.1.3. Large corporates

Large non-financial corporates would no longer be vulnerable to conflicts of interests, as the bank no longer can make improper use of client-related information to increase its own profits (e.g. by means of proprietary trading). Similar to SMEs, while large corporates buys financial services from a wide range of banks, they would still be able to procure a full set of financial services from a single bank should they so wish.

In terms of banks attempting to pass on potential increases in their private costs resulting from separation, large corporates would be able to exercise pressure to reduce such a potential impact. They could access capital markets directly or borrow from a competitor bank.

Following separation, large corporates would be able to choose their bank of preference on the basis of a price/quality ratio comparison that is no longer distorted by artificial funding cost advantages.

The preferred policy options would most likely lead to increased funding costs for activities performed by the trading entity. This is indeed a logical consequence of the reform proposal, which aims at ensuring that the price of an activity internalises the associated risk. This provides the right incentives for the various banking activities carried out within a group. Accordingly, whether this could render certain business lines of EU-based banks less competitive should not be a decisive criterion and should be weighed against other overriding policy principles are being sought, i.e. greater financial stability, easier bank resolvability, less likelihood for public support, and greater benefits for the society as a whole.

5.8.2. Impact on bank employment

Structural reform, as it aims at addressing a misallocation of resources in the economy is likely to have some impact on bank employment. Some executives and traders would be negatively affected by the preferred policy options, as it would become more difficult to engage in proprietary trading given that the funding cost of the trading entity will reflect the risk of such activities.[82] Competition on the merits would keep economic rents lower. It is likely that the private cost increases are at least partially shifted to executives and traders, in terms of reduced remuneration (in particular bonuses).

Other employees of the deposit bank are unlikely to face a significant deterioration of their remuneration package. As relationship-based banking activities tend to be more labour intensive than trading activities, a refocusing of banks towards these activities should not imply a negative impact on employment and could even have a positive impact. The deposit entity will no longer have to act as buffer to absorb shocks and losses from the trading entity.

More generally, the flow of the most talented individuals into financial services may not be desirable, because social returns in other occupations may be higher, even though private returns are not. Philippon and Reshef (2012) find that workers in finance earn the same education- and other characteristics adjusted wages as other workers until 1990, but by 2006 the premium is 50% on average (i.e. the compensation of finance employees is 50% higher than expected). Top executive compensation in finance follows the same pattern and timing, where the wage is on average 250% higher than expected. An almost identical rise in the wage levels of finance workers relative to other business sectors was also seen in the 1920s. It was only in the late 1930s that banking pay fell to levels comparable with other industries where they stayed for almost 5 decades. Philippon (2013) claims that financial deregulation is responsible for the excessive wages in the finance sector, rather than the technology of modern finance.[83] 

Several authors (Baumol (1990), Murphy, Shleifer and Vishny (1991), Philippon (2013)) argue that the flow of talented individuals into financial services from other sectors of the economy (such as manufacturing, IT) may not be entirely desirable, because social returns may be higher in other occupations, even though private returns are not.[84] Many bank employees have strong science or engineering backgrounds. They could also increase the talent base for regulators.

In principle, structural reform is aimed at better directing bank capital and resources to those activities that finance the real economy. Proponents argue that without any structural separation, banks may be incentivised to allocate capital and human resources to trading and intra-financial activity and away from lending activity. Opportunities to engage in socially less useful activities in finance (speculation) can crowd out the provision of useful financial services (lending and banking services) or make them more expensive (Arping (2013)).

5.8.3. Impact on bank shareholders and unsecured creditors

Structural reform should create incentives for shareholders to be more mindful of excessive risk. In particular for the deposit taking entity, risk would be lowered and required returns should be lowered accordingly. So, whereas return on equity will be lowered, the risk-return trade-off does not need to be worse. There is a potential to rebuild a relationship of trust with bank stakeholders.

The financial crisis has underlined the importance of effective scrutiny and the exercise of discipline by creditors. Such discipline has been lacking, in large part as a result of the perceived taxpayer guarantee. The measures to reduce the implicit public subsidies would be the most effective way of correcting this, as bondholders would have a greater incentive to assess credit risk. Market discipline from creditors should encourage banks and their managements better to balance downside and upside risks.

Creditors of the trading entity would require and receive higher returns as a compensation for the higher risk exposure, given that they will no longer be shielded by an implicit subsidy, as before.

5.8.4. Impact on regulators and supervisors

Regulators and supervisors will benefit from increased transparency as separate entities will be subject to separate reporting and governance, and constrained in terms of activities they can conduct. The structural separation should allow simplifying bank regulation and tailoring it to the specific entities.

Philippon and Reshef (2012) argue that regulators do not have the human capital to keep up with the finance industry. Given the wage premium that they document for finance employees, it is impossible for regulators to attract and retain highly skilled financial workers because they cannot compete with private sector wages. They also report that the ratio of executive compensation in finance (the top regulated) to the highest salaries paid to regulators (the top regulators) grew from 10 in 1980 to over 60 in 2005 (or 40, excluding bonuses). This provides a potential explanation for regulatory failures and for circumvention of regulatory requirements.

Following structural reform, regulation and supervision could be better tailored to the nature of the banking activity in the deposit taking entity and trading entity, respectively. Arguably, banking groups engaged in a variety of activities also require much more complex regulation and supervision. More simplicity in terms of corporate structure of banks would normally allow simplifying regulation and supervision of banks, and potentially render supervision and regulation more effective. Likewise, the prudential regulation of banks is difficult for investors to understand. Accordingly, investors do not or are not able to fully exercise the “watch-dog” function under Basel's “pillar 3“ (market discipline). Unsecured bank creditors and investors perceive modern banks as opaque and as black boxes and it is possibly for this reason, inter alia, that they have started to call for structural separation. Institutional investors voiced their concern that banks are too opaque and complex to invest in.[85] If this claim were confirmed there is a prospect that certain forms of structural reform could, in fact, improve banks’ funding strains.

5.8.5. Impact on EU banking industry

Given the TBTF focus, structural reform would target a limited set of banking groups that are large, complex and engage in significant trading activity. Hence, structural reform would not affect the large majority of 8000+ banking groups in the EU, and would in particular exclude small cooperative and savings banks that focus on serving the financing needs of local communities and small businesses. It would therefore preserve the diversity of the EU banking sector, e.g. in terms of different business models. Given the reduced reliance on implicit public subsidies, competition on the merits would become more important, which should lead to inefficient competitors exiting the market and allowing more efficient competitors, including new entrants, to gain market share and revenues. Each part of the group would be subject to its own profitability and resource constraints. To the extent that structural reform facilitates and enhances the effectiveness of bank resolution, exit barriers are being removed, which gives more opportunities for banks that have a sound and prudent business model (European Commission (2011)).

The success and resilience of the universal banking model should not be taken for granted. The current EU financial system is dominated by relatively few large, interconnected and diversified universal banking groups. Whereas several of those large EU universal banking groups have weathered the crisis well, the EU financial system as a whole would have likely imploded due to a system-wide cascade of banking failures without the extraordinary and on-going taxpayer, government and central bank support (European Commission (2011, 2012)). The (contingent) taxpayer support to date amounts to 40% of EU GDP (€5.1 trillion parliamentary committed aid measures) and has undermined the solidity of several Member States' public finances. In the case of some Member States it has contributed to turn a banking crisis into a sovereign crisis (European Commission (2011, 2012)). This has had the effect of further increasing the fragility of the banking system since banks hold large volumes of sovereign bonds on their balance sheet - and hence confidence on these banks depends on the robustness of the public safety nets). Implicit subsidy estimates suggest that the EU universal bank profitability may be artificially high given the indirect and implicit sponsoring by their sovereigns: the largest EU banks are likely to benefit more from implicit subsidies (which represent a sizeable part of their profitability, see Annex A4). 

Bank balance sheets would no longer grow as aggressively as before. Large European banks have expanded and leveraged up rapidly in the run-up to the crisis. They have also expanded internationally, by relying on USD short term wholesale markets and investing in USD claims, effectively intermediating between US savers and US borrowers (Shin (2012)).

Note that the retained structural reform options are not calling for a break up of banking groups (with the exception of proprietary trading activities which constitute a small share of banks’ balance sheets), but simply wants to disentangle the activities that are considered long-term and relationship-oriented from those that are short-term, transaction-oriented, and the most prone to rapid change, while at the same time it wants to maintain banks’ ability to efficiently provide a comprehensive range of financial services to their customers. The proposal is simply to introduce more structure in EU banks, not to break them up. Structural reform through subsidiarisation merely intends to clarify the structure of universal banks. The nature of the activities drives the distinction in underlying cultures. Commercial banking typically involves long-term lending relationships, whereas trading typically involves a short-term perspective. The separation of the two distinct cultures would avoid that a short-term oriented, deal- and fee-based trading culture negatively influences the long-term relationship-based culture of the deposit and lending entity. There is significant public support for structural reform, as highlighted in section 1.1, and as voiced by top economists and financial experts.[86]

Amongst others, banks operate the payment system, make loans to households, businesses and governments, help households and businesses to manage their risks and accommodate their financial needs over time. The purpose of the financial sector and banks should be to serve the “real economy”. A safe and sound banking sector is a pre-condition to fulfil these essential functions, serve the real economy, and allow for sustainable growth. Sustainable economic growth is what counts, not temporarily boosted artificial growth that results in booms and subsequent busts. As such, there is no conflict between stability and growth. As shown in the on-going banking crisis, taxpayer bailouts often prevent the market exit of failing banks, rather than just ensuring the minimum possible (i.e. the continuation of critically important activities and services that cannot easily be provided through other players).

In addition, the increased funding costs are likely to reflect the underlying riskiness of the activities. Genuine economies of scale are found to be exhausted at relatively low levels of assets and banking groups that would be subject to structural reform operate at scales that typically exceed that level (see Annex A9).

Also it is not obvious that any benefits are passed on to consumers. Philippon (2013) find that the unit cost of intermediation has not decreased over the past 30 years, despite advances in information technology, changes in the organisation of the finance industry, and despite the growth of new markets, notably for financial derivatives.

Given the importance of implicit subsidies in terms of banks’ profits, bank profitability is likely to be reduced following the funding cost increase of the trading entity. However, risk-adjusted bank profitability will not necessarily be lower, as structurally reformed banks will be less risky.[87]

Philippon and Reshef (2012) find that much of the increase in financial activity has taken place in the more speculative fields, at the expense of traditional finance. From 1950 to 2006, credit intermediation (lending, including traditional banking) declined relative to “other finance” (including securities, commodities, venture capital, private equity, hedge funds, trusts, and other investment activities like investment banking). Wages in “other finance” sharply increased relative to those in credit intermediation. Bolton et al. (2012) argue that a significant amount of speculation and deal-making is pure rent-seeking. In other words, it is wasteful activity that achieves nothing more than enabling the collection of rents on items that might otherwise be free.

Cecchetti and Kharroubi (2012) empirically find that the enlargement of the financial system, beyond a certain the size, is associated with reductions in real productivity growth. This, in part, may be due to the financial sector competing with the rest of the economy for scarce resources. Arcand et al. (2012) also find that there can be “too much” finance. When private credit reaches 80% to 100% of GDP, which is largely exceeded for several crisis-affected EU Member States such as DK, NL, IE, CY, UK, ES, PT, further private credit is found to be negatively associated with GDP growth. The hypothesis is that excessively large financial systems may reduce economic growth because of the increased probability of a misallocation of resources, the increased probability of large economic crashes[88], or the endogenous feeding of speculative bubbles. Philippon (2008) observes that outstanding economic growth was achieved in the 1960s with a much smaller financial sector.

Furthermore, structural reform could have an important beneficial impact on the stability of the financial system. Structural reform could affect the probability of banking crises such as Lehman Brothers, Northern Rock or Cajas. Following structural reform, Lehman Brothers would have been less connected to deposit taking banks, and therefore the impact of its failure would have been less disruptive. Under lower levels of interconnectedness with deposit entities, resolvability of Lehman Brothers would pose fewer concerns and would thus be more likely. Anticipating this, the scope for aggressive pre-crisis growth and contagion upon failure would have been reduced through increased market discipline. Investment banks would not have been obliged to compete as aggressively as they did, faced with competitors that enjoyed artificial funding cost advantages thanks to their safety net coverage. The repeal of Glass-Steagall may have significantly increased the competitive pressure felt by pure investment banks like Lehman Brothers and Bear Stearns given that they faced commercial banks that were allowed to enter the investment banking area after 1999 and which could do so at artificially low funding costs.

In the case of Nothern Rock and Caja collapses, structural reform could contribute to avoiding the aggressive growth and lending practices of these institutions. Such practices were only made possible by relying on financial innovation, including wholesale funding, securitisation of mortgages, and derivatives to structure these products for distribution to investors, which in turn reflected the expansion of the largest European banks. The rationale of structural reform is to refocus the deposit entities of large banks towards a sustainable relationship-oriented model of banking and move away from a transaction-oriented fee-based and short term oriented business model. Although the rules and restrictions would not apply to Northern Rock and Cajas, they would not feel similar pressure and they would not dispose of similar possibilities to leverage up as quickly as they did. Moreover, the current real estate crises, unlike most in history, imperilled sovereigns because sovereigns were obliged to bail out not only deposit entities but also banking activities that they should not have had to bail out.

Some argue that reputational or informational contagion would still create spill-overs between the two entities in a crisis and that deposit taking entities would not be able to avoid confidence crises if problems arise in the trading entity of the group. However, authorities will have more options to act in dealing with a distressed banking group, if the basic parts are more distinct from each other. Better structured groups allow to isolate the problem better than when the group structure is opaque.[89] Finally, structural reform aims to make the safety net more limited in scope and hence more credible and effective in stopping a run when it does occur.  

6. Monitoring and Evaluation

Monitoring will take place during the phase-in period.  

Ex-post evaluation of all new legislative measures is a priority for the Commission. Evaluations are planned about 4 years after the implementation deadline. The forthcoming Regulation will also be subject to a complete evaluation in order to assess, among other things, how effective and efficient it has been in terms of achieving the objectives presented in this report and to decide whether new measures or amendments are needed.

In terms of sources of information that could be used during the evaluation, the data provided from the national central banks, the national regulators, European bodies such as the ECB, EBA and ESRB and from international organizations such as BIS, OECD, IMF and FSB. Relevant data could also be collected by relevant market participants or intermediaries.

Relevant indicators to evaluate the effectiveness of the reform proposal could include:

· Number and aggregate assets of banks subjected to structural separation requirements;

· Allocation of activities to deposit-taking or trading entity;

· Transaction volumes, spreads or liquidity in relevant markets;

· Trends in market shares of banks subject to structural separation;

· Market concentration in activities subject to structural separation;

· New entrants in activities subject to structural separation; and

· Trends in profitability of banks subject to structural separation, benchmarked against international peers and risk-adjusted.

· Measures of TBTF banks’ funding cost advantage;

· Measures of trading and loan activity by TBTF banks; and notably

· Measures of the size of implicit public subsidies.

References

Acharya, V., T. Cooley, M. Richardson, and I. Walter (2009), "Manufacturing tail risk: a perspective on the financial crisis of 2007-2009", Foundations and Trends in Finance, 4, 4, pp. 247-325.

Admati, A. and M. Hellwig (2013), The Bankers' New Clothes: What's Wrong with Banking and What to Do about It, Princeton University Press.

Adrian, T. and H. Shin (2010a), “Liquidity and leverage”, Journal of Financial Intermediation, 19, 3, pp. 418-437.

Adrian, T. and H. Shin (2010b), “The Changing Nature of Financial Intermediation and the Financial Crisis of 2007–2009”, Annual Review of Economics, 2, 603–18.

Alessandri, P. and A. Haldane (2009), "Banking on the State", paper underlying a presentation delivered at the FRB Chicago International Banking Conference on 25/09/2009.

Arcand, J-L., E. Berkes and U. Panizza (2012), “Too much finance?”, IMF Working Paper 12/161.

Arping, S. (2013), “Proprietary trading and the real economy”, Duisenberg School of Finance – Tinbergen Institute Discussion Paper, TI 13-032/IV/DSF 52.

BIS (2013), “Triennial Central Bank Survey of foreign exchange and derivatives market activity in 2013”, Bank for International Settlements, update 8 December.

Baumol, W. (1990), "Entrepreneurship: Productive, Unproductive, and Destructive", Journal of Political Economy, 98, 5, pp. 893-921.

BCBS (2010), “An assessment of the long-term economic impact of stronger capital and liquidity requirements”, Banking Committee on Banking Supervision, BIS.

BCBS (2012), “Report on intra-group support measures”, February, Banking Committee on Banking Supervision, the Joint Forum, BIS.

Benston, G. (1990), The separation of commercial and investment banking: the Glass-Steagall Act revisited and reconsidered, Oxford University Press.

Bessler, W. and M. Stanzel (2009), "Conflicts of Interest and Research Quality of Affiliated Analysts in the German Universal Banking System: Evidence from IPO Underwriting", European Financial Management, 15, 4, pp. 757-786.

BIS (2013), “The road to a more resilient banking sector”, 83rd BIS Annual report 2012/2013, Chapter V.

Blundell-Wignall, A., P. Atkinson, and C. Roulet (2012), “The business models of large interconnected banks and the lessons of the financial crisis”, National Institute Economic Review, No. 221.

Blundell-Wignall, A., P. Atkinson, and C. Roulet (2013), “Bank business models and the Basel system: complexity and interconnectedness”, Financial Market trends, Issue 2.

Blundell-Wignall, A., G. Wehinger and P. Slovik (2009), “The elephant in the room: The need to deal with what banks do”, Financial Market trends, Issue 2.

Bolton, P., T. Santos, and J. Scheinkman (2012), "Cream Skimming in Financial Markets", working paper, Princeton University and Columbia University.

Boot, A. and L. Ratnovski (2012), “Banking and trading”, IMF Working Paper 12/238.

Boyd, J. and A. Heitz (2012), “The social costs and benefits of too-big-to-fail banks: a bounding exercise”, working paper. Breedon, T. (2012), “Boosting finance options for finance“, report of industry-led working group on alternative debt markets.

Breedon, T. (2012), “Boosting finance options for business”, taskforce report of industry-led working group on alternative debt markets.

Buiter, W. (2008), “Lessons from the North Atlantic financial crisis”, Paper prepared for presentation at the conference “The Role of Money Markets" jointly organised by Columbia Business School and the Federal Reserve Bank of New York on May 29-30, 2008.

Buiter, W. and E. Rahbari (2012), “Debt of nations: Mr. Micawber’s vindication: Causes and consequences of excessive debt”, Citi GPS: Global Perspectives & Solutions, November.

Cecchetti, S. and E. Kharroubi (2012), "Reassessing the impact of finance on growth", BIS Working Paper 381.

Chow, J. and J. Surti (2011), “Making banks safer: Can Volcker and Vickers do it?”, IMF Working Paper 11/236.

Claessens, S. (2009), “Competition in the financial sector: Overview of competition policies”, IMF Working Paper, 09/45.

Claessens, S., R. Herring, and D. Schoenmaker (2011), “A safer world financial system: Improving the resolution of systemic institutions”, CEPR International Centre for Monetary and Banking Studies, Geneva Reports on the World Economy.

Dam, L. and M. Koetter (2012), “Bank bailouts and moral hazard: evidence from Germany”, Review of Financial Studies, 25, 8, pp. 2343-2380.

Davies, R. and B. Tracey (2012), “Too big to be efficient? The impact of too big to fail factors on scale economies for banks”, mimeo.

De Grauwe, P. and Y. Ji (2013), "Strong governments, weak banks", Policy Brief No. 305, Centre for European Policy Studies.

Demirgüc-Kunt, A., Karacaovali, B., and L. Laeven (2005), “Deposit insurance around the world: A comprehensive database”, Working Paper, World Bank.

Demirgüc-Kunt, A. and H. Huizinga (2013), “Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads”, Journal of Banking and Finance, 37, pp. 875-894.

Diamond (1984), “Financial intermediation and delegated monitoring”, Review of Economic Studies, 51, pp. 393-414.

Diamond, D. and P. Dybvig (1983), “Bank runs, Deposit Insurance, and Liquidity”, Journal of Political Economy, 91, pp. 401-419.

Draghi, M. (2012), speech at the London Global Investment Conference, 26 July.

Duffie, D. (2012), “Market Making Under the Proposed Volcker Rule", a report to the Securities Industry and Financial Markets Association and a submission to the Office of the Comptroller of the Currency, the Board of Governors of the Federal Reserve System, the Federal Deposit Insurance Corporation, and the Securities and Exchange Commission, Graduate School of Business, Stanford University.

European Central Bank (2012), "European Financial Integration", April.

European Commission (2011), “The effects of temporary State aid rules adopted in the context of the financial and economic crisis”, Commission Staff Working Paper, 5 October, SEC(2011) 1126 final.

European Commission (2012), “State aid scoreboard: Report on state aid granted by the EU Member States - Autumn Update”, 21 December, SEC(2012) 443 final.

European Commission (2013a), “Towards a stronger financial sector to support growth”, http://ec.europa.eu/internal_market/publications/docs/financial-reform-for-growth_en.pdf.

European Commission (2013b), “Structural reform in the EU banking sector: Motivation, Scope and Consequences”, Chapter 3 of the European Financial Stability and Integration Report 2012, 24 April, SWD(2013) 156 final.

Fecht, F., A. Hackethal, A. and Y. Karabulut (2010), “Is proprietary trading detrimental to retail investors?”, working paper.

Freixas, X., G. Loranth, A. Morrison (2007), ”Regulating financial conglomerates”, Journal of Financial Intermediation, 16, 4, pp. 479-514.

French, K. (2008), “Presidential Address: The Cost of Active Investing”, Journal of Finance, 63, 4, pp. 1537-1573.

FSB (2011), “Effective resolution of systemically important financial institutions”.

FSB (2013), “Progress and Next Steps Towards Ending “Too-Big-To-Fail” (TBTF)”, Progress report to the G20, September 2013.

G20 (2013), G20 Leaders Declaration, St Petersburg, September 2013.

Gambacorta, L. and A. van Rixtel (2013), “Structural bank regulation initiatives: approaches and implications”, BIS Working Paper No 412, April.

Glode, V., R. Green, and R. Lowery (2012), "Financial Expertise as an Arms’ Race”, Journal of Finance, 67, 5, pp. 1723-1759.

Haldane, A. (2010a), “The contribution of the financial sector – miracle or mirage?”, Bank of England.

Haldane, A. (2010b), “The $100bn question”, Bank of England.

Haldane, A. (2012), “On being the right size”, Bank of England.

Hasbrouck, J. (2009), “Trading Costs and Returns for U.S. Equities: Estimating Effective Costs from Daily Data,” Journal of Finance, 64, 3, pp. 1445–1477.

Hebb, G. and D. Fraser (2002), “Conflict of interest in commercial bank security underwritings: Canadian evidence”, Journal of Banking and Finance, 26, 10, pp. 1935-1949.

Hebb, G. and D. Fraser (2003), “Conflict of interest in commercial bank security underwritings: United Kingdom evidence”, Quarterly Journal of Business & Economics, 42, pp. 79-95.

Hermes (2012), “Epidemiology: Next steps in banking regulation”, thought piece annexed to the response to the consultation on the HLEG proposals, Hermes Equity Ownership Services.

Herring, R. and J. Carmassi (2010), “The corporate structure of international financial conglomerates: Complexity and its implications for safety and soundness”, chapter 8 of the Oxford Handbook of Banking, Oxford University Press.

HM Treasury (2013), “Banking reform: a new structure for stability and growth”, February.

Hodgkinson, L. (2001), “Analysts´ Forecasts and the Broker Relationship”, Journal of Business Finance & Accounting, 28, 7 and 8, pp. 943-961.

Hoenig, T. and C. Morris (2011), “Restructuring the banking system to improve safety and soundness”, Federal Reserve Bank of Kansas City, May.

Hu (2012), “Too complex to depict? Innovation, ‘pure information’, and the SEC disclosure paradigm”, Texas Law Review, 90, 7, pp. 1601-1715.

ICB (2011), “Final report: Recommendations”, Independent Commission on Banking.

IMF (2012), “Global Financial Stability Report”, October.

Kashyap, A., R. Rajan, and J. Stein (2002), ”Banks as liquidity providers: an explanation of lending and deposit-taking”, Journal of Finance, 57, 1, pp. 33-73.

Kroszner, R. (1998), “Rethinking banking regulation: a review of historical evidence”, Journal of Applied Corporate Finance, 11, 2.

Kroszner, R. and R. Rajan (1994), “Is the Glass-Steagall Act justified? A study of the U.S. experience with universal banking before 1933”, The American Economic Review, 84, 4, pp. 810-832.

Liikanen (2012), “Final Report of the high-level expert group on reforming the structure of the EU banking sector”, 2 October, http://ec.europa.eu/internal_market/bank/docs/high-level_expert_group/report_en.pdf.

Lumpkin, S. (2011), “Risks in financial group structures”, OECD Journal: Financial Market Trends, Issue 2.

Miller, M., L. Zhang, and H. Li (2013), “When bigger isn’t better: bailouts and bank reform”, Oxford Economic Papers, 65, pp. 7-41.

Moody’s (2013), “EU Bank Resolution: Draft Directive Offers Clarity On Future Support Framework, But Important Questions Remain Unanswered”, Moody’s Investor Services, September 30.

Murphy, K., A. Shleifer, and R. Vishny (1991), "The Allocation of Talent: Implications for Growth". Quarterly Journal of Economics, 106, 2, pp. 503-530.

National Bank of Belgium (2013), “Structural banking reforms in Belgium: Final Report”, July.

Noss, J. and R. Sowerbutts (2012), “The implicit subsidy of banks”, Financial Stability Paper No. 15, May, Bank of England.

Odean, T. (1999), "Do investors trade too much?", American Economic Review, 89, 5, pp. 1279-1298.

Omarova, S. (2011), “From Gramm-Leach-Bliley to Dodd-Frank: The unfulfilled promise of section 23A of the Federal Reserve Act”, North Carolina Law Review, 89, pp. 1683-1769.

Oxera (2011), “Assessing state support to the UK banking sector”.

Philippon, T. (2008), “The evolution of the US financial industry from 1860 to 2007: theory and evidence”, working paper, New York University.

Philippon, T. (2013), "Has the U.S. Finance Industry Become Less Efficient? On the Theory and Measurement of Financial Intermediation", working paper, NYU Stern School of Business.

Philippon, T. and A. Reshef (2008), “Wages and human capital in the US financial industry: 1909-2006”, working paper, University of New York and University of Virginia.

Popov, A. and F. Smets (2011), “Financial markets: Productivity, procyclicality, and policy”, European Central Bank, MIT Press, ISBN 9780262018296, p. 191-211.

Puri, M. (1994), “The long-term default performance of bank underwritten securities issues”, Journal of Banking and Finance, 18, pp. 397-418.

Ranganathan, V. (2013), “A status report on Dodd-Frank”, Central Banking Journal, August 12.

Ratnovski, L. (2013), “Competition policy for modern banks”, IMF Working Paper WP/13/126.

Richardson, M. (2013), "Why the Volcker Rule Is a Useful Tool for Managing Systemic Risk", working paper, New York Stern School of Business.

Schich, S. and S. Lindh (2012), “Implicit Guarantees for Bank Debt: Where Do We Stand?”, OECD Journal: Financial Market Trends.

Schich, S. and B-H Kim (2012), “Developments in the Value of Implicit Guarantees for Bank Debt: The Role of Resolution Regimes and Practices”, OECD Journal: Financial Market Trends.

Shin, H. (2012), "Global banking glut and loan risk premium", Mundell-Fleming Lecture, IMF Economic review, 60, 2, pp. 155-192.

Stiglitz, J. (2010), The financial crisis of 2007-2008 and its macroeconomic consequences, time for a visible hand: Lessons from the 2008 world financial crisis, Oxford University Press, pp. 19-49.

Thakor, A. (2012), “The economic consequences of the Volcker rule”, United States Chamber of Commerce, Center for Capital Markets Competitiveness.

Ueda, K. and B. Mauro (2012), “Quantifying structural subsidy values for systemically important financial institutions”, IMF Working Paper No. 128.

UK Parliament (2012), “First report of the Parliamentary Commission on Banking Standards”, 21 December.

US Government Accountability Office (2011), Proprietary Trading: Regulators Will Need More Comprehensive Information to Fully Monitor Compliance with New Restrictions When Implemented, GAO-11-529, July 2011.

[1]               For reform efforts to date and the complementarity of structural reform, see Section 2.3 and Annex A3.

[2]               Too-big-to-fail is meant to cover too-interconnected-to-fail (TITF), too-complex-to-fail (TCTF), and too-systemically-important-to-fail (TSITF). See also European Commission (2013b).

[3]               For a mandate and list of members, see http://ec.europa.eu/internal_market/bank/docs/high-level_expert_group/mandate_en.pdf

[4]               The other recommendations of HLEG included (2) that a separation of additional activities may be necessary conditional on the recovery and resolution plan; (3) the use of bail-in as a resolution tool; (4) a review of capital requirements on trading assets and real estate related loans; and (5) measures aimed at strengthening the governance and control of banks so as to strengthen bank scrutiny and market discipline. This Impact Assessment focuses on the mandatory separation recommendation.

[5]               See http://ec.europa.eu/transparency/regdoc/rep/10061/2013/EN/10061-2013-2037-EN-F-0.Pdf.

[6]               European Parliament (McCarthy 2013), Reforming the structure of the EU banking sector, 2013/2021 (INI)

[7]               Consultation by the High-level Expert Group on reforming the structure of the EU banking sector, May/June 2012. http://ec.europa.eu/internal_market/consultations/2012/banking_sector_en.htm

[8]               Consultation on the recommendations of the High-level Expert Group on Reforming the structure of the EU banking sector, http://ec.europa.eu/internal_market/consultations/2012/hleg-banking_en.htm

[9]               Consultation by the Commission on the Structural Reform of the Banking Sector, http://ec.europa.eu/internal_market/consultations/2013/banking-structural-reform/index_en.htm

[10]             Next to the higher importance of bank intermediation compared to market intermediation in the EU compared to the US, there are two other important factors that explain the discrepancy of total banking sector size between the EU and US. First, large EU banks apply IFRS to report their financial statements, whereas US banks apply US GAAP. The latter allows to net financial derivatives, implying that an identical bank can report a significantly smaller balance sheet size under US GAAP rules than under IFRS reporting rules. Second, whereas mortgages are recorded on balance for large European groups, a significant amount of US mortgages is recorded on the balance sheets of the US Government Sponsored Entities (GSEs) which are not included in Table 1.

[11]             Commercial banking activities include notably deposit taking and lending to individuals and businesses, traditional investment banking activities include securities underwriting and advisory services.

[12]             The main bank failures (and avoided failures) have been attributed to overreliance on short-term wholesale funding, excessive leverage, excessive trading/derivative/market activity, poor lending decisions due to aggressive credit growth, and weak corporate governance (see Liikanen (2012)). Size relative to GDP matters, but also the speed at which bank balance sheet growth outpaces GDP growth. If total assets outpace GDP growth, it implies that the banking sector increasingly relies on funding that is not coming from households and corporates, but rather from other financial intermediaries and capital market investors, which implies less funding stability.

[13]             In the case of some Member States it has contributed to turn a banking crisis into a sovereign crisis (European Commission (2011, 2012)). This has had the effect of further increasing the fragility of the banking system since banks hold large volumes of sovereign bonds on their balance sheet and since some of their funding sources are explicitly or implicitly insured by their sovereign.

[14]             The findings of the JRC research as well as an extensive review of the relevant literature are provided in Annex A4.1 and A4.2.

[15]             Other activities, notably insurance, are typically carried out in wholly owned but separately capitalised subsidiaries.

[16]             The most common regulatory and legal model for large banking groups in the EU is the “universal banking model”, whereas it is the “holding company model” in the USA. Financial holding company structures have a single holding company that typically holds all shares of the separate capitalised subsidiaries (amongst which may be a bank holding company). There is typically complete legal separation between the parent and the subsidiaries, and in case the holding company is non-operating, there is also operational independence and the latter acts solely as an investment company. Under a holding company structure, a group is headed by one entity which does not itself conduct any business but simply owns a series of other businesses and co-ordinates their strategies. Parent-subsidiary structures may consist of a parent bank that operates directly, with separately capitalised subsidiaries carrying out separate activities.

[17]             Deposit insurance refers to the explicit government guarantee related to certain categories of deposits to a certain extent (in the EU up to 100 000 EUR per retail depositor is 100% insured). Lender of last resort liquidity facilities are provided by central banks. Central banks should lend in principle at penalty rates and against adequate collateral to illiquid but solvent banks (Bagehot principles).

[18]             Freixas et al. (2007) show that financial conglomerates utilise excessive risk-taking due to their access to the public safety net, and that this effect wipes out any diversification benefits.

[19]             See Liikanen (2012) and Buiter and Rahbari (2012) for an overview of the changing business models in this respect.

[20]             At the end of 2011, each of the ten largest EU banking groups had total on-balance-sheet assets exceeding 1000 billion euro. Unlike their US peers, several large EU banking group balance sheets exceed domestic GDP. The geographic scope of large European banking groups’ is also significant in relative terms, as EU banking groups hold a far larger percentage of their assets abroad, compared to North American or Asian banking groups (65% compared to 32% and 26%, respectively, according to Claessens et al. (2011)).

[21]             Popov and Smets (2011) analyse the role of direct intermediation through financial markets with the indirect intermediation through levered banks. They argue that less deep financial markets in the EU relative to those of the US are, to a large extent, responsible for the smaller increase in productivity and slower pace of industrial innovation. They also compare the liquidity spirals, asset fire sales, and interbank market freezes of the recent financial crisis with the much more orderly burst of the dot-com bubble. They argue that the credit boom of the 2000s was driven by debt finance, while the dot-com bubble was mostly driven by an expansion in equity ownership, and equity is not held in levered portfolios.

[22]             Haldane (2010a) discusses the earnings of the financial sector in detail and concludes that “risk illusion, rather than a productivity miracle, appears to have driven high returns to finance”. Philippon and Reshef (2008) study wages earned in the financial sector and conclude that a large part of the observed wage differential between the financial sector and the rest of the economy cannot be explained by observables like skill differences. Philippon (2012) provides a quantitative interpretation of financial intermediation in the USA over the past 130 years and concludes that “…the unit cost of intermediation has increased since the mid-1970s and is now significantly higher than it was at the turn of the twentieth century. In other words, the finance industry that sustained the expansion of railroads, steel and chemical industries, and later the electricity and automobile revolutions seems to have been more efficient than the current finance industry. Surprisingly, the tremendous improvements in information technologies of the past 30 years have not led to a decrease in the average cost of intermediation. One possible explanation for this puzzle is that improvements in information technology have been cancelled out by zero-sum activities, perhaps related to the large increase in secondary market trading”.

[23]             Credit Rating Agencies often provide two different types of credit ratings in their assessment of the probability of default of a bank’s issued debt: a stand-alone rating and a higher support rating which not only takes the intrinsic strength of the bank into account, but in addition the agencies' estimate of the external support that the bank under consideration would receive from public authorities (and parent companies).

[24]             Investors that have replied to the Commission’s consultation on the HLEG report chaired by Liikanen stated that “All banks fail to provide sufficient transparency of their circumstances, meaning that investors tend to mistrust almost all of them with equal fervour” (Hermes 2012, page 5).

[25]             This is reflected in the price-to-book ratios of large and complex EU banking groups, which hover around 0.5 (i.e. the market values their assets at half the price at which they are accounted for), whereas they were as high as 2.0 in the run-up to the crisis. Part of that value destruction reflects the legacy of the past (and possibly on-going forbearance), and another part may reflect weak perceived profitability going forward. However, it is clear that a significant part may reflect the difficulty of valuing the individual components of large and complex banking groups with any degree of certainty.

[26]             See for example Hoenig and Morris (2011) and Haldane (2012).

[27]             Besides minimum capital requirements (pillar 1) and supervisory review process (pillar 2).

[28]             Several international regulators have been investigating suspected cartel arrangements between traders involving a number of international banks. The purpose is to find out if traders of interest-rate derivatives colluded to manipulate the interest rate benchmarks (such as Euribor and Libor) in order to obtain a benefit in their own trading positions. In December 2013, the Commission settled on fines of more than EUR 1.7bn with several international banks and brokers (Deutsche Bank, Société General, RBS, JPMorgan, Citigroup and RP Martin). See http://europa.eu/rapid/press-release_IP-13-1208_en.htm. Two international banks, UBS and Barclays benefited from immunity from the leniency program. Other international regulators have set fines that amount to USD 454m for Barclays, USD 87m for ICAB, USD 1.1bn for Rabobank, USD 1.5bn for UBS, and USD 612m for RBS for Libor rigging (by October 2013).

[29]             G20 Leaders, September 2013: “We recognize that structural banking reforms can facilitate resolvability and call on the FSB, in collaboration with the IMF and the OECD, to assess cross-border consistencies and global financial stability implications.”

G20 Ministers, October 2013: “We will pursue our work to build a safe and reliable financial system by implementing the financial reforms endorsed in our  Leaders’ Declaration, which are aimed at  building upon the significant progress already achieved, including in creating more resilient financial institutions, ending too‐big‐to‐fail, increasing transparency and market integrity, filling  regulatory gaps,  addressing the potential systemic risks from shadow banking and closing information gaps.”

[30]             “The ratio of risk-weighted assets to total assets differs significantly between banks. It is remarkable that the banks with the highest amount of trading assets, notional derivatives, etc. (i.e. banks that are least "traditional") tend to have the lowest ratio.” Report of the HLEG, p. 43.

[31]             “[…] for a sample of 16 large EU banks, the capital requirements for market risks vary between close to 0% to just over 2% of the total value of trading assets, the average being close to 1%.” Report of the HLEG, p. 48. This explains why some measures have been taken, e.g. the use of stressed VaR as part of Basel 2.5’s revisions to the market risk framework.

[32]             European Banking Authority (2013), “Interim results of the EBA review of the consistency of risk-weighted assets. Top-down assessment of the banking book”, February 2013; Basel Committee on Banking Supervision (2013), “Regulatory consistency assessment programme (RCAP) – Analysis of risk-weighted assets for market risk”, January 2013.

[33]             See e.g. response of Algebris Investment’s to the Commission’s 2012 consultation on the recommendations of the HLEG: http://ec.europa.eu/internal_market/consultations/2012/hleg-banking/other-organisations/algebris-investments_en.pdf

[34]             Banks cannot enter normal bankruptcy and suddenly stop performing their special and critical role in the payment system, nor can they freeze their deposits, because of the financial panic that would result and because their business model and raison d’être is to provide liquidity to its depositors.

[35]             See EBA (2012), “Opinion of the European Banking Authority on the recommendations of the High-level Expert Group on reforming the structure of the EU banking sector”, December 2012.

[36]             See e.g. Moody’s (2013) assessment of the BRRD: “Taken at face value, the draft is credit-negative for senior unsecured creditors of the roughly two-thirds of EU banks whose ratings incorporate some level of systemic support uplift. It is unlikely we would remove all systemic support from every EU bank’s rating in the foreseeable future, but a change to our assumptions would imply lower ratings for some or all banks. However, there are a number of important areas in which we need greater clarity before we can take a definitive view on the implications for EU bank ratings. For example, to be able to assess the Directive’s impact we would ideally want to understand […] the plans for broader structural changes in the EU banking industry”

[37]             For example, the forthcoming UK retail ring-fencing rules would apply to UK-incorporated subsidiaries of EU banks (e.g. Santander UK) but not to branches (e.g. Deutsche Bank’s UK branch).

[38]             The former risk has been acknowledged in some of the national reform debates. For example, the UK ICB acknowledged the risk but dismissed it due to disproportional net cost for banks as well as reputational concerns. The National Bank of Belgium (2012, 2013) highlighted the difficulty for a small country like Belgium to impose unilateral structural reforms due to banks’ ability to switch into branches and the unlevel playing field this would give rise to.

[39]             Of the banks that would exceed the thresholds considered in Chapter 5 on the basis of historical data, a majority would be subject to distinct national structural reforms.

[40]             In order to better understand and make markets in commodity derivatives, some banks have in recent years engaged increasingly in the trading of physical commodities. The involvement of banks has given rise to a number of concerns, ranging from possible market manipulation to potential risks to the solvency of banks due to their exposure to volatile commodities prices. Both the French law on banking structural reform and draft secondary legislation published by the UK government restrict trading in physical commodities such as metals, oil or agricultural commodities (only the latter in the case of the French law). In the US, the Federal Reserve currently interprets US legislation such as to allow certain Bank Holding Companies (BHCs) to engage in trading of physical commodities, subject to specific conditions and restrictions. However, on 19 July 2013 the Federal Reserve announced that it was reviewing its 2003 decision that first allowed banks to engage in trading of physical commodities.

[41]             Securities underwriting is a typical investment banking activity in which banks raise investment capital from investors on behalf of corporations and governments that are issuing securities (both equity and debt securities) in return for a fee. It is a way of selling newly issued securities, such as stocks or bonds, to investors.

[42]             Securitisation involves the creation and issuance of tradable securities, such as bonds, that are backed by the income generated by e.g. an asset, a loan or another revenue source.

[43]             The Act originally contained four short sections of the Banking Act of 1933: Section 16 prohibiting banks from underwriting or dealing in securities; Section 21 prohibiting securities firms from taking deposits; and two remaining sections (20, 32) prohibiting banks from being affiliated with firms that are principally or primarily engaged in underwriting or dealing in securities (see also Box 2).

[44]             Global Shadow Banking Monitoring Report 2013, 14 November 2013, FSB

[45]             In order to support the qualitative assessment and comparison of reform options, the Commission services have on a best-effort basis attempted to quantify to the extent possible some of the costs and benefits referred to here. See section 5.7.1 and Annexes A10, A11 and A12 for further detail.

[46]             Note that the direct GDP impact of the funding cost increase will initially be zero, as the increased funding cost implies an offsetting benefit for bank creditors (note that risk is not captured in GDP metrics). However, despite this initial offsetting GDP benefit originating in the higher proceeds for bank creditors, a negative impact on GDP on balance will still materialise over time when the private funding cost increase is passed on to other bank stakeholders. When it is passed on to bank customers through higher borrower rates for households and higher cost of capital for SMEs and firms, household consumption and business investment and hence GDP will be negatively affected (the consumption or investment of creditors will not increase correspondingly, as the higher remuneration reflects higher risk exposure). Alternatively, banks may pass on the cost to their shareholders (lower RoE) or their employees (lower wages), which would again lead to reduced consumption by shareholders and employees and a loss of GDP. However, the evidence so far suggests that this GDP cost may be limited even for a broad trading entity and when introducing conservative estimates about the pass-through rates. In the UK, the long-term GDP level is expected to be reduced with 7.5 basis points when opting for a relatively narrow deposit entity and a relatively strong separation (Impact Assessment of HM Treasury (2013)). This GDP reduction social cost estimate is shown to be almost an order of magnitude smaller than the social benefit of the proposed structural reform, following the estimated reduction in the probability or impact of financial crises (and its impact on economic growth). See also Annex A12 for Commission Services estimates of the impact of funding cost advantages on economic growth.

[47]             Several of the benefits of structural reform listed in Section 3.2 can be interpreted as eliminating potential diseconomies of scope in large and complex banking groups (such as excessive risk taking, lack of resolvability, and conflicts of interest). Whereas the benefits stress the beneficial impact of eliminating distortions on incentives and behaviour, we (should) quantitatively correct for the distortions when assessing the economies of scale and scope. Annex A9 provides a review of the relevant literature.

[48]             See e.g. Davies and Tracey (2012) that found no economies of scale in a sample of large international banks with assets above USD100bn. See Annex A9 for a review of the relevant literature.

[49]             Annex A7 sets out the alternative options and policy levers to design the stronger legal, economic and operational separation.

[50]             For the purposes of this Impact Assessment, “complex” (and “simple“) securitisation will be defined by reference to the ECB ABS eligibility requirements for repo purposes and as currently elaborated and refined by European supervisory authorities (for example EIOPA in its non-public “technical report on standard formula design and calibration for certain long-term investments”). 

[51]             In the context of this Impact Assessment, complex derivative transactions” refer to the structuring, arranging or execution of derivative transactions other than standardised interest rate and foreign exchange derivatives transactions for the prudent management of liquidity, funding and overall balance sheet risk (i.e. to perform its asset and liability management). In principle, the compensation package of the dedicated staff that is to perform the prudent management should reflect the hedging objective and cannot consist of bonuses that per definition are linked to the profitability of the unit or overall banking group.

[52]             In the context of this Impact Assessment, “underwriting” is interpreted in the broad sense, as also incorporating the market making activity that underwriters typically perform in the days following primary market transactions. See Annex A6.

[53]        Many of the leading UK banks have told the UK Parliamentary Commission on Banking Standards that they do not engage in proprietary trading at all. The same message was given by Dutch banks to the Members of the Commission on the structure of Dutch banks. A non-public Febelfin survey provides evidence that proprietary trading amounts to 2% of trading revenues for Belgian banks in the first semester of 2012, down from 13%, 11% and 8% in 2009, 2010, and 2011 respectively. In turn, trading revenues are estimated to amount to 9% of overall bank revenues in the first semester of 2012. The French and the German structural reform proposals propose to subsidiarise proprietary trading. Their cost-benefit analysis findings have not been made public, but BNP Paribas corporate banking and investment banking revenues are estimated by the banks to be impacted by the government plans by less than 2%. Annex 2 of National Bank of Belgium (2013) documents how trading activity more generally has evolved between 2008:Q1 and end 2012 for the four largest Belgian banks.    

[54]             See US GAO (2011) and PCBS (2013).

[55]             Market makers often have signed on to a voluntary code of conduct, which already considers manipulative practices by banks with each other or with customers to be “unacceptable trading behaviour”. However, the multiple financial scandals in the years since signing these voluntary agreements cast doubt on their effectiveness.

[56]             Kroszner and Rajan (1994), Kroszner (1998), Hebb and Fraser (2003) and Stiglitz (2010).

[57]             See Hodgkinson (2001) and Bessler and Stanzel (2009).

[58]             Kroszner and Rajan (1994), Puri (1994), Benston (1990), Hebb and Fraser (2002), Hebb and Fraser (2003).

[59]             Underwriting is a case in point. Underwriting and advisory services require relationship-building with clients. These traditional (investment) banking activities are closely connected to corporate banking. From the corporate client’s perspective, issuing a bond is an alternative way of financing to taking a bank loan. Given that underwriting is not as easily scalable as pure market making, the scope for moral hazard reduction is significant, but smaller than for market making. Underwriting as such does not give rise to significant interconnectedness across financial institutions. The evidence does not suggest that conflicts of interests are obvious between underwriting and loan making.  In fact, it suggests that bonds underwritten by commercial banks default less often than bonds underwritten by investment banks. There may be economies of scope to be enjoyed from spreading fixed costs of acquiring information over multiple outputs; more specifically, concurrent lending and underwriting could be beneficial.

[60]             See footnote 72. 

[61]             See US GAO (2011) and PCBS (2013).

[62]             It might, however, need to be enforced through line-of-business restraints.

[63]             Underwriting is a case in point. Underwriting and advisory services require relationship-building with clients. These traditional (investment) banking activities are closely connected to corporate banking. From the corporate client’s perspective, issuing a bond is an alternative way of financing to taking a bank loan. Given that underwriting is not as easily scalable as pure market making, the scope for moral hazard reduction is significant, but smaller than for market making. Underwriting as such does not give rise to significant interconnectedness across financial institutions. The evidence does not suggest that conflicts of interests are obvious between underwriting and loan making.  In fact, it suggests that bonds underwritten by commercial banks default less often than bonds underwritten by investment banks. There may be economies of scope to be enjoyed from spreading fixed costs of acquiring information over multiple outputs; more specifically, concurrent lending and underwriting could be beneficial.

[64]             Duffie (2012) argues that "market making is inherently a form of proprietary trading". A market maker acquires a position from a client at one price and then lays off the position over time at an uncertain average price. The goal is to "buy low, sell high."  In order to accomplish this goal on average over many trades, with an acceptable level of risk for the expected profit, a market maker relies on its expectation of the investors’ needs and the future path of market price and therefore necessarily this activity involves holding an inventory.

[65]             A supervisor can only do so either through prescriptive rule-making or through a purpose based restriction. In either case it would involve supervisory capital and enforcement can only be done ex post and would be costly and possibly ineffective.

[66]          The treasury function of a bank needs to engage in trades to manage excess liquidity or hedge the risk from for example selling fixed-rate mortgages while being funded with floating rate borrowing. Over time the treasury functions in some banks have become more aggressive traders with strategies that could be seen as resembling proprietary trading. In some cases, Treasury operations no longer merely manage the natural dynamics of the balance sheet arising from customer activity, but increasingly perform a set of trading activities in themselves and become pure profit centres.

[67]             For example, an attempt to identify particular hedges for particular transactions and prohibit all others would be counterproductive because it would encroach on a trader's ability to be creative and innovative as products and product lines become more intricate and new sources of risk emerge (Chatterjee, 2012).

[68]             A type I error is when an activity is identified as proprietary trading when in fact it is not, and type II errors mean that an activity is not identified as proprietary trading when it in fact should have been.

[69]             US experience with the intra-group firewalls provided by Section 23A and B of the Federal Reserve Act are instructive in this regard. According to those rules, the Federal Reserve Board have authority to exempt individual transactions from the requirements of section 23A. During the financial crisis, which at its height posed a threat to systemic financial stability, the FRB exercised this authority on numerous occasions, thereby allowing depository institutions to provide financing to their affiliated securities firms, derivatives dealers and money market funds in order to prevent their failure and the effects this might have had on the financial system and the broader economy. “Crisis containment and systemic risk considerations consistently prevailed over the statutory purpose of preventing the leakage of the federal subsidy outside the depository system” (Omarova (2011)). The UK’s implementation of the ICB recommendation has tried to avoid such concerns by clearly framing exemptions in primary law (e.g. related to simple derivatives).

[70]             As stated in chapter 4.1, there are certain retail activities that under any scenario could not be separated, i.e. deposit-taking and retail payment services.

[71]             Of the banks exceeding the threshold, roughly two thirds are located in a Banking Union participating Member State.

[72]       Although the present proposal is also aimed at facilitating resolution, this conservative assumption is used to avoid double counting some of the benefits calculated in the context of the Impact Assessment of the BRRD.

[73]     The point estimates below are based on JRC analysis and are conditional on the methodology employed and underlying assumptions. Confidence intervals around the reported point estimates could not be calculated.

[74]     For more information on the QUEST model, see http://ec.europa.eu/economy_finance/research/macroeconomic_models_en.htm. Simulations using several hypothetical scenarios of funding cost increases are shown in Annex A12.

[75]     See SWD(2012)166/3, table 23, p. 158 in Annex 13, explained in more detail in appendix 5 thereof.

[76]             US regulators are also considering imposing leverage limits on the largest US banks that would be stricter than the ones foreseen by Basel III.

[77]             For example, while the international standards for global systemically important banks contain numerical thresholds, the provisions in the CRR incorporating those standards do not, but leave it to the EBA to set such thresholds on the basis of principles and guidance set in the CRR.

[78]             Supervisors would always have the possibility to subject additional banks to structural separation.

[79]             This section outlines the likely and stylised impact of the reforms retained above on different stakeholders. The main baseline is the mid-range scenario of the three retained options, i.e. reform option E-. Material differences of options C+ and option C/E- are highlighted where appropriate.

[80]             Zingales (2012): “The third reason why I came to support Glass-Steagall was because I realised it was not simply a coincidence that we witnessed a prospering of securities markets and the blossoming of new ones (options and futures markets) while Glass-Steagall was in place, but since its repeal have seen a demise of public equity markets and an explosion of opaque over-the-counter ones.

[81]             For example, benefits of market liquidity should become smaller with the degree of market liquidity. The additional benefits of the extra liquidity derived from high-frequency trading must be of negligible (or negative) value compared to the benefits from having a market which is reasonably liquid on a day-by-day basis. Moreover, ever greater market liquidity may give rise to destabilising momentum effects, such as cycles of undervaluation and overvaluation.

[82]             Note that this concern needs to be put into perspective. Financial industry employment in London has decreased with 30% between 2007 and 2012, as a result of the financial crisis.

[83]             Philippon (2013) documents that the income share of the finance industry in the US economy starts just below 2% of GDP in 1880. It reaches a first peak of almost 6% of GDP in 1932. Note that this peak occurs during the Great Depression, not in 1929. Between 1929 and 1932 nominal GDP shrinks, but the need to deal with rising default rates and to restructure corporate and household balance sheets keeps financiers busy. Similarly, the post-war peak occurs not in 2007 but in 2010, just below 9% of GDP.

[84]             In a survey of elite US universities, Rampell found that in 2006, just before the financial crisis, 25% of graduating seniors at Harvard University, 24% at Yale, and 46% at Princeton were starting their careers in financial services. Those percentages have fallen somewhat since, but this might be only a temporary effect of the crisis.

[85]             Investors that have replied to the Commission’s consultation on the HLEG report chaired by Liikanen stated that “All banks fail to provide sufficient transparency of their circumstances, meaning that investors tend to mistrust almost all of them with equal fervour” (Hermes 2012, page 5).

[86]             Paul Krugman, Alan Greenspan, Paul Volcker, Thomas Hoenig, Sheila Bair, Andrew Haldane, Simon Johnson, Mervyn King, John Kay, Luigi Zingales, Willem Buiter (Citigroup), David Komansky (Merrill Lynch),, Phil Purcell (Morgan Stanley), etc.

[87]             Alessandri and Haldane (2009) document that large UK banks recorded an average annual return on equity of 7% from 1920 to 1970 (with a standard deviation of 2%), whereas the average annual return on equity amounted to 20% on average in the period 1970 to 2007 (with a standard deviation of 7%).

[88]             Popov and Smets (2011) analyse the role of direct intermediation through financial markets with the indirect intermediation through levered banks. They argue that less deep financial markets in the EU relative to those of the US are, to a large extent, responsible for the smaller increase in productivity and slower pace of industrial innovation. They also compare the liquidity spirals, asset fire sales, and interbank market freezes of the recent financial crisis with the much more orderly burst of the dot-com bubble. They argue that the credit boom of the 2000s was driven by debt finance, while the dot-com bubble was mostly driven by an expansion in equity ownership, and equity is not held in levered portfolios.

[89]             In that sense, the EU state aid control policy and corresponding rescue and restructuring plans would be made easier and more uniform.

LIST OF ANNEXES

ANNEX A1. Overview of structural reforms and reform proposals - PAGE 2

ANNEX A2. Summary of replies to the stakeholder consulta­tion - PAGE 27

ANNEX A3. Assessing the complementarity of structural separation with the current reform agenda - PAGE 43

ANNEX A4.1. Implicit subsidies: Drivers, Distortions, and Empirical Evidence - PAGE 55

ANNEX A4.2. Size and determinants of implicit state guarantees to EU banks - PAGE 79

Annex A1: Overview of existing structural reform proposals

The financial crisis has led to an overhaul of the regulatory and supervisory structures governing the financial system in Europe and beyond. These reforms have been coordinated at international level via the G20. However, reforming the structure of banks – i.e. regulatory interventions to directly amend the organisational structure or specific business models of banks – have not been part of this internationally agreed set of reforms. Instead, bank structural reform was initially pursued by those two countries at the epicentre of the early stages of the financial crisis, i.e. the United States and the United Kingdom. Since then, some other countries have followed suit and have adopted or considered to adopt bank structural reform measures (France, Germany, Belgium, The Netherlands, Switzerland). Furthermore, a High-level Expert Group appointed by the European Commission and chaired by Erkki Liikanen has recommended specific reforms of the structure of European banks.

This chapter assesses and compares some of the most mature reform proposals along three dimensions: (i) the activities that are to be separated, (ii) the type of separation and the impact on economic linkages within the banking group, and (iii) the scope of banks it is likely to apply to. Before assessing on-going reforms, an introductory section places these reforms in a historical context by briefly outlining how the different national banking systems have evolved over time.

1. A long history of evolving bank structures

The debate about reforming the structure of the banks is influenced by the different shapes of national banking systems. That structure largely reflects economic history and the different roles banks have played in financing economic development.

Broadly speaking, banking systems have developed differently in Anglo-Saxon countries compared to continental Europe. In the UK and US the larger financing needs related to industrialisation was predominantly served by capital markets with access to those markets provided by specialised intermediaries (clearing banks in the UK, broker/dealers in the US). Banks were typically small and focused on retail clients. In continental Europe, industrialisation occurred later and substantial amounts of capital were needed to catch up with the industrial forerunners and given that capital markets were less developed, banks became the main source of financing to the large corporates who required a universal set of services. The universal banking model therefore has long historical roots in some countries, whereas in other countries banks were more specialised.

In terms of regulation, approaches have also evolved. In several countries with universal banks, concerns with close links between banks and commerce coupled with the Great Depression saw the introduction of activity restrictions in several countries (e.g. US, IT, BE). However, against the backdrop of the general financial liberalisation and deregulation that occurred from the 1970s onwards, restrictions were relaxed over time due to competitive pressures, perceptions of economies of scope resulting from combining a universal set of activities in one roof and increased faith being put on the ability to manage risks that might arise as a result of these business combinations. Prior to the start of the financial crisis, the universal bank model had accordingly gained prominence internationally.

Box 1: Historical development of national banking structures[1]

United States: in the US, prior to industrialisation, banks tended to be small and fragile. As a result, they avoided concentrated lending and preferred arm’s lengths relations. The large scale financing needs related to the growing economy was primarily done via capital markets, with specialised non-bank institutions acting as gatekeepers. This also reflected long-standing restrictions on banks engaging in securities business. At the turn of the century, banks started to consolidate and a national banking system emerged capable of better serving the growing economy. During that process, some banks gained national prominence and became the main providers of funding to the large US corporate trusts. The increasing power of the largest banks and their close integration with the largest corporates provoked an increasing backlash, with e.g. the establishment of the Federal Reserve system in 1913 ending the ‘central bank’ role provided by the JP Morgan and other large New York banks, and the Sherman Antitrust Act breaking up the trusts and limiting the influence of banks. These changes culminated with the 1933 Glass-Steagall Act, which fully prohibited banks from engaging in securities business. That segmented banking system came under attack in the 1960s, as US commercial banks felt disadvantaged vis-à-vis its competitors both nationally and internationally due to restrictions on their ability to expand their activities. This prompted US regulators to gradually become more permissive as to the extent to which US banks could engage in prohibited activities. In 1999, the Gramm-Leach-Bliley Act finally eliminated the sections of the GSA that prohibited banks from entering into non-banking activities. As a result, US banking groups can organise themselves as financial holding companies, and are then able to provide a universal set of services in different functional subsidiaries.

United Kingdom: the UK until the mid-1980s by and large had a two-tier banking system, with a limited number of nation-wide clearing banks focusing on commercial banking and a number of specialised merchant banks focusing on providing access to securities markets (there were also a number of specialised institutions, such as building societies). The origins of that division of labour can be traced back to the 17th century, when the Bank of England had a monopoly on joint stock banking, small country banks provided local financial services, and London-based merchant banks provided trade finance and placement of government bonds. With the advent of limited liability banking, the small country banks eventually consolidated into nation-wide clearing banks that focused on commercial banking. This focus was not legally imposed but was rather explained by merchant banks occupying the securities market space and commercial banking producing sufficient profit opportunities given the increase in wealth resulting from industrialisation. This system remained stable for most of the 20th century. Following deregulation in the 1960s and 1970s clearing banks began to provide a wider set of services, but their access to investment banking remain hampered by rules of the London Stock Exchange, which required members to specialise either as brokers or market-makers and prevented outsiders from owning a significant financial interest in member firms. Those rules were eliminated in the context of the big bang of October 1986 and as a result, the clearing banks finally also went into investment banking. Since then, the UK banks have provided a universal set of services.

Germany: The German banking system consists of three pillars (private banks, savings banks and cooperative banks) all of which offer a universal set of services. As regards the private banks, they played a central role in the industrialisation process during industrialisation. Given the limited capital markets, banks were the only ones able to pool the amounts of capital necessary and direct it towards the growing sectors of the economy. Commercial banking was eventually dominated by a few large Berlin-based banks that provided a universal set of services to a set of large corporates to which they were closely associated (by banks taking on shareholdings and directorships in the companies they served). Meanwhile, the retail segment was served by local savings and cooperative banks. However, due to their unstable funding structure (e.g. lacking stable retail deposits), the commercial banks suffered heavily during the Great Depression and had to be nationalised (e.g. state acquired 90% of Dresdner Bank, 70% of Commerzbank, and 35% of Deutsche Bank). To address the perception that excessive competition had undermined the solidity of banks, they faced more intrusive regulation (e.g. price caps and branching limits) and essentially became closely associated with the stage in the form of a government controlled cartel. Banks were reprivatized in 1936, but continued to be closely linked to the state. After the war, the commercial banks were broken up along geographic lines (one bank per Land) but no activity restrictions were as such imposed. Following the start of the Cold War, the 1950s saw a gradual reconsolidation of the commercial banks, who by the end of that decade were again allowed to reconstitute themselves as national universal banks. In the 1970s, these universal banks were again questioned due to the financial power and lack of competition resulting from these banks' close links with large corporates. However, policy did not change as a result.[2] However, link between banks and companies have evolved for a variety of reasons (e.g. tax reform making the sale of company stakes more attractive).

France: from the end of World War II until the 1980s, the French banking sector was compartmentalised, with deposit banks on the one hand and banques d’affaires et banques de credit on the other hand. This reflected legal activity restrictions on deposit banks. Large parts of the banking sector were also under public ownership, which was further expanded in the early 1980s. However, the 1984 Banking Act created a single legal and supervisory framework and eliminated activity restrictions, thus paving the way for universal banks. Following this, the French banking sector consolidated rapidly, also exploiting the opportunities for cross-border expansion offered by the European single market. The French banking sector is today dominated by a limited number of large universal banks, some of which have their origins in the savings bank and cooperative banking sectors (BPCE, Crédit Agricole and Crédit Mutuel).

Italy: Italy’s financial system was largely bank-based, given that capital markets were thin and illiquid.  Similar to Germany, Italian banks developed close ties with large corporates, whom they funded through lending and, increasingly, capital ownership as they developed. The banking system went through a number of crises during the late 19th and early 20th century. In the late 1920s and early 30s most of Italy’s universal banks had to be rescued and during those rescue operations, they were forced to divest their industrial stakes and revert to exclusive short-term commercial banking only. These changes were translated into law in 1936, which accordingly effectively separated commercial and investment banking. Similar to Germany, the new law also limited competition by means of e.g. branching limits. The 1936 law contributed to nearly half a century of financial stability and remained in place until the early 1990s. However, this stability came with a price, due to the inefficiencies originated by the limited competition in the Italian banking market. This was compounded by the relevant dimension and market share of publicly-owned banks. Regulatory changes enacted in the 1990s (1990 "Amato-Carli" law, 1993 Banking Law, 1998 "Ciampi" law) privatised and liberalised the banking sector and completed the removal of all the limitations introduced by the 1936 Banking Law, allowing the emergence of universal banks. These changes unleashed a consolidation wave, predominantly domestic in nature. 

Belgium: similar to the United States, Germany and Italy, Belgium also saw a policy debate in the wake of the Great Depression regarding the close links between banking and commerce.[3] Banks having large holdings of shares in industrial and commercial companies exposed them to concentration risk, which could weaken their retail bank activities. A 1934 decree accordingly required mixed banks to separate their deposit taking activities from their investment banking activities by incorporating the latter into a holding company. However, the retail banks also became subsidiaries of the holding. To further guarantee the independence of deposit taking banks vis-à-vis the holding, a 1935 decree notably prohibited banks from holding shares of industrial and commercial companies with the aim of preventing holding companies from using their bank as a vehicle to indirectly finance their investments. These restrictions were relaxed in the 1960s and 1970s and were eventually abolished in 1993 in the context of the single market paving the way for universal banks.

2. United States

Whereas nearly all attention as regards structural reform has focused on the Volcker rule, that rule builds on a long US history of bank structural regulation. This section accordingly places a discussion of the Volcker rule in that broader context.

2.1. Existing structural regulation

US structural regulation is built on three pillars that first, separate banking from commerce, second, restricts and isolate the core banking activities within broader financial groups, and third, limits the institutional concentration of the banking and financial sector.

2.1.1. Separation of banking and commerce

In the US a company that controls a bank holding company (BHC) cannot engage in activities of a commercial nature, be it directly or indirectly through a subsidiary. This reflects long-standing policy concerns that such combinations could give rise to conflicts of interest, a misuse of banking resources, or the creation of firms that are difficult to supervise. There are a few limited exceptions to this rule, including a de minimis exemption (less than 5% investments in commercial firms) and exemptions made for some grand-fathered firms.

2.1.2. Restriction and isolation of core banking activities

During most of the 20th century, the 1933 Glass-Steagall Act 1) separated commercial banking from investment banking, by prohibiting any affiliation between BHCs and firms principally engaged in securities underwriting and dealing. The Act also 2) limited BHCs to a set of core commercial banking activities and other closely related activities (e.g. investment advisory, securities brokerage and leasing). It finally 3) prohibited BHCs from engaging in insurance underwriting and agency activities. These restrictions were substantially liberalised by the 1999 Gramm-Leach-Bliley Act. Under that act, BHCs that meet certain regulatory capital and management criteria may choose to become Financial Holding Companies (FHCs). FHCs are allowed to engage in a broad set of financial activities, including securities underwriting and dealing, insurance underwriting and agency, and merchant banking. However, they are still not allowed to engage in commercial activities.

Nevertheless, structural remnants have remained that restrict and isolate core banking activities. While the banking group (BHC and FHC) may engage in a wide range of financial activities, the insured depositary institutions still face activity restrictions that limit their focus to core banking activities (e.g. taking deposits, lending) and other incidental activities (e.g. custody and asset management).

The US also has rules governing transfers between the different parts of a banking group, which are aimed at isolating the insured depository institution from excessive risks arising from the larger financial firm of which it is part (BHC, FHC) and to prevent the transfer of the subsidy arising from federal assistance to non-depository financial institutions:

· Section 23A of the Federal Reserve Act (i) imposes quantitative limitations on certain extensions of credit and other transactions between a bank and its affiliates that expose a bank to an affiliate’s credit or investment risk, (ii) prohibits banks from purchasing low-quality assets from their nonbank affiliates, and (iii) imposes strict collateral requirements with respect to extensions of credit to affiliates; Section 23A has a number of exemptions, the most important one being for transactions that are between banks where 80% of each bank’s shares are owned by the same company.

· Section 23B of the same act furthermore stipulates that these and other transactions[4] between a bank and its affiliates should take place on market terms, i.e. the same terms and conditions as those for non-affiliated companies;[5] Section 23B excludes banks from the term “affiliate” however. Banks that are part of a chain banking organization are thus exempt from section 23B.

· Regulation W of the Federal Reserve explains and simplifies the interpretation and application of these two acts. The Regulation notably limits a bank’s ’covered transactions’ with any single affiliate to no more than 10% of the bank’s capital and surplus and stipulates that overall covered transactions with all affiliates cannot exceed 20% of capital and surplus. The regulation also states that loans to affiliates should be fully or over-collateralised.[6]

These requirements have been subject to significant debate. In that context, two issues have been identified:

· Not all transactions were subject to the rule. Section 23A has not applied to derivative transactions. Instead, the Federal Reserve Board required banks to maintain internal policies and procedures for managing derivatives exposures to affiliates. Derivatives were nevertheless included in Section 23B.

· The provisions were not consistently implemented and enforced. The Act grants the regulator (Federal Reserve Board) considerable leeway in interpreting the meaning and scope of application. The Board also has the power to exempt transactions and relationships from the requirements of these provisions, if such exemptions are in the public interest. It has been demonstrated that during the financial crisis, the Board granted numerous financial institutions exemptions from these quantitative and qualitative requirements in order to enable banks to use their deposit-taking subsidiaries as a source of emergency financing for other entities within their corporate structure, thereby preventing the failure of their nonbank businesses and to avert broader market dislocations.[7]

2.1.3. Limits on market concentration

US market concentration has historically been limited by e.g. geographic restrictions that limited the ability of banks to establish or buy branches in other states. The 1994 Riegle-Neal Act liberalised these interstate branching limits. To reduce the risk of excessive concentration it imposed a national concentration limit for the deposit market (national deposit cap), prohibiting interstate expansions that would result in one BHC controlling more than 10% of the total deposits of all US insured depository institutions. The Dodd-Frank Act introduced a financial sector concentration limit, which prohibits any firm affiliated with an insured depository institution from expanding if the total consolidated liabilities would exceed 10% of the aggregate consolidated liabilities of all such financial firms.

2.2. Volcker Rule

Building on the existing structural regulation, Section 619 of the Dodd Frank Act and its final implementing rule, known as the Volcker Rule, as voted the relevant US agencies on 10 December 2013, further restricts financial groups that contain such institutions from engaging in certain types of market oriented activity (proprietary trading) and also puts significant limits on their investments in hedge funds and private equity funds.

The rule first prohibits any banking entity (i.e. an insured depository institution, any company that controls such an institution, a bank holding company…) from engaging in proprietary trading. Proprietary trading activity includes any purchase or sale as principal of any security, derivative, commodity future, or option on any such instrument for the purpose of benefitting from short-term price movements or realizing short-term profits. This activity is judged as as incompatible with the appropriate risk profile and customer-driven mission of banking entities.

The prohibition on proprietary trading is subject to exceptions for certain ‘permitted activities’ including market making, underwriting, hedging, organizing and offering a hedge fund or private equity fund and trading in certain government obligations, in particular US government, agency, State and municipal debt obligations and foreign debt instruments under certain conditions. Under these exceptions, banks will be able to engage in market-making activities as long as they prove these activities are aimed at meeting the “reasonably expected near term demands of clients, customers or counterparties”. The final rules exempt, provided certain requirements are met, trading on behalf of a customer in a fiduciary capacity or in riskless principal trades and activities of an insurance company for its general or separate account. Provided certain requirements are met activities are other activities are not considered proprietary trading, including trading solely as an agent, broker, or custodian; through a deferred compensation or similar plan; to satisfy a debt previously contracted; under certain repurchase and securities lending agreements; for liquidity management in accordance with a documented liquidity plan; in connection with certain clearing activities; or to satisfy certain existing legal obligations.

The rule limits these exemptions if they involve a material conflict of interest; a material exposure to high-risk assets or trading strategies; or a threat to the safety and soundness of the banking entity or to U.S. financial stability.

Permitted proprietary trading activities may not result in a material conflict of interest between the bank and its customers, clients, or counterparties; lead to material exposure to high-risk assets or high-risk trading strategies; pose a threat to the safety of the bank; or pose a threat to the financial stability of the US.

Second, the Volcker Rule also provides strict restrictions on banks' investments in hedge funds and private equity funds. It prohibits any banking entity from:

· Owning more than 3% of any individual hedge fund or private equity fund;

· Investing more than 3% of its tier 1 capital in hedge funds and private equity funds in the aggregate;

· Engaging in any asset sale or lending transaction with a hedge fund or private equity fund it sponsors; or

· Bailing out a sponsored hedge fund or private equity in any way.

Similar to the proprietary trading prohibition no permitted investments may involve high-risk trading strategies or assets or a material conflict of interest with customers, clients, or counterparties.

The Volcker rule entails full ownership separation, thus the cease and divestment of the prohibited activities. Compared to existing US regulation, the Volcker Rule applies broadly at the consolidated level and restricts very specific types of activities that, although clearly financial in nature and in some cases very difficult to differentiate from permitted activities, were deemed by the US Congress to be incompatible as a policy matter with the appropriate risk profile and customer-driven mission of banking entities. It is therefore different from the regime in place under the Glass-Steagall Act between 1933 and 1999, which more bluntly separated commercial banking from investment banking and did not attempt to make fine distinctions between “high-risk” and “low-risk” activities, or “proprietary” and “customer-driven” activities, in regulating bank structure.

Concerning corporate governance, to hold bank chief executives accountable, they must attest in writing that their banks are setting up processes to maintain, enforce and review compliance programmes. The rule does not impose on bank executives however that they certify that the bank is not engaged in proprietary trading. The rules also grant a broader exemption for banks’ market-making desks, on the condition that traders aren’t paid in a way that rewards proprietary trading. Trading-desk compensation must not reward either proprietary trading or “excessive or imprudent risk-trading”.

These rules are likely to affect a relatively limited part of US banks’ balance sheet. An in-depth examination by the US Government Accountability Office (GAO) of the proprietary trading activities of six largest US banks that carry out the bulk of trading overall highlighted that proprietary trading revenues were generally small compared to overall trading revenues and overall activities of these bank holding companies.[8] The combined revenues between 2006 and 2010 from stand-alone proprietary trading[9] represented between 1.4% - 12.4 of combined quarterly revenues for all trading, and between 0.2 – 3.1% of combined quarterly revenues for all activities at the bank holding companies. As regards revenues from hedge fund and private equity fund investments these were also small compared to overall revenues of the concerned bank holding companies, representing between 0.08% - 3.5% of the bank holding companies’ combined revenues between 2006 and 2010.[10]

Given the attempt to make these distinctions but the difficulty of providing clear ex-ante definitions by means of legislation, the Volcker Rule relies on a reporting and compliance regime that will collect new data to be monitored over time and to a certain extent on supervisory judgment. The compliance requirements under the final rules vary based on the size of the banking entity and the scope of activities conducted. Banking entities with significant trading operations will be required to establish a detailed compliance program. Small banks, particularly US community banks, will have no compliance obligations under the final rule if they do not engage in any covered activities other than trading in certain government, agency, State or municipal obligations.

The proposed rulemaking was published for consultation in November 2011. The consultation closed in February 2012 and generated a substantial number of responses.[11] Key concerns raised by commenters during the consultation process included: (1) the appropriate scope of the exemption for market making-related activities, including concerns about (i) the impact on market liquidity if such exemption is not construed broadly and (ii) the potential for arbitrage and evasion if the exemption is not construed narrowly; (2) the scope and potential burden of compliance program and data reporting requirements included in the proposed rule; (3) the over breadth of the statutory definition of “hedge fund and private equity fund”; and (4) the extraterritorial implication of the Volcker Rule’s statutory application to (i) the non-U.S. activities of foreign banks and (ii) proprietary trading in non-U.S. sovereign debt by both U.S. and foreign banks.

US relevant regulatory agencies (Fed, FDIC, OCC, CFTC and SEC) have voted and adopted the final rules on 10 December 2013. The rules go into effect on April 1 2014, although the compliance date will be delayed for a year to July 21 2015. Starting in June 2014, large banks are required to begin reporting certain information.

In sum, in terms of the three dimensions highlighted at the outset, the Volcker rule’s provisions amounts to full ownership separation of proprietary trading and investments in certain funds and applies to all US financial groups containing insured depository institutions as well as to any foreign bank with a US branch or agency subject to certain exemptions.

2.3. Swaps Push-out

Section 716 of the Dodd-Frank Act, known as the Swaps Push-out provision, states that if banks want to continue to benefit from federal assistance (deposit insurance and Federal Reserve discount window) they either have to stop engaging in certain swaps (certain credit derivatives, all equity and most commodity derivatives) or do such swaps in a separate legal entity, registered as a swap dealer and subject to capital requirements and margin requirements under the derivatives sections of the Act. The provision would not affect those derivatives judged to be important for banks’ management of risks (i.e. interest rate, foreign exchange, gold/silver, credit derivatives where underlying is an investment-grade security), which could thus continue to be provided within the bank. The concerned derivatives are likely to be a fraction of the overall derivatives business of major US banks.[12]

The large US banks that dominate US derivatives activity already have separately capitalised affiliates that could house the pushed-out derivatives. Moreover, the additional capital required is likely to be incremental. The main cost of the provision appear to be for clients that could lose the benefits related to cross-derivative product netting and who would have to deal with one more counterparty (the separately capitalised affiliate).

This provision is scheduled to enter into effect in July 2013. However, insured depository institutions can apply for a two-year extension, with a further one-year extension available afterwards. Furthermore, the US Congress is debating potential changes to Section 716, which would significantly broaden the range of permissible derivatives.[13]

2.4. On-going discussions on further structural measures

US policymakers continue to discuss potential additional structural measures to address banks that are too-big-to-fail. These discussions are primarily driven by some members of Congress and some supervisors. While these bills are at an early stage of debate and do not necessarily muster sufficient political support, they are indicative of an ongoing US debate on structural measures as regards TBTF banks. Different alternatives arise in these discussions. One draft legislative bill has proposed to significantly increase capital standards of the largest banks at both consolidated level and the level of individual subsidiaries; introduce stricter firewalls between these subsidiaries; and, would prohibit non-depository institutions having access to the safety net.[14] Another draft has instead proposed to introduce absolute size limits[15], whereas a third draft bill[16] would effectively reinstate the prohibition on ownership affiliations between commercial banks and securities broker/dealers of the Glass-Steagall Act.

3. United Kingdom

In June 2010, the UK government set up and Independent Commission on Banking (ICB) with the task of assessing bank structural reform. In September 2011, the ICB recommended that large UK banks should ring-fence their retail bank operations into separate legal subsidiaries subject to their own prudential safeguards. It also recommended that such groups should increase their loss absorbing capacity both at group level and at the level of the ring-fenced retail entity. The UK government has put forward draft primary legislation aimed at implementing the ICB’s recommendations, aimed at enabling pre-legislative scrutiny by a Parliamentary Commission on Banking Standards (PCBS), set up in October 2012 as part of a broader exercise aimed at assessing the UK banking sector.[17] The PCBS made a number of suggestions for amending the legislative proposal, some of which were taken up by the UK Government in the bill finally submitted to the Parliament in February 2013.[18] Secondary legislation is forthcoming and the government has indicated that all legislation will be in place by May 2015. The framework is scheduled to enter into force by the start of 2019.

3.1. Activities to be separated

The ICB divided activities into ‘mandated’, ‘prohibited’ and ‘ancillary’ services. The UK draft bill distinguishes between core activities (and related services) and excluded activities. More specifically:

· Mandated services / core activities: mandated services are services where even temporary interruption has significant economic costs and where customers are not well equipped to plan for such interruption. The ICB defined such services as taking of deposits from and providing overdrafts to individuals and SMEs. These services would have to be separated and could only be provided by the ring-fenced retail bank. The bill takes this on board, specifying accepting deposits as a ‘core activity’ and outlining three ‘core services’ that only ring-fenced banks can provide (including withdrawal/payments and overdraft). The bill also grants the Treasury the ability to include additional activities as well as exempt certain activities and services subject to certain conditions. The government has indicated that it will use the exemption possibility in secondary legislation to exempt high-net worth individuals and large corporate entities from the ring-fence obligation.

· Prohibited services / excluded activities: some services cannot be provided by the ring-fenced entity. The ICB stated that those are services that (i) make it harder/more costly to resolve ring-fenced bank, (ii) increase exposure to global financial markets, (iii) involve risk taking and are not integral to provision of payments services to customers or direct intermediation of funds between savers and borrowers within non-financial sector, and (iv) services that threaten the objectives of the ring-fence. The ICB did not exhaustively define those services, but highlighted that it would include (i) services provided to customers outside EEA, (ii) services that result in exposure to non-ring-fenced bank or non-bank financial organisation (except payment services), (iii) services that result in trading book asset, (iv) services that result in having to hold regulatory capital against market risk, (v) purchase/origination of derivatives /other contracts resulting in having to hold regulatory capital against counterparty credit risk, and (vi) secondary markets activity including purchase of loans/securities. The government bill states that dealing in investments as principal is an excluded activity. Similar to core activities, it grants Treasury the power to rule other activities as excluded, as well as a power to provide for exceptions to the ban on dealing in investments as principal.

This has an impact on derivatives. The ICB recommended that ring-fenced banks would only be able to act as agents for derivative products sold by others, not as principal. In its response to the ICB, the government argued that ring-fenced banks should be able to sell simple derivatives subject to certain conditions (e.g. only certain types, to certain customers, centrally managed risks, and subject to certain limits). The government has indicated that those conditions will be set in secondary legislation. Meanwhile, the PCBS has concurred with the government in that that there might be a case for permitting ring-fenced banks to sell simple derivatives but has proposed additional conditions: (i) safeguards to prevent mis-selling, (ii) a limited and durable definition in legislation of ‘simple derivatives’, and (iii) that there would be an additional cap on the gross volume of derivative sales for ring-fenced banks, and on the total value of derivatives used for hedging. It also called on the regulator to report annually on ring-fenced banks’ sale of derivatives. The government has committed to incorporate the PCBS’ suggestions into secondary legislation.

· Ancillary services: activities that are necessary for the efficient provision of mandated services may be provided by the ring-fenced bank. This concept has not been reflected in legislation so far.

Possible conditions governing ring-fenced banks’ dealing in derivatives as principal

In allowing ring-fenced banks to sell derivatives, a number of safeguards have been considered to ensure that this activity does not expose the ring-fenced banks to additional risks and/or make it more difficult to resolve. The safeguards are aimed at restricting the sale of derivatives to more simple ones and to limit the economic importance of this potential business.

The government[19] has previously outlined that simple derivatives would only mean certain types of derivative contracts (standardised interest rate and foreign exchange derivatives) and could only be sold to certain customers (only to holders of mandated deposits, i.e. households and SMEs, and non-financial institutions). It has also considered conditioning the sale of derivatives on their risk management characteristics. Along that line, the ring-fenced bank should show that it can manage counterparty and market risk and that the portfolio can be unwound or transferred if the bank fails. Accordingly, ring-fenced banks could be able to hedge its net market risk if it used standardised derivatives on arm's length/third party basis. Hedging derivatives could also be permitted if they were centrally cleared on a CCP or bilaterally cleared with daily margin calls and exchange of collateral. However, residual market exposures should be capped at a low percentage of the ring-fenced bank's Tier 1 capital, and netting arrangements should respect ring-fenced banks’ independence. 

The PCBS[20] has proposed three additional safeguards: (i) prevent mis-selling, (ii) a limited and durable definition in legislation of ‘simple derivatives’, and (iii) an additional cap on the gross volume of derivative sales for ring-fenced banks, and on the total value of derivatives used for hedging.

According to UK government and ICB estimates, this separation would lead to a relatively narrow deposit bank and a broad trading entity. It would retain 18%-36% of the UK total banking assets in the deposit-taking ring-fenced retail bank and would shift the remaining 64%-82% to the other legal entity that does trading and is not allowed to take insured deposits.[21]

3.2. Strength of separation

As regards the strength of the separation, the ICB put forward recommendations related to (i) the legal and operational links, and (ii) economic links of ring-fenced banks that are part of a wider group.

As regards legal and operational links, the ICB recommended that ring-fenced part should be possible to isolate from group in some days and able to continue provide services without solvency support. For that to be the case, the following conditions should be met.  First, the ring-fenced bank should be a separate legal entity. Second, if the ring-fenced bank owns other organisations, these should only do activities permitted within retail ring-fence. Third, the wider corporate group should ensure that ring-fenced bank has access to all services it needs to continue its operations irrespective of the state of wider group. Finally, the ring-fenced bank should be direct member of payment systems, or use another ring-fenced bank as agent.

As regards economic links, the ICB stated that ring-fenced bank's relations with other parts of group should take place on a third party basis. Ring-fenced bank should not be dependent on group's continued financial health for its solvency or liquidity. Accordingly, relationships with other group entities should for regulatory purposes be treated similar to relations with third parties. Second, transactions with other group entities should be conducted on commercial and arm's length basis. Third, assets sold to other group entities should be at market value. Fourth, ring-fenced bank should meet regulatory requirements, including for capital, large exposures, liquidity and funding on a solo basis. Fifth, ring-fenced bank should only distribute dividends/transfer capital if its board believes it has resources to do so. If this leads it to breach capital requirements, they would need explicit regulatory approval. Sixth, the board of ring-fenced bank should be independent. Furthermore, ring-fenced bank make regulatory disclosures on a solo basis and, finally, the board of ring-fenced bank and board of parent company have duty to maintain integrity of ring-fence.

The government has accepted these recommendations. The proposed primary legislation has structured the rules that ring-fenced banks have to respect to ensure their legal, operational and economic independence vis-à-vis the wider group along five dimensions highlighted during PCBS hearings by Andy Haldane of the Bank of England and Sir John Vickers:

· Separate governance;

· Separate risk management;

· Separate balance sheet (treasury) management;

· Separate remuneration structure and human resourcing; and

· Independence of capital and liquidity.

In addition, debate continues on additional measures to strengthen the separation. First, the PCBS debate has highlighted a concern with the durability of the ring-fence in light of regulatory arbitrage eroding it over time. To counter that risk, the PCBS has proposed granting additional reserve powers to ‘electrify’ the ring-fence if banks do not comply, i.e. to force full ownership separation. This, it is argued, would discourage banks from testing the limits of the ring-fence. A first reserve power would empower regulators to, subject to conditions force a specific banking group to divest itself fully of either its ring-fenced bank or its non-ring-fenced bank. In addition, the PCBS argues that the ring-fence framework should be subject to regular independent review, which should assess whether ring-fencing is achieving its objectives and, as a second reserve power, assess whether there is a case for a move to full ownership separation across the banking sector as a whole.[22] Whereas the government has agreed to amend the bill to reflect the first reserve power it has not accepted the independent review or the second reserve power. The PCBS has accordingly suggested amendments to the bill.[23]

Second, as regards group corporate structures, apart from stating that ring-fenced banks could not own entities that provide prohibited services, the ICB was not prescriptive about groups with ring-fenced banks should be structured. It notably did not suggest prohibiting non-ring-fenced banks from owning ring-fenced banks. The case has since been made that the bill should include a power to prohibit such ‘parent-child’ ownership structures on the grounds that it would undermine the strength of the ring-fence and notably the provision that relations should be on an arm’s length basis. A parent-subsidiary structure based on control would contradict this principle. A ‘sibling’ ownership structure, where the ring-fenced and non-ring-fenced banks are subsidiaries of a holding company, would facilitate the objectives of the ring-fence. The PCBS have accordingly called for the bill to be amended so as to require concerned banking groups to organise themselves as sibling structures with holding companies on top.[24] The government has rejected this call, arguing that the first reserve power combined with additional powers being considered as part of the EU Bank Resolution and Recovery Directive would grant sufficient powers to regulators to intervene in group structures to this end. The PCBS has accordingly put forward an amendment requiring that shares in a ring-fenced body are held only by another member of the group that is not carrying on an excluded activity.[25]

3.3. Institutions to be covered by separation requirement

The ICB recommended that the ring-fence obligations would apply to all UK domiciled banking groups. The bill excludes building societies from ring-fencing.[26] The bill also grants the Treasury power to exclude other institutions from the ring-fencing obligation. The government has indicated that it would use this power to exempt smaller institutions from the ring-fencing requirement by means of a de minimis exemption whereby which banks with less than £25bn in mandated deposits would not have to implement the ring-fence.

3.4. Loss absorbency

As regards loss absorbency, the Government endorses the ICB's recommendations on loss absorbency. Ring-fenced banks should accordingly hold a minimum level of 17% of primary loss absorbing capacity (PLAC) consisting of an equity buffer of 3% and a choice for banks whether to hold the remaining 7 percentage points in either equity or highest quality loss absorbing debt. For non-ring-fenced banks, the Government proposes that the most systemic UK G-SIBs (those subject to 2.5% G-SIB surcharge) should hold at least 17% PLAC against all domestic and non-exempted overseas operations.[27] The ICB proposed to scale these requirements in line with an institution's size; the Government proposes to await conclusion of international negotiations on D-SIBs before making a decision on how to scale.

The government furthermore takes up the ICB’s suggestion to change the creditor hierarchy so that insured deposits (those covered by the UK deposit guarantee system – FSCS) are preferred in insolvency, i.e. depositors rank ahead of other creditors (depositor preference). The aim is to sharpen the incentives for other senior unsecured creditors to exert discipline on banks' behaviour.

Contrary to the ICB, the government does not propose to require larger ring-fenced banks to respect stricter leverage limits than those considered internationally. Accordingly, it endorses the 3% limit currently contained in Basel III. This is being contested by the PCBS that instead endorses stricter and more immediately applicable leverage limits.[28]

The government supports bail-in and expects to implement it when transposing the European Commission's proposal for a Recovery and Resolution Directive (RRD) that contains bail-in powers, and not earlier as requested by the PCBS. PLAC should accordingly consist of capital (equity, Additional Tier 1 and Tier 2) and long-term unsecured debt that is subject to the bail-in power.

4. France

The French government in December 2012 put forward a proposal for a law separating and regulating banking activities.[29] The proposal has since been approved with modifications by the Assemblée nationale and the Sénat.[30] Both assemblies will now be working towards a common version. The law would amend the legislative section of the Monetary and Financial Code. Subsequent changes to the regulatory section are likely to follow to further spell out the exact provisions.

4.1. Activities to be separated

The French draft reform proposes that French banking groups with trading activities above a certain limit (to be defined by administrative provisions) would only be allowed to engage in (i) certain own-account dealings – hereinafter 'proprietary trading' –d and (ii) in certain dealings with leveraged funds by means of ring-fenced subsidiaries that are not funded by insured deposits.

As regards the provision separating proprietary trading, it is stated that French banking groups would still be allowed to engage in certain own-account activities that are deemed useful for the real economy, subject to certain conditions:

· Provision of investment services (notably risk hedging) to clients. In order to determine that these activities are client oriented, the services in question should be remunerated by the client and the risks should be prudently managed;

· Risk hedging for the credit institution or group, i.e. with the aim of reducing risk exposures related to credit and market risks, where the hedges must present an economic link with the identified risk (to be defined by administrative provisions);

· The clearing of financial instruments;

· Market-making, i.e. either the simultaneous publication of firm and competitive two-way quotes for comparable volumes with the result of providing liquidity to the market, or involving as part of its usual activity the execution of buy and sell orders on behalf of clients. The draft law proposes a set of indicators that would need to be monitored so as to determine that the market-making activity is not concealed proprietary trading. The draft law would also grant the power to the Minister of the Economy to set a threshold above which banks would no longer benefit from the exemption[31];

· Prudent group treasury management, subject to forthcoming conditions (the forthcoming conditions related to prudent management have been added to avoid JP Morgan CIO style events); and

· Group investment operations, i.e. long-term investments.

In addition, any ownership interests or unsecured exposures towards certain leveraged funds (e.g. hedge funds) would also have to be transferred to the investment entity.

The French government impact analysis did not publish any figures for the proportion of assets that would be affected by the separation requirement.[32] This was due to the limited number of banks[33] concerned and the associated confidentiality problems. The analysis did neither disclose an average number, as the extent of proprietary trading vary substantially between the banks in question. However, external analyses suggest that proprietary trading constitute a very limited proportion of bank assets.[34] This has prompted analysts to conclude that this reform would have a moderate impact on the French universal banking model.[35]

4.2. Strength of separation

The segregated entities would be authorised as investment firms or, by derogation, as credit institutions. These would have to respect the following conditions:

· They would be barred from taking guaranteed deposits and from providing payment services to clients with guaranteed deposits;

· They would have to respect prudential requirements on an individual or sub-consolidated basis. The banking group controlling these entities would furthermore have to deduct the exposures to these entities when calculating their own fund requirements;

· As regards large exposures, the segregated entity is regarded as being part of the group; and

· The segregated entity should have a separate commercial identity (e.g. name) so as to avoid confusion;

· The segregated entity should have separate corporate governance, in the sense that the persons that determine its business focus cannot be the same ones as the wider group;

· Finally, the segregated entities would be prohibited from carrying out (a) at high frequency trading and (b) trading in agricultural commodity derivatives.

The law also foresees a strengthening of the supervision of market activity, as well as granting the supervisor (ACPR) extended powers to ban banks from engaging in specific activities or providing certain products.

4.3. Institutions to be covered by separation requirement

The scope of the proposed reform is limited to banking groups (credit institutions, financial holding companies, and mixed financial holding companies) with significant trading activities, measured as a share of their total assets. The threshold for such a de minimis exemption will be defined by decree.[36]

5. Germany

In February 2013, the German government put forward a draft bank-separation law and new criminal-law provisions for the financial sector.[37] The draft law was modified in the parliamentary procedure.[38] The law, which was promulgated in the Federal Law Gazette (Bundesgesetzblatt) on 12 August 2013[39], amends in particular the German Banking Act. The provisions dealing with separation of activities (Article 2 of the law) are similar to the French approach, with some differences as regards the activities to be separated and strength of separation. The law also contains a fleshed-out de minimis exemption. Article 2 of the law will enter into force on 31 January 2014.

5.1. Activities to be separated

In terms of activities, it prohibits banks and enterprises belonging to the same group as a bank from engaging in (i) proprietary business, (ii) lending and guarantee business with hedge funds and European or foreign alternative investment funds (AIFs), and (iii) proprietary trading with the exception of market-making unless that activity is conducted in a separate financial trading institution (FTI). An FTI would not be allowed to provide payment services or e-money services.

However, business that is (i) aimed at hedging transactions with clients (except from AIFs) or that serve the institution's or its alliance's[40] interest rate, currency, liquidity or credit risk management, (ii) that involves the purchase or sale of long-term holdings or business which does not serve the purpose of drawing profit from the short-term use of existing or expected differences between the purchase and sale prices or volatilities of prices or interest rates, would not be prohibited.

Irrespective of the de minimis exemption, the supervisor BaFin will be able to prohibit a bank or an enterprise belonging to the same group as a bank to carry on certain activities (market-making activities, activities that when exceeding the thresholds are prohibited by law or activities that are comparable in terms of risk to the these activities) and ask it either to cease these activities or to transfer them to an FTI if there is a concern that this business threatens to jeopardise the solvency of the bank or the enterprise belonging to the same group as a bank due to the volume, profit or risk structure of the activity.

The explanatory notes accompanying the German proposal did not publish any figures for the proportion of assets that would be affected by the separation requirement. This may be due to similar confidentiality problems. Similar to French banks, proprietary trading is likely to constitute a very limited proportion of German banks’ assets.

5.2. Strength of separation

The business to be separated would be operated at a commercially, organisationally and legally separate enterprise (i.e. FTI).

The FTI would have to secure its own refinancing independently. Dealings between the banking group and its other entities with the financial trading institution should be dealt on a third party basis.

In terms of governance, BaFin will be able to issue orders to ensure appropriate business organisation in terms of separation.

Specific reporting obligations for the bank and the superordinated enterprise of a group encompassing a bank towards the BaFin may be imposed by secondary legislation. The supervisory boards of the FTI, the bank and the superordinated enterprise are obliged to inform themselves regularly and as required on the activities of the FTI, the risks involved and compliance with requirements.

5.3. Institutions to be covered by separation requirement

The law sets a de minimis exemption regarding the statutory prohibition of activities for banks and groups comprising a bank which either have (i) assets held for trading and available for sale under €100bn, or (ii) have total assets of less than €90bn, subject to the above-mentioned trading assets not constituting more than 20% of total assets. This exemption does, however, not apply for the power of the BaFin to prohibit activities and ask for their cessation or transfer.

The statutory prohibition of activities will apply as of 1 July 2015. However, the relevant business is only prohibited and has to be ceased or transferred to an FTI twelve months after exceeding one of the thresholds. The BaFin's power to prohibit activities and ask for their cessation or transfer will apply as of 1 July 2016.

6. High-level Expert Group on reforming the structure of the EU banking sector (Liikanen)

The High-level Expert Group on reforming the structure of the EU banking sector, chaired by Erkki Liikanen, delivered its report in October 2012. It notably recommended the mandatory separation of certain high-risk trading activities into a separate legal entity.

6.1. Activities to be separated

The Group concluded that it is necessary to require the legal separation of certain particularly risky financial activities from deposit-taking banks within the banking group. The Group therefore recommended that the following activities would need to be assigned to a separate legal entity (“trading entity”):

· Proprietary trading;

· All assets or derivative positions incurred in the process of market-making, other than the activities exempted below; and

· Any loans, loan commitments or unsecured credit exposures to hedge funds (including prime brokerage for hedge funds), SIVs and other such entities of comparable nature, as well as private equity investments.

The group argued that these activities naturally belong to each other and accordingly should be conducted within the same entity.

The following activities would be exempted from the separation requirement:

· Hedging services to non-banking clients (e.g. using forex and interest rate options and swaps) which fall within narrow position risk limits in relation to own funds, to be defined in regulation;

· Securities underwriting;

· Use of derivatives for own asset and liability management purposes; and

· Sales and purchases of assets to manage the assets in the liquidity portfolio.

Only the deposit bank would be allowed to supply retail payment services. All other banking business would be permitted to remain in the entity which uses insured deposits as a source of funding (“deposit bank”), unless firm-specific recovery and resolution plans require otherwise.[41]

The Group furthermore recommended that authorities should be able to require a bank to separate a wider set of trading activities, if the recovery and resolution planning highlighted that this activity involved (i) particularly complex trading instruments, (ii) particularly complex governance and legal structures, and (iii) led to risk positions that were large in relation to overall market size for the particular instrument.

6.2. Strength of separation

In terms of strength of separation, the Group put forward the following recommendations:

· Separate capitalisation: Both the deposit bank and the trading entity would each individually be subject to all the regulatory requirements (e.g. CRR/CRDIV and consolidated supervision) which pertain to EU financial institutions. They would accordingly need to be separately capitalized according to the respective capital adequacy rules, including the maintenance of the required capital buffers and possible additional Pillar 2 capital requirements. In addition, the trading entity and deposit bank only pay dividends if respect capital requirements;

· Limits on intra-group transfers: intra-group transfers would need to be on market-based terms and should be restricted according to normal large exposure rules. Direct or indirect transfers would furthermore not be allowed to the extent that capital adequacy, including additional capital buffer requirements on top of the minimum capital requirements, would be endangered; and

· Corporate structure: The requirements would apply on the consolidated level and the level of subsidiaries. The Group furthermore stated that the legally-separate deposit bank and trading entity could operate within a bank holding company structure.  At any rate, the deposit bank would need to be sufficiently insulated from the risks of the trading entity.

6.3. Institutions to be covered by separation requirement

Structural separation would only be mandatory if the activities to be separated amount to a significant share of a bank’s business, or if the volume of these activities can be considered significant from the viewpoint of financial stability. Accordingly, the smallest banks would be considered to be fully excluded from the separation requirement. The Group suggested that the decision to require mandatory separation should proceed in two stages:

· Examination thresholds: In the first stage, if a bank’s assets held for trading and available for sale exceed (1) a relative examination threshold of 15-25% of the bank’s total assets or (2) an absolute examination threshold of EUR100bn, the banks would advance to the second stage examination.

· Separation threshold: In the second stage, supervisors would determine the need for separation based on the share of assets to which the separation requirement would apply (i.e. proprietary trading, market-making, and loans and unsecured exposures to certain funds). This threshold, as share of a bank’s total assets, would need to be calibrated by the Commission. Mandatory separation would apply to all banks for which the activities to be separated are significant, as compared to the total balance sheet.

Once a bank exceeds the final threshold, all the activity concerned should be transferred to the legally-separate trading entity.

6.4. Other

The Group furthermore made a number of recommendations in other areas.

It recommended possible amendments to the use of bail-in instruments as a resolution tool, notably that the bail-in requirement would be used explicitly only to a certain category of clearly defined debt instruments and that the bail-in instruments would not be held within the banking sector.

It called for a review of capital requirements on trading assets and real estate related loans, once on-going international reviews finalised, and at a general level called for strong and coordinated actions to improve the consistency of internal models across banks. Furthermore, to strengthen the macro-prudential framework, it called on Member States to include caps on loan-to-income and loan-to-deposits in their macro-prudential toolboxes; and

It recommended measures aimed at strengthening the governance and control of banks, including e.g. fit and proper tests to ensure management ability to run large and complex banks; measures related to remuneration, and more detailed risk disclosure at legal entity level so as to strengthen scrutiny and market discipline.

7. Conclusion

From the above, whereas the different proposals share the ambition of better protecting retail deposits, they differ as regards (i) the activities they formally suggest separating, (ii) the strength of separation considered, and (iii) the scope of institutions they are intended to apply to. The proposals are summarised in the table below.

Table 1 – Overview of structural reform measures in place or being proposed

|| Measure || Type of separation || Activities || Strength || Institutions || Other

US || Bank Holding Company Act (BHC) || Functional separation || · BHC cannot engage in activities of commercial nature; · Insured depository institution can only engage in core banking activities (e.g. taking deposits, lending) and incidental activities (e.g. custody and asset management) || · Quantitative limits on certain transactions between a BHC and its affiliates (FRA Section 23A); · Transactions between BHC and affiliates to be on market-based terms (FRA Section 23B). || BHC ||

Prohibition on proprietary trading and investments in certain funds || Ownership separation || · Proprietary trading, except permitted activities: - Underwriting and market-making; - Risk-mitigating hedging activities; - Trading in US government securities - … · PE/hedge fund investments, unless (de minimis): - Less than 3% of individual fund; - Less than 3% of bank’s Tier1 capital in aggregate. Permitted /de minimis activities may not lead to: · Material exposure to high-risk trading strategies or assets; · Material conflict of interest with customers & counterparts; · Threat to safety of bank; or · Threat to US financial stability. || Prohibition Prohibition with de minimis exemption || Any US depository institution; Any firm affiliated with a US depository institution; and Any foreign bank with a US branch or agency || Reporting and compliance regime adapted to magnitude of banks’ trading activity.

Ring-fencing of certain derivatives (“Swaps Push-out”) || Functional separation || Banks benefiting from federal assistance cannot engage in certain swaps (certain credit derivatives, all equity and most commodity derivatives). If they do, such swaps must be transferred to a separate legal entity, registered as a swap dealer and subject to capital requirements and margin requirements under the derivatives sections of the Dodd-Frank Act. Swap entities do not have access to federal assistance. || || Any US depository institution; Any firm affiliated with a US depository institution; and Any foreign bank with a US branch or agency ||

UK || Ring-fencing of retail banking activities || Functional separation || · Core activities (deposit taking) · Core services (incl. withdrawal, payments and overdraft) || Legal separation - Separate legal entity ring-fenced from group; - If RFB owns other organisations, those only do activities permitted to RFB; Separate governance - RFB independent board; - RFB separate disclosure; - Boards RFB and NRFB have duty to uphold ring-fence. Operational links - Operational independence irrespective of state wider group; - RFB member of payment systems; - Separate remuneration structure and human resourcing; Economic links - Relations between RFB and non-RFB on third party basis; - Transactions RFB-NRFB on commercial, arm’s length basis; - Assets sold at market value - Independence of capital and liquidity. - Separate risk management - Separate balance sheet (treasury) management; - Dividends/transfer of capital only if board believes sound; || All UK incorporated banks subject to exemptions: - Not building societies; - Not banks below certain size (Secondary legislation, indication GPB25bn) || Additional loss absorbency (equity, debt): - Ring-fenced bank: primary loss absorbing capacity (PLAC) of 17% RWA (equity buffer 3%, remainder either equity or loss absorbing debt); - Non-ring-fenced bank: at least 17% PLAC against all domestic and non-exempt overseas operations. Annual review of ring-fence effectiveness by PRA Reserve powers (electrification of fence) to regulators to require ownership separation from specific banks if attempt to circumvent.

FR || Ring-fencing of certain trading activities || Functional separation || 1. Proprietary trading 2. Unsecured lending for own account to leveraged entities || Legal separation - Separate legal entity ring-fenced from group Economic links - Independence of capital and liquidity - TE not take guaranteed deposits - TE not provide retail payment services - Tighter intra-group large exposure limits Separate governance - Different commercial identity; - Different leadership || De minimis exemption for banks with trading activity below a certain threshold (secondary legislation) ||

Prohibition of certain trading strategies and trading in certain instruments || Ownership separation || Segregated entity cannot engage in following activities: - High frequency trading - Trading in agricultural commodity derivatives || || ||

DE || || Functional separation (ring-fencing of certain trading activities) || 1. Proprietary business; 2. Lending and guarantee business with hedge funds and AIFs; 3. Proprietary trading with the exception of market-making || - Commercial separation; - Legal separation; - Organisational separation; - Own refinancing of financial trading institution (FTI) to be secured independently; - Group transactions on third party basis; - Reporting obligations towards the BaFin (secondary legislation) || Banks and enterprises belonging to the same group as a bank. De minimis exemption for entities with trading assets (HFT+AFS) that are: - below €100bn; or - below 20% of total assets if total assets are below 90bn. ||

HLEG || || Functional separation (ring-fencing of certain trading activities) || 1. Proprietary trading; 2. Market-making; and 3. Loans/unsecured exposures related to hedge funds etc. || - Separate legal entity – trading entity (TE) - Can operate within holding company structure; DB must be sufficiently protected from TE risk. - RRP to ensure operational continuity of IT/payment system infrastructures in crisis. - Independence of capital and liquidity - Intra-group  transfers on market-based terms - Intra-group transfers according to normal large exposure rules - Dividends and direct/indirect intra-group transfers only possible if still respect capital requirements (incl. buffers) - TE not take guaranteed deposits - TE not provide retail payment services || De minimis exemption, as only apply to banks with significant trading activity: - Examination thresholds: Relative threshold of 15-25% of the bank’s total assets; absolute threshold of EUR100bn; - Separation threshold: supervisors determine need for separation based on share of assets to which the separation requirement would apply. Threshold, to be calibrated by the Commission. ||

Annex A2 – Summary of replies to the stakeholder consultation on structural reform of the banking sector

In the context of the impact assessment accompanying a potential legislative proposal on reforming the structure of large EU banks, the Commission services have conducted a public stakeholder consultation. The consultation was open for eleven weeks (16th May 2013– 11th July 2013). It contained both qualitative and quantitative sections, with the former focusing on questions related to the need for EU action and the different options for legislative reform, and the latter containing a data template for banks to provide data on short and medium term implications of different reform scenarios on their balance sheets. This document summarises the responses to the qualitative part. This note follows the structure of the consultation document and provides a high-level summary of the nature of responses of different stakeholders. The following graphs accompanying each section of the consultation document indicate the proportions of each category of respondents that gave a certain answer[42].

The Commission services received 540 replies. These responses came from the expected type of respondents: banks and other financial institutions, corporate clients, investors, public authorities, and consumer associations and individuals (analysis shown in Graph 1). However, while the composition is fairly traditional, the number of responses from individuals (439) and consumer associations (11) stand out. The majority of these 439 replies took either the exact, or abbreviated, form of a recently-publicised Finance Watch response to the consultation. 

Graph 1: Overall breakdown of consultation respondent by stakeholder type.

The following graph shows the breakdown of all other responses by nationality.  It may be worth noting that the relatively high number of responses from Belgium is composed mainly of international organisations headquartered in Belgium, but representing several European countries and member associations.

Breakdown by nationality of the 439 responses from consumer individuals is excluded, as respondents did not indicate within their response from which Member State they wrote. The majority of consumer individuals responded in English (270), followed by French (131), German (22), Italian (15), and 1 response was written in Danish. 

Graph 2: Overall summary of responses by nationality

1. Problem drivers

The consultation document outlined the problems that continue to affect the EU banking sector and outlined the potential contribution structural reform could make in addressing those problems. Stakeholders were then asked whether structural reform of the largest and most complex banking groups could address and alleviate these problems. The answers are summarised in Graph 3.

Graph 3: Are banking structure reforms relevant?

The above graph, showing the type of response as a percentage of overall responses from each group of stakeholder, excludes the responses of individual consumers. This large group's inclusion skews the graphic representation.  Individuals responded in large numbers in favour of the reform proposal (433 out of 439 in favour, but with 4 disagreeing with the proposal). Along with the above graph, it is clear that there is a distinct fault line between the responses of banks, on the one hand, and consumers and non-bank financials on the other hand. The former are to an overwhelming extent against structural separation (with the exception of some cooperative banks). The latter are largely in favour. The views of other categories are more balanced. Corporate customers, while acknowledging the need to address TBTF, express opposition, based on the potential impact of such reforms on the cost of financing.

As shown in the graph, some cooperative banks, consumer associations, non-bank financial companies, and public authorities added the comment that they agreed with the proposal, but that it should target smaller banks and should consider the detriment of a shadow banking sector in the proposal process. 

2. Subsidiarity

The consultation document then highlighted the on-going reforms within Member States and outlined the potential benefits of action at the EU level, that is, to preserve the integrity of the internal market. It then asked for stakeholders’ views on whether they considered an EU proposal in the field of structural reform necessary.

Graph 4: Is EU action needed?

Views on the need for EU action mirror the views on the merits of such reforms in the first place. There is accordingly a large opposition from most banks (excluding cooperatives) and non-financial corporates. Consumer associations and individuals are again unanimously in favour. It may also be interesting to note the relatively high number of consumer responses that supported allowing Member State legislation to "go further," than supranational reform, if possible.

3. Policy options

The remainder of the consultation document asked for views of stakeholders on the three different elements of bank structural reform: the scope of banks to be subject to potential separation, the activities to be separated, and the strength of separation. It also asked for stakeholders’ views on the best combination of activities and strength.

3.1. Scope of banks

As regards the scope of banks, the consultation document asked questions related to, first, the threshold for banks to becomes subject to separation, and second, the extent of supervisory judgement in applying the threshold.

As regards the threshold for separation, the document outlined four options:

(1) Using the HLEG definition (Assets held for trading and available for sale);

(2) A more narrow definition for separation, which excludes available-for-sale assets, as they are mostly composed of securities held for liquidity purposes;

(3) A definition focused on the gross volume of trading activity, which is likely to focus on proprietary traders and market-makers;

(4) A definition focused on net volumes, which is likely to only capture those institutions that have a higher share of unbalanced risk trading (proprietary traders).

It then asked stakeholders which of the four definitions would be the best indicator to identify systemically risky trading activities. If none of the above, it asked stakeholders to propose an alternative indicator. The views of stakeholders are summarised in Graph 5a. Many stakeholders, particularly consumers, did not respond to this rather technical question. Of those who did, many rejected the HLEG recommendation (option 1). Most banks, as well as public authorities, argue in favour of a risk-based approach. For those banks that expressed an opinion on the four options given (standard and cooperatives and savings), the majority favours option 4 (net volumes). Among other stakeholders, the debate lays between gross and net volumes, with the latter having the most widespread constituency.

Graph 5a: Threshold options by total consultation response.

Graph 5b: Threshold options by respondent

As regards the degree of supervisory judgement, the consultation document outlined three options:

(1) Ex post separation subject to constrained discretion by the supervisor;

(2) Ex ante separation subject to evaluation by the supervisor; or

(3) Ex ante separation.

It then asked for stakeholders’ views on which would be the most appropriate approach, including suggestions for alternatives. The pattern of responses is similar to views of the fundamental merits of structural separation. Bank respondents however are more divided, already showing some acceptance of separation with limited supervisory discretion, mostly to ensure a level playing field in the market and ensure legal certainty. The graph below does not reflect the responses of individual consumers, as the volume of these respondents again will skew the graphic illustration.  The consumers who did answer this question (216) expressed complete support for no supervisory discretion.

Graph 6: The degree of supervisory judgement (excluding individual consumers)

Of those non-individuals who responded to this question, 3 of the 18 total standard banks responded favourably to leaving no supervisory discretion, with ex ante separation.  The graph below depicts the composition of respondents (excluding the 216 individual consumers) who responded in the same way.

Contrastingly, a third of all standard banks who responded to the consultation replied in favour of either entire or important discretion:

3.2. Activities

The consultation document highlighted the broad range of activities banks may engage in and provided three scenarios for separation, ranging from only some activities (e.g. proprietary trading, PT) being separated and the trading entity thus remaining ‘narrow’ and the deposit-taking entity ‘broad’; to a scenario where many activities (e.g. all wholesale and investment bank activities, WIB, would be separated and the trading entity accordingly becoming ‘broad’ and the deposit entity correspondingly ‘narrow’. More specifically:

(1) “Narrow” trading entity and “broad” deposit bank;

(2) “Medium” trading entity and “medium” deposit bank; or

(3) “Broad” trading entity and “narrow” deposit bank.

It then asked a question specifically related to the separation of market-making (MM) and underwriting activities. This question has served as a foundation for classifying responses into the three options highlighted above.

Responses here again reflect the general pattern of replies. Bank responses are divided between those who argue against separation, and those who argue that if there is to be separation, then it should only focus on proprietary trading. This sentiment is echoed by corporates, who argue in favour of an as narrow separation as possible in order not to affect their access to (low cost) financial services. Individuals and consumer associations on the other hand favour option 3, i.e. separation of all investment bank activities.

Graph 7:  Which activities should be separated?

Consumer individuals are excluded from the above graph, as 243 extra favourable respondents skewed the scale of the chart. The individuals who were in favour of the reform and who responded to the question, however, were all in favour of separating all wholesale and investment activities. As shown in the graphs, standard banks who accept a certain degree of separation favour separating proprietary trading only from the deposit-taking entity, much like non-financial companies and some public authorities.

Graph 8: Activities to be separated, by respondent

The consultation document also highlighted the particular case of risk management products, where the banking group and its constituent entities not only need to be able to engage in prudent risk management practices for treasury purposes, but where the deposit entity may provide risk management products to its clients. The consultation document therefore asked if deposit-taking entities should be allowed to directly provide risk management services to clients, and if so, whether any additional safeguards should be considered.

Many stakeholders did not respond to this rather technical question. Of those who did, many (most banks) responded in favour of the deposit-taking entity being able to provide such services. Some argued in favour of safeguards (e.g. simple derivatives, caps). Consumers, some non-bank financials (investors) and public authorities were against such provisions.

Graph 9: Should the deposit entity be able to provide risk management products?

3.3. Strength of separation

The consultation document subsequently highlighted different forms of separation and indicated that the Commission services were considering three degrees of strength:

(1) Functional separation with economic and governance links restricted according to current rules;

(2) Functional separation with tighter restrictions on economic and governance links; or

(3) Ownership separation (full prohibition).

It then asked stakeholders for their views on the pros and cons of stricter legal and economic separation, as well as views on full ownership separation.

· As regards legal and economic separation, the classical pattern can be observed, with some cooperatives being concerned about the potential implication of structural reform on their reverse-ownership structure;

· As regards ownership separation, this has polarised opinion, with most banks highlighting the costs, and consumer associations and individuals and some non-bank financials highlighting it as the simplest and most effective option in the longer term.

Graph 10: Is there a case for stricter legal separation?

Graph 11: Is there a case for stricter economic separation?

Graph 12: Is there a case for full ownership separation?

Non-financial companies have been excluded from the above graphs, as they consistently did not reply to the questions on the strength of eventual activity separation.

As shown by comparing the three graphs, there is a high level of consistency across respondents.  That is, a large proportion of respondents in favour of increased economic separation are also in favour of increased legal separation, for example.

3.4 Options

The consultation document then highlighted a number of preliminary and illustrative combinations of different degrees of activity restrictions and separation degrees that would be subject to further assessment in terms of costs and benefits (Table 1).

Table 1: Overview of policy options

Activities strength || Functional separation 1 (SUB) Current requirements || Functional separation 2 (SUB+) Stricter requirements || Ownership separation Ownership separation

Narrow trading entity/ broad deposit entity E.g. Proprietary trading  + exposures to HF (PT ) || Option A || Option B [≈ FR, DE baseline] || Option C [≈ US Volcker]

Medium trading entity/ medium deposit entity E.g. PT + market-making (MM) || Option D || Option E [≈ HLEG; ≈ FR, DE if wider separation activated] || Option F

Broad trading entity/ narrow deposit entity E.g. all investment banking activities || Option G || Option H [≈ US BHC; ≈ UK] || Option I [≈ Glass-Steagall]

It then asked stakeholders whether (i) the above matrix captured a sufficiently broad range of structural reform options and (ii) which option would, in stakeholders’ views, best address the problems identified. Fundamental views show up again, with a large portion of banks expressing a blanket opposition to structural reforms or endorsing option A (PT Only). Consumer associations and individuals, however, argue that option E (HLEG) is the minimum effective option and express a preference for either option H (UK ICB) or I (Glass-Steagall).

Graph 13: Preferred reform approach

As with several of the other responses, it is worthy to note that the respondents do not necessarily add up to the total number of responses received by the Commission.  For this question in particular this is especially evident, as many stakeholders expressed their preference for Option E as a minimum option, but said that combinations such as H and I or F and I would also work.

Only one public authority cited capping total assets as the preferred reform method. Others were divided fairly evenly over the range of options.

With regards to consumer associations, four respondents preferred option H (~UK), while 14 respondents preferred the Glass-Steagall type option on structure, or option I for separation. Only Option G received no support from any respondent.

4. Conclusion

While consumers are overwhelmingly in favour of the proposal for structural reform and banks are predominantly against, where respondents addressed the numerous detailed choices within the proposal, they displayed a wide range of views.  Of course especially for the final policy option and the divide between banks and consumers regarding supervisory discretion, some dominant choices appear.  Accompanied by several detailed, analytical responses to the consultation document, as well as the quantitative feedback from banks, the Commission has gleaned much information from the varied group of respondents.

Annex A3 – Assessing the complementarity of structural separation

Since the start of the financial crisis, the European Commission (the "Commission") has – in parallel to the management of the crisis by national authorities, state aid control by the Commission, and non-standard interventions by central banks – engaged in an ambitious overhaul of financial system regulation and supervision.[43] The objectives of those reforms have been to create a safer, sounder, more transparent and responsible financial system that works for the economy and society as a whole. In particular this has been done by: (i) strengthening capital requirements related to trading, securitisation and derivatives activities (the CRDII and CRDIII)[44] and to improve both the quantity and quality of bank capital more broadly (the CRR/CRDIV)[45]; (ii) introducing rules that enable authorities to better manage failing banks (the BRRD)[46]; (iii) creating a stronger micro- and macro-prudential system of supervision at EU level by the creation of the European Supervisory Authorities and the European Systemic Risk Board (the "ESRB") and the proposal on the Single Supervisory Mechanism (the "SSM") as a fundamental pillar of the Banking Union; (iv) strengthening the regulatory framework for the trading of different classes securities, the provision of investment services and activities and the treatment of OTC derivatives (the MiFID review and EMIR).

The case for structural reform depends on demonstrating in what ways such reforms could complement those reforms. The purpose of this annex is not to provide an in-depth, comprehensive description of the details of all these large and complex reforms. Instead, this annex: (i) briefly outlines the objectives and principles of the major reforms either agreed or currently under negotiation; and, (ii) assesses the extent to which bank structural reform could complement already existing tools that provide powers to take measures aimed at: (a) strengthening bank capital, governance, supervision, recovery and resolution, derivatives clearing and trading, and, taxation of financial transactions, and (b) restricting or limiting the legal and economic organization and operation of banks. The annex also takes stock of how bank structural separation would affect the general tension of increased regulation leading to a shift of activities towards the less regulated shadow banking sector.

1. Capital

Banks play a critical role for the functioning of the financial system and the economy as a whole. The business of credit intermediation is nevertheless inherently unstable.  Prudential requirements to back up a bank's balance sheet by a certain level of own funds in order to absorb losses have therefore since a long time been a key instrument to control and limit these inherent risks. The financial crisis highlighted shortcomings both as regards the level and loss-absorbing nature of banks’ own funds.

Efforts have therefore been undertaken at both international (Basel Committee on Banking Supervision (the "BCBS")) and European level (changes to the Capital Requirements Directive (the "CRD")) to improve the quantity and quality of bank capital. At the early stages of the crisis, capital requirements related to trading, securitisation and derivatives activities were strengthened in order to address the materially undercapitalised trading book exposures (the CRDII and CRDIII). The crisis also initiated a more fundamental review of the capital requirements framework at international level (Basel III), requiring banks to have more and better quality capital overall. The BCBS has also developed additional capital buffers for banks that are of global systemic importance. These changes have been incorporated into European law by the recent CRDIV.[47] 

The reforms to bank capital requirements will reduce incentives to take excessive risks. It will also enable banks to absorb more losses before defaulting. These two effects will reduce the probability of default. The increased capital requirements on banks’ trading books may also reduce banks’ rapid balance sheet growth.

Even so, risk weighted assets as compared to total assets for large cross-border banks, which typically have an important trading book, are significantly lower than for other banks.[48]  Moreover, the risk-based capital requirements based on value-at risk ("VaR") model calculations can still be small compared to the size of trading assets.[49] This explains why some measures have been taken[50] and that standard setters at both international (BCBS) and European level (by the European Banking Authority (the "EBA")) are assessing the consistency and accuracy of the risk-weighted asset approach.[51]

The subsequent revisions to the Basel agreement and the CRD build on the same regulatory approach as the previous Basel framework (risk-based capital requirements), even though it also introduces a leverage ratio designed to serve as a backstop to risk-based capital requirements. Trading, securitisation, short term wholesale funding and other activities were all supposed to be assessed by supervisors as part of the Pillar 2 supervisory review process introduced by Basel 2, as well as partially by the Pillar 3 process of market disclosure. It therefore remains to be seen whether these revisions will address all the difficulties of supervision and market monitoring.

In addition, CRDIV enhances the Pillar 2 supervisory review process by providing supervisors with powers to impose certain structural measures on a bank when it is at risk or pose a risk to the entire financial system.

The capital requirements framework sets capital and liquidity requirements that depend on the riskiness of the individual entities and/or of the consolidated group. Structural reform could therefore complement the capital requirements framework by imposing ex ante stricter capital constraints on specific activities which are likely to generate the most risk. If as a result, complex activities no longer benefit from being covered by the public safety net, this would remove one source for excessive growth of such activities.

More fundamentally, capital requirements do not address conflicts of interest and cultural problems, and as such do not address the problem drivers associated with structural reform. Structural reform (i.e., separation) combined with strengthened governance rules could therefore prevent the investment bank side to affect the business of the commercial banking part of an integrated group.[52]

Moreover, irrespective of the changes to capital requirements that increase the amount of capital that needs to be held, banks may still have significant incentives for engaging in risky activities that yield substantial profits, as they do not have to fully cover the potential losses arising from such activities.[53] Given limited liability as regards losses, the activities in question would accordingly yield excess private returns. This could induce a broad-based shift towards these activities, with an increase in systemic risk being the consequence.

Finally, structural reform could also complement the CRDIV powers to impose structural measures by setting a uniform standard for how relevant authorities should review the need for and subsequently apply structural measures.

Structural bank reforms could complement the reforms related to capital requirements by:

– Imposing ex ante constraints on specific activities, as opposed to ex post capital requirements that depend on the riskiness of the individual entity and/or of the consolidated group;

– Ex ante addressing remaining incentives to engage in excessive trading activities over credit activities. Structural reform would be a more direct way of addressing excess returns resulting from insured deposit the benefits of which currently flow freely throughout integrated groups. It could lead to a better pricing of risk, which over time could help reducing excessive trading activity and thereby limit the build-up of systemic risk. This could also complement the systemic risk charges for SIBs by adding another disincentive towards banks excessively expanding their risky trading activities, thus putting a break to the main source of bank growth in recent years;

– Allowing a more effective and transparent tailoring of capital requirements to the different legal entities. A structural separation would entail different entities holding separate capital and liquidity buffers aligning the prudential requirements more closely with the risk. This promotes market discipline;

– Setting a uniform standard for how relevant authorities should review the need for and subsequently apply structural measures; and

– Facilitate Pillar 3 market monitoring by providing more transparent group structures that match main business lines, and by providing more disclosure of the data of the segregated business entities.

2. Governance

One of the key contributors to the financial crisis has been the implicit or explicit guarantees of bank deposits and other liabilities, which induced excessive risk-taking (including, for example, high leverage and own-position taking) and bad lending and acquisition decisions. The problem was exacerbated by ineffective governance arrangements within banks, which did not offer sufficient checks and balances within their internal processes, resulting in a lack of effective oversight of management decision-making and excessively short-term and risky management strategies. This was compounded by ineffective ownership and control due to fragmented shareholder structures. Is that not the same thing?

Another contributing factor has been passive debt holders. This passivity is partly due to the complexity of banks. Banks are engaged in many activities and active across many markets and their balance sheets are accordingly complex and opaque. This limits the ability of investors to fully understand banks and exercise effective scrutiny. The complexity is mirrored in e.g., the prudential regulation of banks (e.g., the Basel capital adequacy framework), which is also complex and difficult for investors to understand. Accordingly, investors have not fully exercised the "watch-dog" function granted to them under Basel's pillar 3 (market discipline) framework.

The high leverage and guarantees have encouraged boards to give undue focus to return on equity (and make decisions that do not meet a bank's true overall cost of capital).

Current reforms address many of these problems, notably by imposing leverage limits (the CRR/the CRDIV), by strengthening banks’ and investment firms’ internal governance structures, enhancing risk oversight by boards, improving the status of the risk management function and ensuring effective monitoring of risk governance by supervisors (the CRDIII and CRDIV and the MiFID review) and by more clearly exposing debt holders to risks of failure (the BRRD provisions on bail-in). However, current reforms do not address the intra-group subsidies arising from the public guarantees on deposits or the issues arising from the complexity of group structures. Furthermore, internal management structures (e.g., boards) are an imperfect substitute for market discipline.

Structural reform could therefore strengthen bank governance by e.g., providing for separate and independent governance structures for deposit taking banks and the trading entity of a group. Such decentralised governance structures could lead to better managerial focus on managing the risks specific to those business lines.

3. Supervision

Bank supervision is essential to monitor a bank's risk profile and intervene if the bank's own management does not take appropriate action to reduce excessive risks. Since the start of the crisis, the CRD has been strengthened to enhance and harmonise national supervisors' powers. Also the review of the MiFID has reinforced the powers of securities regulators, at national and European level (the European Securities and Markets Authority). Moreover, most if not all supervisors in the EU have reviewed their supervisory approaches and increased their resources.

Supervision of cross-border banks represents particular challenges. The financial crisis laid bare a gap between transnational banking markets and cross-border groups and essentially national supervisors. The CRDII package recognised the important role of Colleges of Supervisors for these purposes. Since 1 January 2011, the newly created EBA coordinates national banking supervisors. The EBA has binding mediation powers, notably as regards approving banks’ internal risk models, and the Pillar 2 supervisory review. In order to strengthen macro-prudential supervision, the EU has set up the European Systemic Risk Board (the "ESRB") that is responsible for issuing early warnings and recommendations in relation to the build-up of systemic risk. In order to address the negative feedback loops between bank stability and Member States' finances, the European Council in June 2012 asked for a road map to achieve a genuine Economic and Monetary Union. Following a specific call from the Euro Area Summit, on 12 September 2012 the Commission presented legislative proposals for the establishment of a single supervisory mechanism within the EU, as a first step towards Banking Union. Political agreement was reached in December 2012 and the SSM should become fully operational in summer 2014.

While the above measures have and will significantly strengthen the supervision of cross-border banks in Europe, supervision of the largest and most complex banks in the EU will remain challenging, given their complexity, size, and interconnectedness and with business lines where risk profiles can change significantly in a very short time and thereby risk outpacing supervisory control.

Structural reforms could complement these reforms by adding further transparency by e.g., separating the activities that are most complex into a separate legal entity. This could facilitate supervision and could allow market discipline to more effectively assist supervisors.

4. Recovery and resolution

Neither strict prudential rules nor close supervision can exclude bank failures. In order to ensure that such failures can be managed without impact on the stability of other financial institutions or financial markets, and without recourse to public resources, the Commission proposed in June 2012 a Bank Recovery and Resolution Directive (the BRRD) requiring all Member States to have in place resolution regimes, consisting of a number of elements: (i) an obligation on banks and certain investment firms to draw up recovery plans and on authorities to prepare resolution plans; (ii) a power to authorities to require changes to banks legal or operational structure if they present an obstacle to resolvability; (iii) early intervention powers; (iv) a harmonised set of resolution powers (including e.g., bail-in, whereby the bank would be recapitalised with shareholders being wiped out or diluted and creditors having their claims reduced or converted to shares); and, (v) the creation of dedicated resolution funds in order to ensure that resolution authorities can resolve banks without recourse to taxpayers’ money. Negotiations in the European Parliament and the Council on the proposal have reached an advanced stage and are about to close.

The implementation of this proposal would make it easier to stem bank problems at an early stage as well as making it easier to resolve banks once they are beyond the point of repair. It would thus help to address one of the major problems highlighted by the crisis, i.e., the inability of banks to default in an orderly manner without leading to systemic disruptions.

This framework, (combined with the proposals to maintain resolution and supervision of larger euro area banks at the supranational level through the Single Resolution Mechanism and the SMM) addresses impediments to resolving banks in an orderly way. Nevertheless, structural separation could facilitate the effective and timely resolution of banks. First, structural separation along the lines considered in the main report would facilitate resolution because risky activities would either no longer feature on a bank's balance sheet or be lodged in a separate subsidiary ring-fenced from the rest of the group. In either case, the risks would have less potential to contaminate the remaining part of a bank's activities.

Second, while the implementation of an effective bank recovery and resolution framework is a critical part of the on-going financial reform programme, the preparation of recovery and resolution plans (RRPs) will be particularly challenging for especially the largest EU banking groups, given their organisational complexity, interconnectedness, international scope, and ability to rapidly expand their balance sheet.[54] Structural banking reform can therefore facilitate resolvability.[55] More specifically, by complementing the RRP efforts, structural reform will be able to benefit from and build on the resolvability assessment conducted under the BRRD framework, and spur a mutually reinforcing progress toward removing impediments to recovery and resolution for the institutions concerned.[56] 

Furthermore, resolution may be rendered more complex by the co-mingling of activities within legal entities and financial dependencies within financial groups.[57] Large EU banking groups have a complex legal and corporate structure, in some cases involving a myriad of different legal entities. During early intervention or resolution this can make it difficult to identify and isolate the root of the problem. Structural reform could enhance the simplification of groups by providing for competent supervisor with the ex ante ability to separate certain activities. This simpler structure could facilitate the exercise of the discretionary powers provided for by the BRRD to reset the group structure and to remove further impediments to resolvability stemming from the group operational or legal structures and therefore improve the overall resolvability of banks. Resolving complex banking groups may also be difficult because of the interconnectedness of their subsidiaries and their centralised support functions. Central to the resolution and restructuring of a failing firm (and potentially restoring the viability of the bank, or parts of it) is the ability to identify and assess the losses and the capital required to meet these losses. The writing down of creditor claims and the conversion of debt claims to equity are important steps early in a resolution process. Timing is critical in this process. While it is important to value these losses accurately, it is also important that creditors are treated fairly and quickly, and that any restructuring is done efficiently. Perceptions of a disorderly process can encourage fears of contagion.

Impediments to resolution may appear gradually, for example as a bank's business model or activities change or as deficiencies in the governance structure and performance emerge; but the risk that those impediments carry can materialize suddenly. It could be that a process to remove impediments (operational or governance-related) that rely on multi-agency consultation and cooperation (and in the case of cross-border groups, different authorities and supervisory colleges) would sometimes face challenges in responding quickly enough to ensure that these risks do not materialize. Ex-ante separation could mitigate for this by separating the banks that would be the most difficult to resolve, before risks materialize and threaten its viability.

The early intervention and resolution tools are new and authorities need to gain practical experience. For these tools to be effective, authorities need to invest significant resources to scrutinise banks as to their effective recovery and resolvability. This applies in particular to the potential changes to banks' legal and operational structure resulting from the RRP process. The burden of proof to demonstrate the need for such measures will be on authorities, and may be tested in court.

Structural reforms could complement these reforms in a number of ways:

– The top-down – i.e. horizontal legislation setting ex ante rules applying to similar entities – structural change implied by structural reform could complement the preventative powers of the BRRD that may result in a bottom-up reorganisation (i.e. individual ad hoc actions addressing specific impediments to resolution at individual banks) of banking groups. The combination of these two processes could over time lead to a greater consistency between business lines and legal structures. Structural reform could also facilitate the preventative powers by providing a base upon which they would be exercised, thus limiting the degree of divergent outcomes.

– It is an ex ante tool to address a broader set of objectives beyond facilitating the orderly recovery and resolution of a banking group. Structural reform also aims at reducing the complexity, interconnectedness, cultural problems and conflicts of interest between the different banking entities within a given group, and aims at reducing the excessive growth of bank balance sheets by constraining the coverage of the public safety nets to specific activities only.

– By simplifying balance sheets and by introducing functionally segregated and autonomous balance sheets rather than one large complex one, separation could expand the options for resolving banking groups. Also, separation could limit the scale of the task on an entity basis and thereby make it more feasible to apply the different resolution tools. This could also allow a swifter resolution process, as it would be easier to identify problems and apply targeted solutions;

– It would give authorities the option of treating different entities of the group in a separate way. Different resolution strategies can be applied to different parts of the group;

– Separation could make it easier to identify and assess losses as they would either no longer be accounted for, or be accounted for in separate entities. This could e.g., make it possible to apply differentiated bail-in requirements, if appropriate, to the different parts of the group;

– Separation could also improve market discipline by ensuring that the losses arising from certain risky activities have to be borne solely by the creditors and investors of that entity, and not be diluted across the creditors and investors of the whole group. Those creditors and investors would accordingly have a greater interest in the risk that the entity is running, and creditors and investors in other less risky activities would be better protected.  

5. Trading and clearing of financial instruments, including derivatives

Market developments and experiences amid the crisis showed that the existing EU regulatory framework centred on trading activities on shares and on the operation of regulated markets needed updating. Regulation needed to address challenges posed by a more complex market reality and an increasing diversity in financial instruments and methods of trading. This was the basis for important regulatory interventions in this area, also in line with G20 commitments, such as the reviews of the Markets in Financial Instruments Directive (the "MiFID") and the Market Abuse Directive and the legislation on short selling and certain aspects of credit default swaps. Focusing on derivatives, these instruments have been growing significantly over the last decade. Derivatives markets today constitute a major part of some banks' investment banking activities. The failure of U.S. Lehman Brothers highlighted the size and interconnectedness of derivatives exposures and the difficulties to manage and orderly resolve counterparty risk. It also underlined the lack of transparency in derivatives market, which is mostly carried out in private over-the-counter (“OTC”) venues rather than on public trading venues.

In order to limit contagion between banks from OTC derivatives positions, the European Market Infrastructure Regulation (the "EMIR") requires that, as of 2013, OTC derivatives must be collateralised and that standardised transactions be cleared by Central Counterparties ("CCPs") that interpose themselves between the parties and assume their counterparty risk. Moreover, the EMIR establishes trade repositories that will collect information on non-standardised derivatives, with the aim to increase transparency for regulators.

The MiFID review includes a proposal to require that all standardised derivatives be traded on trading venues, it introduces transparency requirements for non-equities, including derivatives and strengthens powers of supervisors, including in the area of commodity derivatives.

The increased use of CCPs and higher collateralisation of derivatives rights and obligations will limit the risk of contagion. The cost of OTC derivatives will increase as a result of counterparty requirements to better internalise the associated risk. This may contribute to reducing the growth of the balance sheet of those EU banks that are significant users and/or providers (i.e., dealers) of derivatives.

Structural reforms could still complement these by reducing the extent to which depositors are exposed to counterparty risks. Moreover, multilateral clearing on CCPs will only apply to those derivatives that are sufficiently standardised.[58] The remaining ones will be subject to an enhanced bilateral clearing process (e.g., higher collateral requirements).

Furthermore, the MiFID review includes a measure that would provide for an organisational separation between a bank dealing-on-own-account business and the operation of a trading venue. While this will address one instance of conflicts of interest (the MiFID II is focussed on protecting market integrity and investors and is not intended to deal with issues concerning the setting-up of different legal entities for the provision of different services or activities) it will not require separate subsidiaries with stand-alone capital. Neither does it deal with the separation of traditional banking activities (e.g., deposit taking and lending) from the provision of investment services and activities. It is accordingly unlikely to affect intra-group subsidies.

Structural separation could complement this provision by requiring that certain forms of trading activity would have to be located in a separately capitalised and funded entity.

6. Taxation of financial transactions

The Commission initially proposed a Financial Transaction Tax (the "FTT") to be implemented by all 27 Member States by September 2011. However, following discussions in the Council, it became clear that unanimous support for such a tax would not be reached. In October 2012, following a request from 11 Member States[59], the Commission proposed a decision to allow enhanced cooperation on the FTT. After an agreement had been reached between the European Parliament and the Council, the Commission put forward a detailed proposal for a FTT in February 2013.

The objective of the FTT is to: (i) tackle fragmentation of the Single Market that an uncoordinated patchwork of national financial transaction taxes would create; (ii) ensure that the financial sector makes a fair and substantial contribution to public finances; and, (iii) create disincentives for financial transactions that do not contribute to the efficiency of financial markets or to the real economy.

The base for the FTT is broad, covering all transactions carried out by financial institutions on all financial instruments and markets that have an economic link to the FTT-zone. While the FTT base is broad, the tax rates are low.[60] Even so; the FTT would have to be paid by each financial institution involved in the transaction.

If adopted in current form, the FTT could have a significant impact on the cost base of some of the risky activities that are being considered for separation. These activities could accordingly be expected to decrease in importance. For example, the impact assessment accompanying the FTT foresees a significant reduction in derivatives activity, albeit with a strong degree of uncertainty. The FTT would in that case achieve one of the objectives of this exercise, namely to reduce incentives to engage in excess trading activity at the expense of real economy activities.

However, the FTT is at an early stage of negotiations and the contours let alone detailed specifics of the proposal are far from settled. Second, even if adopted, the FTT would only apply to banks incorporated in the FTT zone. This excludes banks incorporated in some of the major EU financial centres, which would make it susceptible to circumvention by means of relocation.[61]

Nevertheless, given the significant impact of the FTT on transaction costs, banks would have strong incentives to find a way around the safeguards.

Accordingly, bank structural reform applicable to EU28 could provide an additional safeguard to the FTT and limit regulatory arbitrage.

7. Shadow banking

Structural reform of banks may trigger a migration of activities away from the regulated banking sector to the unregulated or less regulated "shadow banking sector", i.e., institutions of different kinds that provide services similar to the core banking functions (e.g., liquidity transformation and lending) thereby falling outside the scope of application of the EU banking prudential rules which can increase systemic risk and lead to financial stability issues. Although, when risks move to a less regulated sphere and out of the scope of the supervisors’ screen it is also the case that the activity would no longer benefit from the implicit public safety net.[62]

In pursuing structural reform, policymakers therefore need to be vigilant to avoid that much of the regulated activities migrate elsewhere. Banks may actually invest in products provided by shadow banking entities or even acquire some of these alternative institutions providing credit intermediation functions.[63]

Accordingly structural reform would create a need for parallel regulatory actions in order to address a potential risk of regulatory arbitrage. 

This would particularly be the case if some trading activities were to be isolated in a separate entity within a banking group.  In this case, they should remain covered by high prudential standards and appropriate supervisory arrangements. The status of such “trading entities” should normally be the one of "investment firm" subject to the MiFID and to the relevant prudential requirements of the CRDIV/CRR. Unless these entities are controlled, the overall net reduction of risk could be limited. [Also a strict ban on proprietary trading may require stricter prudential requirements such as, for example, stricter exposure limits between deposit taking institutions and "shadow banks".]

Structural reforms only offer an incomplete response to the above, and needs to be completed by rules addressing risks posed by the potential increase of shadow banking.

The Commission's strategy so far has been to try and tackle all financial risks in a global and comprehensive way and by ensuring that the positive effects expected from the strengthening of supervision of financial players and markets are not neutralised by a transfer of financial risks towards less regulated sectors. This strategy has been articulated in the Commission's Green Paper on shadow banking published in March 2012[64]. The overall concern regarding shadow banking is also shared by G20 leaders which have mandated the Financial Stability Board[65] to prepare policy recommendations on the topic.

Following the public consultation on the Commission’s Green Paper, the Commission now intends to set out its roadmap in a communication. This Communication will comprehensively set out the issues at stake in relation to the shadow banking system and outline the priorities for a Commission initiative such as: (i) increasing the transparency of the shadow banking sector; (ii) adopting a harmonised framework for money market funds; (iii) developing securities law to further address risks associated with securities financing transactions (i.e., securities lending and repurchase transactions); (iv) strengthening provisions for interactions between shadow banking and banks including consolidation of activities which could be moved outside the banking sector; (v) identifying bank-like activities which may be provided outside the regulated sector; and (vi) providing for supervision arrangements to ensure that specific risks are adequately addressed.

As the shadow banking sector changes over time certain areas will require further and continuous analysis, particularly on the basis of the analyses of the Commission services and the G20 final recommendations.

8. Conclusion

The on-going banking regulatory reform agenda is of vital importance. It will significantly increase the resilience of both individual banks and the banking sector as a whole. Steps have been taken to address systemic risks and an EU resolution framework will soon be in place.

Current and on-going reform measures are intended to improve financial stability and provide a more robust base for economic growth and a well-functioning Single Market for financial services. Measures such as increased bank capital requirements, improved bank regulation and supervision will all help to avoid excessive risk-taking of banks.

However, the current reforms do not directly address the problems and distorted incentives that originate in the implicit public safety net, banks’ complexity and interconnectedness. Structural reform would therefore be an important complement to on-going regulatory reforms, as it would offer one way of more directly addressing intra-group complexity, intra-group subsidies, and excessive risk-taking incentives. It also targets a broader set of objectives, such as ensuring that the deposit taking bank is not unduly influenced by a short term oriented trading culture and prone to conflicts of interest at the detriment of its customers.

Annex A4.1: Implicit subsidies: Drivers, Distortions, and Empirical Evidence

1. Introduction

In a perfectly competitive market, inefficient companies or firms may fail and exit the market. However, in the banking sector, profits are often claimed to be privatised while losses are socialised as governments have often intervened to avoid bank failures. In order to prevent widespread failure, governments have been designing bail-out programmes for banks across the globe.  The recent financial crisis has shown that policymakers are prone to do so particularly in order to bail-out large or otherwise important banks. These institutions deemed to be systemically important are typically referred to as too-big-to-fail ("TBTF").  Anticipating public support for these institutions, bondholders and depositors are willing to lower their requested return on these banks' liabilities. Therefore, these banks benefit from a lower funding cost. These benefits may stem from explicit government measures, such as deposit guarantee schemes (to the extent they are inadequately priced) as well as from the expectation that certain holders of banks' debt would not face the (full) risk of loss an explicit safety net ("implicit subsidy"). While the government safety nets can contribute to the prevention of systemic crises, they also have several adverse effects. In addition to the direct impact of imposing strains on public finances, they also lead to several market distortions. By making banking losses social, implicit subsidies lead to moral hazard as the market has a significantly reduced incentive to monitor banks' activities resulting in excessive risk-taking. Also, they distort the level playing field across large and small banks (as large banks are more likely to benefit from the funding cost advantage) and between banks headquartered in sovereigns with different ability to provide such support (depending on the state of their public finances). The implicit subsidy also leads to allocative inefficiency as it makes the financial sector more lucrative thereby luring resources from other sectors. The first part of the Annex explores these issues and explains the motivation for public intervention in banking as well as the side effects of the implicit subsidies.

In the second part of this Annex, we review the empirical evidence on the quantification of implicit subsidies. There has been significant interest by academics and policymakers on "putting a figure" on the size of the implicit government guarantee to the banking sector. By definition, the implicit subsidy is not transparent, and therefore not observable and not easy to estimate with great precision. There have been several strands of literature on the valuation of the implicit subsidy, including the funding advantage models, the contingent claims model, event study methodology, and by measuring market distortions. Empirical analyses typically confirm that implicit subsidies exist and in most cases are very significant, with subsidies reaching levels of several billion euros annually that represent a significant share of banks' profitability. However, the precise estimate of the level of the implicit subsidies is highly dependent on the exact methodology used, as well as on the sample period and countries under consideration. A summary table of the methodology and results of empirical papers is provided in Appendix A of this Annex. The Joint Research Centre (JRC) has also performed an in-depth empirical analysis based on the funding advantage model (rating methodology) in order to determine the size of the implicit subsidies and to find its main determinants, which is provided in Annex A4.2 (and is briefly discussed below).

2. Implicit subsidies: Drivers and Distortions 2.1. Why do governments intervene?

Deposit-taking banks are vulnerable to bank runs given their asset maturity transformation role. Their mix of illiquid assets and liquid liabilities (deposits that may be withdrawn at any time) may give rise to self-fulfilling confidence crises and force banks to liquidate illiquid long term assets at a loss even when they are, in reality, solvent (as explained in the seminal paper by Diamond and Dybvig (1983)). The stakes from bank runs are even larger as very often a bank run may have a contagion effect, given that banks are interconnected. If a large bank fails then other banks that rely upon this bank and its creditors to fulfil their obligations may collapse as well, and so on.[66] By creating a domino effect (through a direct contagion effect but also indirectly due to reputational or informational contagion), the failure of a TBTF bank threatens to adversely affect the real economy. A TBTF bank would receive support if the regulatory authorities consider that its failure could impose severe negative externalities upon society at large. Anticipating government interventions, depositors are less likely to run on banks. Therefore, ensuring depositor confidence may help prevent wide-scale collapse of the banking sector and facilitates banks' ability to engage in effective maturity transformation. Also, when banks fail the information capital they have developed may disappear resulting in many borrowers not having access to funds to pursue productive investment opportunities. As a result of the above, widespread banking crises significantly affect the economy's ability to channel funds to productive investment opportunities, possibly leading to a full-scale economic crisis and a large decline in investment and output.[67] In order to address these social costs of financial crises there are public safety nets measures to restore depositor confidence. These measures include deposit insurance, lender of last resort facilities (discount window central banks role) and government bailout of these institutions.

These explicit and implicit public safety nets allow banks to enjoy significant benefits, as their funding costs are artificially lowered given that creditors take into account the lower credit risk. Explicit deposit insurance measures have been introduced since the 1929 Great Depression and now they exist in more than 90 countries worldwide. In July 2010, following concerns about some Member States' banking sectors, the level of deposit protection in the EU increased from a minimum of EUR 20 000 to a uniform level of EUR 100 000, with a maximum pay-out delay of 7 days. In addition to these explicit guarantees, policymakers will be inclined to bail-out institutions that are considered to be systemically important and therefore whose potential failure could threaten the stability of the entire financial system. There is an expectation from the market participants that the government might step in and bail-out at least some stakeholders.  For example, in several cases during the recent financial crisis, bondholders did not face the (full) risk of loss, even if equity holdings were diluted. Consequently, these banks may in turn enjoy favourable treatment from market participants exploiting these safety nets. While for explicit subsidies it is possible for governments to recoup the cost of intervention by levying a charge for it (or at least for a great part of it), for implicit guarantees this is typically not possible. Therefore, to the extent they are not recouped, implicit subsidies are a transfer of resources from the government to the financial sector. Bank creditors, customers, staff, and shareholders may benefit at the expense of taxpayers.

2.2. What determines governments' decision to intervene and bail-out banks?

Concerning implicit subsidies, the authorities do not have any explicit ex ante commitment to intervene.  Therefore, by definition, implicit subsidies are not ex ante transparent and market participants cannot know with certainty which institutions and which creditors the public authorities would bail-out. Instead, when referring to implicit subsidies, it is useful to think about the bail-out as a probabilistic event (that is, assigning a probability that some institutions' creditors will be subject to a bail-out).

Stern and Feldman (2004) consider that there are three motivations for policymakers to engage in TBTF policies. Firstly, public authorities worry about the system-wide consequences of bank failures and therefore act in the public interest. Secondly, regulators and supervisors may have incentives that differ from the public. For example, they might bail-out banks to avoid looking bad in case of bank runs, or to preserve prospective career opportunities in banks for supervisors. Thirdly, public authorities may want to direct credit: protecting large banks that are often government-controlled helps encourage the public to put their savings in the institutions, giving them resources to lend according to the government's wishes. They also acknowledge that during a financial crisis it is inevitable that some decisions are made as new information emerges, and it is easier for policymakers to err on the side of caution. That is, if policymakers bail-out a bank at risk, they would risk wasting taxpayer money and provoking public outrage. If, though, policymakers wrongly do not intervene, then the whole financial system may collapse (with consequences for the real economy).

López-Espinosa et al (2013) analyse a large sample of banks operating in 24 countries and investigate whether firm-specific characteristics affect the likelihood of recapitalisation and the volume of these capitalisations. They find that banks' sources of funding, management quality, and the degree of international diversification help predict future fiscal liabilities. Leveraged banks relying on non-stable funding and fewer deposits, as well as banks with low management quality, and those more exposed to fair value securities, are more likely to require public capital. Similarly, a bank with domestic operations is more likely to be rescued, as national authorities are not likely to weigh to the same extent the global externalities from the disorderly failure of an internationally-active bank.

Credit rating agencies (CRAs) also assess the likelihood of government intervention. For example, Moody's (2011) explains that the availability of support will reflect the sovereign's willingness to provide public finds and its ability to intervene. Governments will be more willing to intervene if banks have a greater systemic importance, and will depend on the extent and type of resources required to prevent failure, and the availability of alternative policy instruments (such as resolution mechanisms).  A government's ability to intervene will depend on the government's debt rating. Also, the more widespread the problems in the banking sector are, the less likely that governments would be in a position to intervene.

2.3. Side effects of implicit (and explicit) guarantees

To the extent that they are under-priced there are budgetary implications from the implicit (and explicit) subsidies, as well as market distortions due to this implicit guarantee, which tend to create efficiency costs in the economy. Morrison (2011) argues that the most important costs of the TBTF problem are incurred ex ante, in the form of distorted incentives that arise as a consequence of distortions to the capital markets, and to the choice of banks' scale and scope.

1. Negative impact on taxpayers and public finance of ex post bail-out

As explained above, implicit subsidies are probabilistic events (that is one cannot anticipate with certainty which institutions and under which conditions governments will intervene). Sometimes bail-outs (of at least some categories of creditors) arise as governments intervene to avoid a devastating impact on real economic activity. Governments are very reluctant to let large and interconnected banks fail given the high potential costs, or if they do they would like to bail-out some categories of creditors. For example in the 1984 Continental Illinois case in the US the government also guaranteed all accounts (even if the deposit insurance was limited to a fixed amount) and prevented losses for the banks' bondholders. As shown by the recent financial crisis, governments have stepped in in several cases with significant exposure for taxpayers. For the UK alone, Morrison (2011) estimates taxpayer exposure to banking sector losses at GBP 955 billion, and taxpayers' paper loss on RBS and Lloyds shares at around GBP 12.5 billion. This UK example illustrates that implicit subsidies, at some point, become explicit, and divert resources from the public budget which might have greater added value to the society.

Furthermore, the massive scale of the adopted public recapitalisation measures, while succeeding in some cases in preventing the collapse of the financial sector, has magnified spillover effects from banking institutions to sovereigns. Negative feedback loops in banking and sovereign balance sheets have been observed as massive public interventions lead to higher sovereign risk, which in turn negatively affects banks (for example, through their exposure to their sovereign bond risk). Acharya et al (2010) find that sovereigns that announced bank bail-out programmes during the financial crisis had a substantial increase in market perception of their default risk, measured by the price of the credit default swaps (CDS). Also, the close link between banks and sovereigns is confirmed by the high correlation in the subsequent movements in CDS prices for sovereigns and large banks.

Therefore, public bail-out can impose significant financial burden on taxpayers in the form of direct funding costs, greater costs of sovereign borrowing, and inefficient allocations of public finances.

2. Moral hazard and excessive risk

From a theoretical perspective, the impact of bail-out expectations on bank risk-taking is ambiguous. On the one hand, excessive risk-taking on behalf of bank managers and personnel would arise as a result of the market discipline hypothesis. According to this hypothesis, due to the expectation that unsecured depositors, bondholders, and even shareholders might be bailed-out, they would have a reduced incentive to monitor the bank's activities closely and ask for a higher premium if the bank is taking too much risk.  Risk-shifting may occur if deposit insurance is not fairly priced (Merton, 1977), or if governments provide guarantees to holders of bank debt (Flannery and Sorescu, 1996). This increased moral hazard arises with any safety net or insurance that is under-priced (see Freixas et al (2004)). Furthermore, an adverse feedback loop exists, as implicit subsidies lead to higher risk-taking, which in turn increases the likelihood of implicit subsidies and further increases the level of implicit subsidies. For example, Kane (2009) relates the market's appetite for structured securitisation to the implicit safety net and supervisory cover provided to the sponsors of these products. Similarly, Blundell-Wignall et al (2009) suggest that the TBTF problem is also the result of the excessive growth through derivatives and structured products trading, which present significant risks.[68] On the other hand, under the charter value hypothesis, government support decreases banks’ funding costs with depositors and creditors demanding lower rates. The decline in funding costs increases the interest margin and raises banks’ charter values, which leads to banks taking fewer risks to protect future rents (Keeley, 1990). Therefore, according to this latter hypothesis, implicit subsidies and the threat of losing future rents act as a deterrent of risk-taking.

Empirical evidence, however, provides support to the market discipline hypothesis. Marques, Correa and Sapriza (2013) find that the intensity of government support is positively related to bank risk-taking, using an international sample of bank data and government support for 2003-2004 and 2009-2010. They also show that this relationship is stronger in 2009-2010 compared to 2003-2004. Acharya, Anginer and Warburton (2013)’s findings from a large US sample also provide support for the market discipline hypothesis. They show that for institutions that achieve systemically important status, bond spreads are less sensitive to risk as the bond spread-risk relationship diminishes with TBTF status. They also find that larger financial institutions follow risker strategies than smaller ones. Gropp et al (2010a) find evidence that public guarantees may be associated with substantial moral hazard effects. They use the removal of government guarantees by court decision in 2001 for savings banks in Germany as a natural experiment on the effect of public guarantees. They show that when these guarantees were removed, those banks’ credit risk reduced, as they cut the riskiest borrowers from credit (with the Z-score of average borrowers increasing by 7.5%). Banks also increased interest rates on remaining borrowers (by around 46 basis points), and banks' bond yield spreads increased by 5 basis points. They conclude that "a credible removal of guarantees will be essential in reducing the risk of potential future financial instability."[69] Dam and Koetter (2012) also explore German data of actual bail-outs and official records of distress during 1995-2006 and show that significant increases in expectations of bail-outs for banks (measured through political factors and historical bailout probabilities) lead to significant increases in risk-taking.

Furthermore, Gropp et al (2010b) argue that there is a third effect (in addition to the market discipline and charter value), as government guarantees may also affect the risk-taking of, not only, the protected banks but also, through competition, of the protected banks' competitors. This indirect effect arises as government subsidies reduce the profit margins of competitor banks, due to fiercer competition from banks that are able to refinance at subsidised levels. This leads to greater risk-taking of the competitor banks. Gropp et al (2010b) find evidence from a sample of banks from OECD countries that the main channel of increased risk-taking is through increasingly risky competitive conduct of other banks. They conclude that bail-out policies increase market expectations of bail-outs in the future, which may distort competition and increase risk-taking of all banks, and lead to greater financial instability in the future.

There is some empirical evidence that structural reform can address excessive risk-taking. For example, Marques, Correa and Sapriza (2013) find that capital supervision and regulation were not enough to fully prevent additional risk-taking by banks with more government support, but banks that faced more restrictions in terms of activities they were allowed to perform were less likely to take on more risk (see also Blundell-Wignall et al. (2009)). Besides, in order to curtail the excessive risk-taking and expansion of banks resulting from the existence of public safety nets in the US, when the first safety nets were introduced, they were accompanied by a series of regulations that (i) prohibit deposit-taking banks to underwrite or deal in securities, (ii) limit commercial banks'  access to deposit insurance, as well as (iii) set a saving deposit rate ceiling to avoid destabilising competition among banks.

3. Competitive advantage for banks that benefit from the implicit subsidy.

There is a competitive advantage for banks that benefit from the implicit subsidy in terms of lower funding costs. Banks, however, are likely to benefit from such an implicit subsidy to differing extents. Even if implicit subsidies are hard to estimate with great precision, there are several findings that are recurring in the literature. According to several studies, and as also confirmed by the JRC work, implicit subsidies benefit the largest banks disproportionately (see Rime (2005), Noss and Sowerbutts (2010), Schich and Lindh (2012) and Oxera (2011)) and therefore larger banks are more likely to benefit from such an implicit subsidy. This would entrench the too-big-to-fail banks, and induce a competitive barrier for smaller banks. Another common finding in the literature, also confirmed by JRC work, is that the implicit subsidy advantage is higher the greater the creditworthiness of the sovereign in which the bank is headquartered. Therefore, a level playing field across EU Member States is not guaranteed, as banks in Member States with higher sovereign rating benefit disproportionately more from such implicit subsidies. This distortion is likely to have become more acute as the debt crisis in several Member States reflects increasing disparities in the public finances across Member States.

4. Allocative inefficiency

Another source of distortion is an allocative (and productive and dynamic) inefficiency. The implicit subsidy makes the financial sector artificially more lucrative and this lead to increases in the size of the financial sector diverting resources (including human capital) from other sectors of the economy. Therefore, from a static perspective, banks that benefit from implicit subsidy due to lower cost of capital (funding costs) will be in a position to over-expand. Furthermore, if banks are more likely to be bailed-out if they are larger and more interconnected, as also confirmed by empirical analysis, then banks would have an incentive to increase both their scale and their exposures to other large banks. As a result, an institution to obtain or maintain its TBTF status would have an incentive to have an aggressive program of mergers and acquisitions and expand their activities' portfolio as much as possible, making the institutions even more gigantic, ever more complex (and thus harder to resolve) and more politically influential. In addition to this TBTF bias, Kane (2009) suggests that implicit subsidies may further create geographic distortions in the financial market as financial firms have incentives to book risky positions in jurisdictions where supervisory loopholes would allow maximum benefits of subsidies.

Furthermore, there are productive and dynamic inefficiencies stemming from implicit subsidies. Stern and Feldman (2004) claim that the possibility of a bank bailout makes it more likely that banks will not operate in a cost efficient way and may also innovate less.

Also if public interventions take a specific form, such as in the form of bailing out debtors, they also create a distortion in the liability/equity structure of banks leading to sub-optimal levels of leverage (see Admati and Hellwig (2013)).

2.4. Why do governments still intervene if public intervention leads to high TBTF costs?

As explained above, implicit public interventions have very significant side effects. Despite these adverse effects, policy makers often bail-out failed banks. Stern and Feldman (2004) explain that the TBTF problem stem from a lack of credibility of policymakers' commitment to not bail-out large banks. This lack of credibility is another manifestation of the time inconsistency problem as discussed by Kydland and Prescott (1977). Even if policymakers recognise that bail-out policies lead to long run moral hazard and distortions, they face a commitment problem with respect to their pledge that they will not intervene in cases of bank crises.  When a large bank is at the brink of failure, policymakers will want to renege on their pledge in order to avoid systemic risk. In turn unsecured creditors anticipating such policymakers' incentives will not monitor large banks sufficiently leading to such TBTF problem.

Besides, governments often like to leave some discretion on their hands on their decision to intervene or not in order to cater for unforeseen situations. Governments have often deliberately left an ambiguity on whether they will intervene and they do not announce their willingness to support institutions they consider TBTF. Even if this provides some short term comfort to policymakers it also accentuates the time inconsistency problem they face.

Furthermore, several academics claim that the incentives to intervene (and the associated implicit subsidy) are likely to persist even if a clear resolution mechanism is in place (see Admati and Hellwig (2013)). They argue that no bail-out commitments are not credible as banks’ interconnectedness remains high and the (social) costs of a systemic crisis are also too high. Also credit rating agencies such as Moody’s consider that when they assess the likelihood of public intervention they must strike a balance between "clear policy intent to impose losses on credits […] and on the other hand, despite stated policy preferences, there are still cases where governments will feel bound, or trapped, by the complexity and interconnectedness of banks (and the painful economic repercussions that their failure will entail) to continue to extend systemic support to troubled institutions".[70]

As other complementary policy initiatives, structural reform is a way to (at least to some extent) address this time inconsistency problem. By separating trading and deposit taking activities it is made more credible that governments would not renege on their no bailout pledges for the trading entities. One would expect a lower probability of intervention for trading entities (for example due to the fact they will have access to deposits). Therefore, with less of an incentive for policymakers to renege on no bailout pledges then the TBTF problem would be (at least to some extent) reduced.

The complementarity of structural reform with other reform proposals, and in particular with resolution, is discussed in detail in Annex A3.

3. Valuing the implicit guarantee

There has been a lot of research on the quantification of the implicit government guarantee to the banking sector. Measuring the level of implicit subsides is important to provide insight on the potential magnitude of their adverse impact (both for the taxpayers and in terms of distortions in the market place). The implicit subsidy is not transparent and therefore not observable and as a result it is hard to reach a consensus on its size. The recurrent finding in the literature is that implicit subsidies exist and are sizeable; with subsidies reaching levels of several billion euros annually for large banking groups. However, the precise estimate of the level of the implicit subsidies is highly dependent on the exact methodology used as well as on the sample period/countries under consideration, with estimates differing to a significant extent. There have been several strands of literature on the valuation of the implicit subsidy, including the funding advantage models, the contingent claims model, event study methodology, and measuring market distortions. Also, different studies on implicit subsidies focus either on bondholders, shareholders or depositors. Kane (2000) suggests that the first place to look for implicit subsidies is in markets for uninsured bank and bank-holding debt and indeed most literature focuses on the advantages of banks on their funding through debt. There is however a number of studies on equity benefits (typically through event studies) and some work on uninsured depositors (Jacewitz and Pogach (2013)).

In this section we first summarise the findings of the literature in the field of quantifying the implicit subsides and identifying its determinants and discuss the methodologies used. We also discuss two studies of two large US banks. Before concluding we discuss the work undertaken by JRC. Appendix 1 summarises the findings of the literature and the methodology used in the literature.

1. Funding advantage models

In this class of papers the subsidy is valued as the aggregate reduction in the cost of bank funding following an implicit government guarantee. Implicitly there is a comparison with a higher counterfactual cost that the bank would face in the absence of this implicit government support. Then multiplying the difference between the actual and counterfactual cost of funding by the size of each bank's risk-sensitive liabilities gives an estimate of the implicit subsidy. However, this approach assumes that banks' liability structures are independent of the existence of the implicit government guarantee but in practice when government support is likely withdrawn banks might seek to reduce their more expensive liabilities/shrink. There are two main approaches within funding advantage models the size based approach and the rating based approach.

a. Size based approach

Size based models test the hypothesis that only large (TBTF) banks would be supported by governments and consequently enjoy a reduced cost of funding compared with their smaller counterparts. Consequently, these models assume that in the absence of government support large banks would face the same cost of funding as their smaller peers. A number of papers that follow this size based approach they use events to identify the implicit subsidies and examine how large (TBTF) banks funding costs diverge from the funding costs of small banks in the bond market.

A simple example of this approach is Baker and MacArthur (2009) that use US FDIC data for depository institutions and assume that all banks with assets in excess of $100 billion (18 BHCs) will receive government support in the event of their failure. Using bailout events they find that larger banks have 49bps lower funding costs post such bailout events, which translates to a government subsidy of USD 34.1 billion a year for these 18 banks. Acharya, Anginer and Warburton (2013) find evidence that expectations of state support are embedded in credit spreads on bonds issued by more than 500 US financial institutions. They analyse the determinants of the bond spread by regressing them on a number of institution (including size and risk), bond, and macro characteristics. They consider that an institution is TBTF if it is in the top 90th percentile in terms of size. They find that the implicit subsidy provided large banks with an annual funding cost advantage of approximately 28 bps before the financial crisis increasing to 120 bps during the crisis. The total value of subsidy amounted to USD 20billion per year before crisis reaching USD 200billion during the crisis. They also examine events related to the recent financial crisis and study their impact in the bond market. Following the governments' rescue of Bear Stearns larger financial institutions experienced greater reductions in bond spreads than smaller institutions and following Lehman Brothers collapse larger financial institutions experienced greater increases in their spreads compared to smaller institutions. They also show that the passage of Dodd-Frank in the summer of 2010 did not eliminate investors' expectations of government support.

Similar to debt holders, uninsured depositors face a potential loss in the event of bank failure, and therefore should require less compensation for risk from a bank they feel will likely receive a bailout. Jacewitz and Pogach (2013) consider that since uninsured depositors face potential losses in the event of closure the market will ask a smaller risk premium of banks for which it considers a government bailout to be more likely. Their results confirm that largest banks receive a deposit premium.  They estimate differences in the cost of funding using the actual deposit rates offered at the branch level between 2005 and 2008. They focus on the premium paid on USD 100 000 money market deposit accounts and use the USD 25 000 money market deposit accounts as a threshold to account for non-risk factors across banks (such as having a larger branch network, a broader range of services etc.). The advantage of using such a difference in difference approach is that they can eliminate many non-risk based bank characteristics as well as many standard risk metrics. They find that largest banks paid 15-40 bps lower than other banks for comparable deposits (deposits with at least USD 100 000). They estimate that if this advantage is extended to all consolidated uninsured funds of banks it would represent 70% of their pre-tax profits. They also show that this difference in risk premiums cannot be attributed to usual balance sheet measures of risk. They conclude that since large banks' risks are differentially priced, the competitive environment will tend to favour them.

The main advantage of the size based approach is that (at least in its simpler versions) it is relatively simple to estimate. However, the main drawback is that the size based approach makes the crude assumption that only large banks will receive government support. Furthermore the related question is at what level should this threshold apply, and why. A commonly used TBTF threshold in the literature is $100billion in assets. But even ex post the history of the applications of bailouts makes a precise definition of a TBTF threshold impossible.[71] These models also typically do not take into consideration that size alone may not be the only factor that determines whether a bank will get bailed out or not and other factors such as interconnectedness are important.

b. Rating based approach

Some credit rating agencies (including Fitch, Moody's and Standard and Poor's) often issue two credit ratings for a bank in their assessment of the probability of default: a stand-alone rating and a higher support (all-in) rating. The all-in rating factors in the agencies’ estimate of the external support that the bank under consideration would receive from public authorities and parent companies while the former only considers the intrinsic strength of the bank. More recently credit rating agencies also publish adjusted stand-alone ratings that take into account also the likely parental support (adjusted stand-alone rating). As a result, the difference between the adjusted stand-alone and all-in ratings would reflect a decrease in the cost of servicing one's debt and is functionally equivalent to an implicit guarantee for the debt. This difference is often called "uplift". Several studies exploit the rating approach to determine the value of the implicit subsidy and/or to assess its determinants.

Studies find consistently several notches of uplift which translate into significant levels of implicit subsidy. The size of the implicit subsidy depends on the sample used as well as the definition used for rating sensitive liabilities, as only for these liabilities the funding advantage would apply. Haldane (2012) looks at 29 institutions deemed by FSB to be world's most systemically important. In pre-crisis the difference between the stand alone and support ratings averaged 1.3 notches using Moody's data (note that he uses stand-alone and not adjusted stand-alone ratings). He argues that even small notches can translate into big implicit subsidies if balance sheets are large. He estimates the implicit subsidies to be on average around USD 70 billion per year for 2002-2007 (which corresponds to roughly 50% of the average post-tax profits of these banks). By 2009, the ratings uplift reached 3 notches and the average implied subsidy exploded to over USD 700 billion per year. Schich and Lindh (2012) also estimate the rating uplift using Moody's data focusing on a sample of 118 large European banks between 2007 until March 2012. They also find that the implicit guarantee increased in the wake of the global financial crisis (uplift of around 2.2 points in 2007 and 2012 which peaked in 2009 to 3.14 points). In March 2012 the estimated yearly reduction in funding costs were around USD 35 billion in Germany, around USD 15 billion in France and USD 10 billion in the UK. Bijlsma and Mocking (2013) using data from Moody's on 151 relatively large European banks show that the size of the implicit subsidy reached almost EUR 150 billion for the sample of these banks in mid-2011 to fall to around EUR 100 billion at July 2012. Ueda and di Mauro (2012) use Fitch data on around 900 banks from 16 OECD countries and find a funding advantage of around 60bp in 2007 and 80bp in 2009.

Several papers also attempt to identify the source of determinants of credit rating uplifts by regressing the uplifts on several bank and other relevant characteristics (such as variables relating to the rating of sovereigns in which banks are headquartered). Schich and Lindh (2012) use cross section analysis for March 2012 with country fixed effects to control for unobserved heterogeneity across countries. Their results show that larger firms receive (on average) larger such subsidy implicit guarantees and implicit subsidies are higher the larger the bank relative to its peers in the same country. Estrella and Schich (2012) follow a similar approach but focus on the effect of government rating on the uplift. They find that the higher the sovereign rating the higher the uplift. A 2 notch increase/decrease in sovereign/stand-alone rating leads to 1 notch uplift. Rime (2005) finds that bank size has a positive and significant impact on all-in rating and such impact is larger for banks with low individual ratings.

JRC has also undertaken an empirical work on implicit subsidies based on the ratings approach which is discussed below. The full analysis is provided in Annex A4.2.

An advantage of the rating approach is that it controls for the relative risk of different banks' business models (as incorporated in the ratings). Also it allows taking into account the likelihood of receiving government support as agencies typically distinguish between stand alone and support ratings. However, the ratings based approach has been criticised due to reliance on subjective rating agency judgment and to be informative it requires that credit rating agencies have good insight of banks' risks and the probability of government support. For example, Goldman Sacks (2013) considers that the link between credit ratings and funding costs may not be as clear as ratings take a different time horizon giving more long term view of credit fundamentals and typically lag market assessments. However, even if subjective it would be hard to argue that credit ratings are not informative. Credit rating agencies are important market participants and markets use credit ratings in pricing debt instruments. Therefore while subjective their views are likely to reflect market sentiment and views on the likelihood of government support. Currently all three major credit rating agencies provide proxies for government support.[72] Furthermore, there is some evidence that ratings of credit agencies have predictive power. Marques, Correa and Sapriza (2013) also investigate whether the Moody's support ratings are able to predict actual bail-outs. They find that banks that enjoyed a nonnegative support rating were more likely to be rescued in 2008-2010 by 30 percentage points.

2. Contingent claims ("CC") models

Contingent claims models determine the implicit subsidy in an option pricing framework. The value of the subsidy is seen as the expected annual payment from the government to the banking system necessary to prevent default. The subsidy is modelled as the shortfall between the value of banks' assets and some threshold based on their minimum capital requirements at some future time (for example Oxera (2011) uses a one year horizon). Failure is assumed to arise when total assets of all banks falls below this minimum requirement. The value of government support is assumed to be the sum necessary to restore the value of assets to this minimum amount, weighted by the probability of their falling below that level.

CC models require the modelling of the dynamics of banks' future asset values and their statistical distribution. This is similar to pricing an option with the banking system viewed as a residual claimant on the government. The value of the implicit subsidy is similar to a put option. If banks' assets are greater than the threshold minimum asset value when the option expires then it would not be exercised. If the value of the assets is below the trigger point its payoff is equal to the difference between the two. There are two possible methods to model the dynamics and distribution of future assets: the equity option price approach in which the distribution of bank's equity values is estimated based on the prices of equity options and the historical approach which bases its estimates on historical prices of bank equity).

Oxera (2011) uses an equity option price approach. Equity options pay out if banks' equity experiences price changes of different magnitude and they give an insight into investors' expectations of relative likelihood of changes of different magnitudes in banks' equity prices. The central base case estimate of the forward looking state support corresponds to overall GBP5.9 billion annual transfers for 2010 in the UK for the five largest banks. However, as Noss and Sowerbutts (2012) notes, the conclusions are sensitive to a number of assumptions. For example, the timing of state intervention; the implicit subsidy is modelled as the expected shortfall at a horizon of one year (European option i.e. at year-end). But state support could be extended at any time were the value of banks assets to fall below the threshold. Noss and Sowerbutts (2012) suggest that it would be more realistic to consider the subsidy as a look back option whose value is determined by maximum shortfall at any time over a year's horizon (subsidy would rise to GBP30billion). Other critical assumptions include the level of the discount factor, flatter tail distribution of asset returns. Overall, the subsidy can reach up to GBP120 billion under some assumptions. Under a historical approach the estimated subsidy is lower at around GBP 30 billion Noss and Sowerbutts (2012). Therefore estimates for the same period and banks range from GBP5.9 billion to GBP120billion depending on the underlying assumptions.

Noss and Sowerbutts (2012) note further drawbacks of these models. They note that CC models require modelling of the future path of banks' assets and make simplifying assumptions (for example that banks fail when assets fall to a value commensurate with banks' minimum capital ratio). Also they consider that banks' equity prices may be distorted by investor's expectation of support, increasing their value and thereby reducing the size of the estimated subsidy.

On the other hand, an advantage of the CC models is the use of market price data. Oxera (2011) for example claims that market prices should reflect aggregate expectations of actual investors in the market and are available almost continuously. In this respect, CC models are able to estimate directly the expected value transfer from government based on likelihood of a systemic shock occurring and expected payment from state to avoid systemic failure. On the contrary, funding cost advantage models look at difference with and without support, albeit often it is not obvious how to establish the relevant counterfactual.

3. Event studies

Another stream of literature looks at how specific events such as policy announcements related to TBTF policy or policy interventions, as well as bank mergers affect market expectations on the implicit subsidies by studying banks' equity prices. These of papers on event studies have looked at equity prices of banks rather than debt prices. Equity holders may benefit from implicit subsidies either due to an expectation of bail-out as well as from an indirect effect as implicit subsidies will impact a bank's stock price by reducing a bank's cost of funding thereby increasing its profitability.

A number of papers explore the impact of events that affect market's expectations concerning TBTF policy. For example, O'Hara and Shaw (1990) conduct an event study to estimate the value of TBTF subsidy using the 1984 congressional testimony from the US Comptroller of the Currency in which he indicated that the 11 largest US banks were subject to a TBTF policy (whereby for these banks total deposit insurance would be provided). They find that there was a significant positive average residual return of 1.3% in these banks' equity value on the day of the announcement. Pop and Pop (2009) find evidence that the announcement that shareholders would not incur losses in the 2003 Resona Holdings bailout in Japan resulted in significant abnormal returns on the larger Japanese banks.

Other studies instead examine the impact of mergers on public expectations of TBTF. The underlying idea is that when two banks merge and as a result they become TBTF (or increase their likelihood of being TBTF) they would earn a premium compared to other mergers. In particular, any merger that strengthens market presumptions that a bank acquirer is TBTF would lower that entity’s financing cost and this may be reflected as a benefit for shareholders.  Kane (2000) studies the mergers of the top 12 US banks between 1991 and 1998 and finds that they gave greater value to shareholders when the target was a deposit-taking institution which was large. He considers that implicit subsidies have inflated incentives for giant banks to merge with other banks and this way shift risk into taxpayers.  Brewer and Jagtiani (2011) use data from 1991-2004 to look for the premium paid in 8 US mergers that bought organisation over USD 100billion in assets in the US (their defined threshold for TBTF), in order to test the hypothesis that banks are willing to pay more to become TBTF. They find that acquirers paid a total premium of around USD 15 billion. Molyneux et al (2010)' results are similar for nine EU countries with EU banks paying higher merger premiums if their targets are larger. Penas and Unal (2004) study the impact of 66 US mergers.  They find that mergers that push the combined bank's asset size above the TBTF threshold asset size (2% of assets of all depository institutions) realise the highest returns whereas returns of megamergers (mergers among banks that are already TBTF) or smaller bank mergers (that do not bring the combined entity above the TBTF threshold) earn relatively lower adjusted returns.  This can be explained as megabanks are already TBTF while small banks do not become TBTF as a result of the merger. These "medium mergers", which are most beneficial for shareholders, are found to benefit from a funding cost advantage as they lead to a 15 basis point decrease in credit spreads of new debt issues of the acquiring bank post-merger.[73] On the other hand, a study that does not find evidence of TBTF subsidy is Benston, Hunter and Wall (1995) which analyse the prices that acquirers bid in an earlier US sample, 1980-1989. They conclude that most of mergers in the 1980s were motivated by earrings diversification rather than TBTF considerations.

A caveat of event studies based on share prices is to ensure that the events under consideration are truly exogenous and therefore are not reflected in these prices pre-announcement (of the intervention or the merger). To the extent that the studied event is not exogenous the analysis would not be in a position to identify the claimed effect. An additional caveat of these studies is that they can only identify the market price changes and not the level of such implicit subsidies as the banks in question may have been benefiting from the implicit subsidy even before the examined event. Thereby these studies are likely to provide a lower bound for the size of the implicit subsidies. On the other hand, some studies do not adequately control for the efficiency effects of mergers and therefore any identified impact may confound both market power/efficiency and TBTF considerations (as in Kane (2000)) thereby possibly overestimating the TBTF effect. Similar to the size based funding cost models, a problem inherent in several of these studies is the need to make an assumption on the TBTF threshold.

4. Distortions of market prices – CDS prices

Another approach to measure the size of the implicit subsidy is to analyse the distortions of market prices caused by the TBTF policy as reflected in the CDS market. The underlying idea is that the implicit subsidy would lower the CDS spreads for these banks as the default probability would be lower.

Völz and Wedow (2011) analyse an international sample of banks and find that CDS prices tend to be distorted if the banks are larger and therefore less market discipline is exercised on these larger banks. A one percentage increase in mean size of a bank relative to the country's GDP reduces the CDS spread by around 2 basis points. They also find that this relationship has decreasing returns as banks would presumably become too big to save (with the turning point located at around 2 times the home country's GDP). Schweikhard and Tsesmelidakis (2012) consider that there is an asymmetric treatment of debt and equity during bailouts which tend to favour creditors, as typically equity holders are the first to be hit. In order to exploit this asymmetry, they compare credit default swap (CDS) premiums to their theoretical equity counterparts for around 500 US companies between 2003-2009. They find that while CDS and stock-market-implied CDS spreads are closely aligned during the pre-crises period, during the 2007-2009 crisis the stock-market-implied CDS prices for banks are significantly higher than market CDS prices (that is there is higher stock-implied default risk compared to CDS observations). This difference is also shown to be positively related to the size of the banks and by revaluing the offering price of bonds in the US they estimate the total subsidies to amount to around USD 130 billion over the period 2008-2009.

Industry studies

There are some studies from the industry that claim the TBTF banks do not actually enjoy a funding advantage due to the implicit subsidy.

Goldman Sachs (2013) for example argues that while a small funding advantage was present during the financial crisis it has totally disappeared post crisis.  The funding advantage for US has been modest, around 30 bps on average since 1999.[74] They claim that large firms in other industries also benefit from a comparable or even larger funding advantage, which suggests that size confers benefits that appear distinct from the issue of government support. In addition, they find that bonds of the largest banks provide investors with far greater liquidity and this added liquidity which, coupled with the superior loss history for large banks, can explain the funding advantage that the largest banks have experienced. In other words, investors are willing to pay for the benefits of liquidity and large banks have more liquid bonds. They make a second point that the largest banks generate fewer realised losses than do the failures of smaller banks per dollar of deposits. They provide the example of the Troubled Asset Relief Program (TARP) in which the government made a 15% profit on the assistance it provided to the six largest banks.

However, their paper sets aside the issue that TBTF subsidy is about the expectation of government support lowering the cost of funds relative to what they would have been otherwise. Carney (2013) argues that TBTF banks take on far more risk (see Schnabel 2004, 2009) and more debt than smaller banks. He points out this study of Goldman Sachs does not control for risk and leverage metrics and therefore the fact that that risky debt-laden large banks pay only 10 basis point more on their debt would be an evidence of implicit subsidies. Related to the returns to taxpayers, one would need to look into more aggregate data and not only one program. Laeven and Valencia (2012) find that recent financial crisis 2007-2009 has increased the burden of public debt and size of government contingent liabilities, creating concerns about fiscal sustainability in some countries. They estimate that there are direct fiscal costs reaching 6% of GDP in advanced economies for measures such as capitalisation of banks, provision of liquidity and guarantees and asset purchases. Furthermore, Morrison (2011) claims that one should also take into account that funds are diverted from alternative uses, potentially at a significant opportunity cost and therefore the profit net of opportunity costs will surely be much lower.  Also, irrespective of whether to government makes a loss or a benefit this should not have an impact on creditors and therefore the distortions of the implicit subsidy. It would imply that bailouts do not really create a subsidy for large banks as they instead make money for taxpayers.

Araten and Turner (2012) (JP MorganChase) also provide an estimate of the funding cost advantage that large banks enjoy and end up with lower estimate of implicit guarantees compared to other studies. They look in globally Systemically Important Banks (G-SIBs defined as institutions with more than USD 500 billion in assets) in the US and find moderate cost advantages associated with the GSIB status of 9 bps taking into account all funding cost sources (and if one considers only interest bearing funding sources this increases to 18bps). The main funding advantage is with regard to domestic deposits (23bps) but there are also smaller cost advantages with respect to credit spreads on senior, unsecured debt (and higher costs for Fed Funds). They explain that this relatively small difference in funding costs is due to the difference in funding mix between GSIB and non-GSIBs as non-GSIBs rely more heavily on deposits (that have relatively lower funding costs than other categories).

This analysis shows that there are composition effects that however are not accounted for in their analysis. The fact that larger banks are able to expand their balance sheet with lower share of deposits makes comparisons of individual categories of funding less informative if one does not control for the underlying structure of liabilities of different banks. Also risk factors are not accounted for. Furthermore, Noss and Sowerbutts (2012) suggest that this paper assumes that banks' funding advantage due to TBTF status to be invariant over time and that a number of endogenous variables are included (e.g. profitability which is itself affected by spread).

JRC work

As highlighted above, the JRC has also worked on the quantification of implicit subsidies. This analysis follows the rating based methodology of funding cost models and in addition to quantifying the implicit subsidy it also investigates the determinants of the government support on a sample of 112 large EU banks representing around 60-70% of EU banking total assets. Based on Moody's data on all-in ratings and stand-alone data JRC first computes the uplift across these banks that is due to government support and finds that it lies between 2 and 3 notches. Also, JRC has constructed a yield curve to translate the uplift into an estimate of the total funding cost advantage in euros. The estimate of the implicit subsidy for their sample is UR 72-95 billion and EUR 59-82 billion in 2011 and 2012, respectively, for the banks considered. In relative terms, the implicit subsidy estimate is in the range of 0.5% to 0.8% of EU-27 GDP. However there is some variation across countries as the estimates are largely dependent on the size of the banking sector in each country and the sovereign credit rating, as well as the sample construction. The importance of the subsidies is also illustrated by the fact that it account between one-third and one-half of the aggregate annual pre-impairment operating profit of the banks.

The econometric analysis confirmed the findings of the literature showing that the uplift is associated with the size of the bank, the banks' stand-alone strength and the sovereign (in which the bank is headquartered) ratings. Larger banks are more likely to receive a relatively higher uplift. Banks are likely to be TBTF due to their size, their interconnectedness or importance to the financial systemic. As size is likely to be highly correlated to TBTF status, this indicates that larger banks are likely to benefit more from government support. A common finding in the literature is that the rating of the country in which banks are headquartered is important. JRC's work confirms that banks that are headquartered in a country with a strong sovereign credit rating are more likely to benefit from a higher uplift for a bank, raising questions on the level playing field in the internal market. Furthermore, a higher stand-alone rating increases the likelihood of observing lower uplift. Components of the individual strength rating of a company relate to the riskiness of the banks (see Moody's methodology (2011)). Therefore, a bank that is considered more risky is likely to receive a higher uplift. The results also show that banks with stronger parents are more likely to receive support within the group rather than from the government.

JRC also extended the analysis by including several balance sheet indicators to investigate whether the uplift is likely to be affected by bank business model. Before commenting on the results of this extended model, the Commission notes that it is demanding to require from a reduced form model to provide unequivocal evidence on the importance of business model. Overall, however, the Commission considers that the regression analysis provides some evidence that the business model variables affect the level of the uplift. Two variables that are significant are the variables that relate to the bank’s level of interconnectedness. Banks that hold a larger proportion of net loans to banks in their total assets and banks that rely more on wholesale market for funding are more likely to benefit from a higher implicit guarantee. Also, banks that are better capitalised benefit from lower uplift.

The full report is provided in Annex A4.2.

4. Conclusion

There are several explicit or implicit safety nets for banks to deal as even solvent banks are susceptible to bank runs and given the importance of the banking sector for the real economy. Governments often step in and bail-out banks as evidenced during the financial crisis. To the extent that these guarantees are mispriced they lead to several distortions. Anticipating public support for the banks, their creditors have lower incentive to monitor banks' activities. As a result moral hazard arises and banks' managers are willing to take higher risks. Implicit subsidies for banks also imply artificially higher returns for the sector attracting resources from other sectors of the economy. Furthermore, as certain types of banks are more likely to benefit from the implicit subsidy this also distorts the level playing field across banks. For example, large banks, which are more likely to be TBTF, benefit disproportionally from implicit subsidies compared to their smaller counterparts. Also, as the value of the implicit subsidy depends on the credibility of the guarantor, banks headquartered in countries with good public finances are more likely to benefit from such a subsidy.

There are several studies on the quantification of implicit subsidies. The results of these studies are highly dependent on the methodology employed as well as the sample (banks, geographic scope, time period) under consideration. Overall, however, they point out that implicit subsidies are present, sizeable and can represent a significant advantage for larger banks. Debt holders, depositors and shareholders are likely to benefit from implicit subsidies. However, studies typically focus on one type of stakeholder even though the overall benefits to TBTF banks might be expected to accrue to several parties. Implicit subsidies are found to represent a significant share of countries' GDP (typically more than 0.5%) and of banks' profits (more than 30% in some studies). JRC findings confirm previous findings of the literature: larger banks are likely to benefit relatively more from implicit subsidies. Also, banks headquartered in a country with higher rating are benefiting from a higher uplift. When investigating the role of business model variables, it is found that the degree of banks’ inter-connectedness also affects the uplift, with more interconnected banks enjoying a higher uplift.

                                                                                                    

5. Bibliography

Acharya, V., I. Drechsler and P. Schnabl (2010), “A Pyrrhic Victory? Bank Bailouts and Sovereign Credit Risk”, Working paper, NYU-Stern, CEPR Discussion Paper 8679.

Acharya, V., D. Anginer, and J. Warburton (2013), “The End of Market Discipline? Investor Expectations of Implicit State Guarantees”, working paper, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1961656

Admati, A. and M. Hellwig (2013), “The Bankers’ New Clothes: What’s Wrong with Banking and What to Do about It”, Princeton University Press

Araten, M., and C. Turner (2012), ‘Understanding the funding cost differences between globally systemically important banks (G-SIBs) and non G-SIBs in the United States’, mimeo, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2226939.

Baker, D.  and T. McArthur (2009), ‘The value of the ‘too big to fail’ big bank subsidy,’ CEPR Reports and Issue Briefs 2009-36, Center for Economic and Policy Research.

Benston, G., W. Hunter, and L. Wall, (1995), “Motivations for Bank mergers and acquisitions: enhancing the deposit insurance put option versus earnings diversification.” Journal of Money, Credit, and Banking, Vol. 27, No. 3, pp. 777-788.

Bijlsma M. and R. Mocking (2013), “The private value of too-big-to-fail guarantees”, CPB Netherlands Bureau for Economic Policy Analysis, CPB Discussion Paper 240.

Blundell-Wignall, A., G. Wihinger, and P. Slovik (2009), “The Elephant in the Room: The Need to Deal with What Banks Do”, OECD Financial Market Trends Vol. 2009/2.

Brewer, E. and J. Jagtiani (2011), "How much did banks pay to become Too-Big-To-Fail and to become systemically Important?", Federal Reserve Bank of Philadelphia, working paper No. 11-37.

Dam, L. and M. Koetter, (2012), "Bank Bailouts and Moral Hazard: Evidence from Germany," Review of Financial Studies, Society for Financial Studies, vol. 25(8), pages 2343-2380.

Diamond, D. (1984), “Financial intermediation and delegated monitoring”, Review of Economic Studies, 51, pp. 393-414.

Diamond, D. and P. Dybvig (1983), “Bank runs, deposit insurance and liquidity”, Journal of Political Economy, 91, pp.401-19.

Estrella, A. and S. Schich (2011), “Sovereign and banking sector debt: Interconnections through implicit guarantees”, OECD Financial Market Trends Vol. 2011/2.

Flannery, M. and S. Sorescu, (1996). “Evidence of Bank Market Discipline in Subordinated Debenture Yields: 1983-1991,” Journal of Finance, American Finance Association, vol. 51(4), pages 1347-77, September.

Freixas X., J-C. Rochet, and B. Parigi (2004), "The Lender of Last Resort: A Twenty-First Century Approach," Journal of the European Economic Association, MIT Press, vol. 2(6), pages 1085-1115.

Goldman Sachs (2013), “Measuring the TBTF effect on bond pricing”, Global Markets Institute.

Gropp, R., C. Gruendl, and A. Guettler  (2010a), “ The Impact of Public Guarantees on Bank Risk Taking: Evidence from a Natural Experiment”, European Central Bank, Working Paper Series,, No 1272.

Gropp R., H. Hakenes, and I. Schnabel (2010b) “Competition, Risk-Shifting, and Public Bail-out Policies, Max Planck Institute for Research on Collective Goods, Bonn, 2010/05.

Haldane, A. (2012), “On being the right size”, The 2012 Beesley Lectures, at the Institute of Directors, London.

Jacewitz, S and J. Pogach (2013), “Deposit Rate Advantages at the Largest Banks”, working paper, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2018474

Kane, E. (2000), “Incentives for Banking Megamergers: What Motives Might Regulators Infer From Event-Study Evidence?,” Journal of Money, Credit and Banking, 32(3), 671–701.

Kane, E. (2009), "Extracting Non-transparent Safety Net Subsidies by Strategically Expanding and Contracting a Financial Institution’s Accounting Balance Sheet," Journal of Financial Services Research, Springer, vol. 36(2), pages 161-168, December.

Keeley, M. (1990): “Deposit Insurance, Risk, and Market Power in Banking,” American Economic Review, 80, 1183 – 1200.

Kydland, F. and E. Prescott, (1977), "Rules Rather Than Discretion: The Inconsistency of Optimal Plans," Journal of Political Economy, University of Chicago Press, vol. 85(3), pages 473-91, June.

Li, Z., S. Qu, and J. Zhang (2011), ‘Quantifying the value of implicit government guarantees for large financial institutions’, Moody’s Analytics Quantitative Research group, January.

López-Espinosa, G., A. Moreno, A. Rubia, L. Valderrana, and R. Calderón (2013), “Drivers of Fiscal Outlays in Banks Recapitalizations”, Working paper, http://gbif.ch/files/content/sites/irene/files/shared/documents/s%C3%A9minaires/Lopez.pdf

Marques, L., R. Correa, and H. Sapriza (2013), “International Evidence on Government Support and Risk Taking in the Banking Sector”, IMF Working Paper, 13/94.

Merton, R. (1977), "An analytic derivation of the cost of deposit insurance and loan guarantees An application of modern option pricing theory," Journal of Banking & Finance, Elsevier, vol. 1(1), pages 3-11, June.

Mishkin, F. (2006) “How big a problem is Too Big to Fail? A review of Gary Stern and Ron Feldman's Too Big to Fail: The Hazards of Bank Bailouts”. Journal of Economic Literature, Vol. XLIV, pp. 988-1004.

Molyneux, P., K. Schaeck, and T. Zhou  (2010), ““Too-Big-to-Fail” and its Impact on Safety Net Subsidies and Systemic Risk”, Center for Applied Research in Finance, Bocconi University, Working paper 09/2010

Moody’s (2011), “Status Report on Systemic Support Incorporated in Moody’s Bank Debt Ratings Globally”, Moody’s Investor Service, Special Comment.

Moody’s (2012a) “Incorporation of joint-default Analysis into Moody’s Bank Ratings: Global Methodology”, Moody’s Investor Service.

Moody’s (2012b), “Moody’s Consolidated Global Bank rating Methodology”, Moody’s Investor Service.

Morgan, D., and K. Stiroh (2005), “Too Big to Fail after all these years”, Federal Reserve Bank of New York, Staff Report, No. 220.

Morrison, A. (2011), “Systemic Risks and the “too-big-to-fail” problem”, Oxford Review of Economic Policy, Volume 27/3, pp.498-516.

Noss, J. and R. Sowerbutts (2012), “The implicit subsidy of banks”, Bank of England, Financial Stability Paper No. 15.

O’Hara, M. and W. Shaw (1990), “Deposit Insurance and wealth effects: the value of being “Too Big To Fail””. Journal of Finance, Vol. 45, pp. 1587-1600.

Oxera (2011), “Assessing State Support to the UK Banking Sector‟, prepared at the request of the Royal Bank of Scotland, March.

Penas, M.F. and H. Unal (2004), “Gains in Bank Mergers: Evidence from the Bond Markets”. Journal of Financial Economics, vol. 74, pp. 149-179.

Pop, A. and D. Pop (2009), “Requiem for market discipline and the specter of TBTF in Japanese banking”. The Quarterly Review of Economics and Finance, vol. 49, pp. 1429-1459.

Rime, B. (2005), “Do “Too Big To Fail” Expectations Boost Large Banks Issuer Ratings?” Working Paper, Systemic Stability Section, Swiss National Bank.

Schich, S. and S. Lindh (2012), “Implicit Guarantees for Bank Debt: Where Do We Stand?”, OECD Journal: Financial Market Trends, vol. 2012 Issue 1.

Schweikhard, F. and Z. Tsesmelidakis (2012). “The Impact of Government Interventions on CDS and Equity Markets”. Available at SSRN: http://ssrn.com/abstract=1573377.

Sironi, A. (2003), “Testing for Market Discipline in the European Banking Industry: Evidence from Subordinated Debt Issues” Journal of Money, Credit, and Banking, Vol. 35, No. 3, pp. 443-472.

Stern, G. and R. Feldman (2004), “Too Big to Fail: The Hazards of Bank Bailouts”, Washington, D.C.,  Brooking Institution Press.

Ueda, K and B. Di Mauro (2012), ‘Quantifying structural subsidy values for systemically important financial institutions’, IMF Working Paper No. 128.

Völz, M. and M. Wedow (2009) “Does Banks‟ Size Distort Market Prices? Evidence for Too-Big-to-Fail in the CDS Market”. Discussion Paper, No. 06/2009, Deutsche Bundesbank.

APPENDIX A: Summary of literature

2013

J Cariboni, H. Joensson, L. Kazemi Veisari, D.Magos, E. Papanagiotou, C. Planas

Annex A4.2 : Size and determinants of implicit state guarantees to EU banks

Executive summary

The European Commission is bringing forward a proposal for a structural reform of the EU banking sector to tackle problems arising from banks being Too-Big-To-Fail (TBTF), including too-important-to-fail, too-complex-to-fail and too-interconnected-to-fail. It has been argued that banks that are viewed to be TBTF might enjoy an implicit government guarantee, because governments will not allow these banks to fail. This implicit guarantee thus improves these banks perceived credit worthiness and is observed as an upgrade in their long term credit ratings. As the bank’s funding cost (measured in basis points) depends on its credit rating, the implicit guarantee gives rise to a funding cost advantage: bank creditors do not demand full compensation for being exposed to banks’ risks, because bank creditors expect government bailouts to avoid troubled banks to fail. Multiplying the funding cost advantage with a share of the outstanding debt results in a monetary estimate which can be viewed as an implicit subsidy given to banks.

The present report investigates the size and determinants of the implicit state guarantee enjoyed by a sample of 112 EU banks covering 60-70% of the total bank assets in the EU over the period 2011-2013. It also estimates the implicit subsidy.

The implicit guarantee is derived from Moody’s credit ratings. Long-term credit ratings take into account the bank’s stand-alone financial strength, a rating upgrade due to possible support from a parent or a cooperative group, and a rating upgrade due to potential government support. This third component is used to measure the implicit government guarantee. The implicit guarantee is translated into a funding cost advantage by comparing the funding cost inferred from the long-term credit rating of a bank with the funding cost inferred from its stand-alone credit rating.

Results can be summarized as follows:

1) Estimation of the implicit guarantee using ratings

For about 80% of the banks the rating uplift due to potential government support corresponds to a one to three notches upgrade. For the largest 25% of the banks this upgrade is up to 26% higher than for the other banks. The average upgrade due to implicit guarantee is relatively stable over the period under study, while a decrease in the average long term credit rating can be observed.

2) Estimation of the implicit subsidy

An estimate for the total implicit subsidy is EUR 72-95 billion and EUR 59-82 billion in 2011 and 2012, respectively, for the banks considered. In relative terms, the implicit subsidy estimate is in the range of 0.5% to 0.8% of EU-27 GDP. We also find that the total implicit subsidy is between one-third and one-half of the aggregate annual pre-impairment operating profit of the banks.

3) Determinants of the implicit guarantee

In our sample of EU banks, results of an econometric analysis show that the implicit guarantee (measured through the rating upgrade) appears to be driven by:

· The bank individual strength: a higher individual strength rating increases the probability of observing lower implicit guarantee;

· The parental support: banks with stronger parents receive a lower implicit guarantee.

· The credit rating of the country where a bank is headquartered: being headquartered in a country with a strong sovereign credit rating increases the probability of observing a higher implicit guarantee. This evidence also appears if the debt to GDP ratio is used as a proxy for the sovereign credit worthiness.

· The size of the bank: larger banks are more likely to enjoy a greater implicit guarantee.

Concerning the business model we find the following evidence:

· More inter-connected banks that hold a larger proportion of net loans to banks in their total assets are more likely to benefit from a higher implicit guarantee.

· Banks that rely more on the wholesale market for funding (i.e. a high wholesale funding ratio) are also more likely to benefit from a higher implicit guarantee.

· Better capitalized banks are more likely to receive a lower implicit guarantee.

Table of Contents

Executive summary.. 81

1.Introduction.. 85

2.Literature review... 86

2.1 Credit ratings as indicators of implicit state guarantee.. 86

2.2 Quantification of the implicit subsidy.. 87

3.Estimation of the implicit guarantee 89

3.1 Data description.. 89

3.2 Descriptive statistics on the UPLIFT/UPLIFT*. 92

4.Estimation of the implicit subsidy.. 103

4.1 Methodology.. 103

4.2 The rating-yield curve map. 103

4.3 Proxies of rating sensitive liabilities. 107

4.4 Implicit subsidy estimates. 108

4.5 Comparing implicit subsidy with pre-impairment operating profit, 113

4.6 Relation between implicit subsidy and bank characteristics. 113

5.Determinants of the implicit guarantee.. 116

5.1 Dataset and explanatory variables. 116

5.2 Econometric methodology.. 122

5.3 Empirical results without business model variables. 123

5.4 Empirical results with business model variables. 128

6.Conclusions. 133

References. 135

APPENDIX A: List of Banks. 137

APPENDIX B: Statistics on Moody’s rating data.. 140

APPENDIX C: Comparison with the existing literature.. 145

APPENDIX D: UPLIFT* for selected large EU groups. 151

Annex E: Descriptive statistics of the sample used for building the rating-yield map. 152

APPENDIX F: Description of the proxies for the rating-sensitive liabilities  154

APPENDIX G: Implicit subsidy versus pre-impairment operating profit  156

APPENDIX H: Implicit subsidy versus bank characteristics. 158

APPENDIX I: SNL business model variables description.. 161

APPENDIX J: Marginal effects in Probit regression.. 163

1. Introduction[75]

The European Commission is bringing forward a proposal for a structural reform of the EU banking sector to tackle problems arising from banks being Too-Big-To-Fail (TBTF), including too-important-to-fail, too-complex-to-fail and too-interconnected-to-fail. It has been argued that banks that are viewed to be TBTF might enjoy an implicit government guarantee, because governments will not allow these banks to fail. This implicit guarantee thus improves these banks perceived credit worthiness and is observed as an upgrade in their long term credit ratings. As the bank’s funding cost (measured in basis points) depends on its credit rating, the implicit guarantee gives rise to a funding cost advantage: bank creditors do not demand full compensation for being exposed to banks’ risks, because bank creditors expect government bailouts to avoid troubled banks to fail. Multiplying the funding cost advantage with a share of the outstanding debt results in a monetary estimate which can be viewed as an implicit subsidy given to banks.

The aim of this report is two-fold. The first aim is to estimate the size of the implicit government guarantee (measured through the rating upgrade) of EU banks and to investigate its determinants via an econometric analysis. The second aim is to estimate the implicit subsidy due to this guarantee.

To measure the size of the implicit guarantee, we rely on the rating-based approach using Moody’s credit ratings. Long-term credit ratings take into account the bank’s stand-alone financial strength rating, a rating upgrade due to possible support from a parent or a cooperative group, and a rating upgrade due to potential government support. This third component is used to measure the implicit government guarantee. The implicit guarantee is translated into a funding cost advantage by comparing the funding cost inferred from the long-term credit rating of a bank with the funding cost inferred from its stand-alone credit rating.

Studies in the literature (Schich and Lindh (2012) and Schich and Kim (2012)) suggest that the implicit guarantee depends on the strength of the sovereign where the bank is headquartered and also on the size of the bank. This document extends the set of possible drivers of the implicit guarantee and investigates whether bank business profile also plays a role in the size of the implicit guarantee.

The rest of this document is structured as follows. Section 2 summarizes the available literature. Section 3 describes the Moody’s ratings used to estimate the implicit guarantee and presents some statistics. Section 4 focusses on the quantification of the implicit subsidy. Section 5 presents the econometric analysis performed to assess the determinants of the implicit guarantee and the last section concludes.

Some technical Annexes are enclosed to present additional statistics on the datasets employed and to present the Probit model used for the econometric analyses.

2. Literature review

The methodology followed in this work relies on the previous work of Schich and Lindh (2012), Ueda and Di Mauro (2012) and Schich and Kim (2012). A brief description of the methodology is provided below.

The three largest credit rating agencies provide two types of ratings for a bank:

· An "all-in credit rating" (AICR) that factors in the possibility and likelihood of external support that the bank receives when needed from its parent, a cooperative or public authorities (including government support);

· a “stand-alone credit rating” (SACR) that abstracts from such support.

The difference between these two ratings is referred to as the rating UPLIFT:

                                                            UPLIFT = AICR – SACR.                                        (1)

The UPLIFT thus factors in the effects of guarantees from the government and from other types of support, such as parental and cooperative. Since 2011, Moody’s has also started publishing an adjusted SACR (SACR*), that includes parental and cooperative support but not government support. Using SACR*, an adjusted uplift (UPLIFT*) can be calculated:

UPLIFT* = AICR – SACR*.                                                (2)

The UPLIFT* measures the increase in the rating due to government support.

2.1 Credit ratings as indicators of implicit state guarantee

Schich and Lindh (2012) employ a dataset of 118 European banks from 17 European countries for calculating the UPLIFT and the UPLIFT*. This study uses ratings from Moody’s and covers the period between end 2007 and first quarter of 2012. A general conclusion is that EU banks typically enjoy implicit guarantee and that the sample of banks enjoying an implicit state guarantee extends beyond the list of systemically important banks defined by the Financial Stability Board. They find that implicit guarantees have increased during 2009 and 2010. However, overall implicit state and parental support are persistent, as they provide evidence that there is not a major difference in the UPLIFT for the time period under study. When the UPLIFT* is examined on a country by county basis it is shown that implicit support is present for all the countries under study, although its magnitude varies across counties. Schich and Lindh (2012) also attempt to provide an insight into the determinants of the UPLIFT* by using a cross-section regression with country-fixed effects. They run a least squares regression where the adjusted stand-alone credit rating (SACR*), the domestic sovereign credit rating and the relative size of each bank are used as explanatory variables. They conclude that: (i) the lower the adjusted stand-alone credit rating (SACR*), (ii) the better the domestic sovereign rating, and (iii) the larger the bank, the higher the implicit state guarantee.[76]

In contrast to Schich and Lindh (2012), who fit a linear probability model, Ueda and Di Mauro (2012) employ ordered probit regressions with country-fixed effects. They use ratings from Fitch in order to estimate implicit state guarantee for 895 global banks in two points in time: 2007 and 2009. They also confirm that the funding cost advantage increased substantially during the crisis: from 60 basis points (bps) at the end of 2007 80 bps in end 2009.

Bijlsma and Mocking (2013) employ Moody’s ratings to calculate the UPLIFT for 151 European banks for the time period 2006 to 2011. They conclude that the average UPLIFT reached its maximum in 2010 (i.e. 3.6 notches) and then dropped to 2.5 notches in 2012. Their results show that larger banks (above some threshold value based on total assets) enjoy on average higher UPLIFT and a higher sovereign rating of a bank's home country leads on average to a higher UPLIFT for that bank.

Noss and Sowerbutts (2012) report the average UPLIFT for four UK banks from 2007 to 2010 by using Moody’s ratings. They show that the average UPLIFT for the UK banks in their sample has declined from about 3.5 notches in 2009 to one notch in 2010.

Haldane (2010) using Moody’s ratings reports that expected government support, for a sample of about  16 UK banks and 26 global banks, has increased on average 2 notches between 2007 and 2009. Moreover, his analyses also show that the UPLIFT is higher for large than for small banks in the considered sample.

In a subsequent paper, Haldane (2012) reports the evolution of the UPLIFT for the 29 world’s most systemically-important institutions calculated from Moody’s ratings. The reported time period is from 2007 to 2012. Once more, in Haldane (2012) it is confirmed that levels of support are higher after the crisis. Indeed, the average UPLIFT has increased around 2.8 notches in 2012 from about 1.5 notches in 2007.

All available studies using the credit rating methodology thus conclude that banks in the EU and worldwide benefit from a considerable implicit state support, as reflected by an increase in their credit rating.

2.2 Quantification of the implicit subsidy

On the basis of bank bond data one can construct a map between credit ratings and bond yields. To estimate the funding advantage we compare the difference between the bank's cost of funding given its all in credit rating with the funding cost based on its stand-alone rating. This latter cost of funding is assumed to be the funding cost the bank would experience if the implicit government guarantee were not present. An estimate of the implicit subsidy can be obtained by multiplying the funding advantage in basis points with the amount of rating sensitive liabilities of the bank. One must be cautious when making a comparison of results among different studies due to differences in: i) the definition of rating sensitive liabilities, ii) the sample of banks and the time period considered, iii) the sample of bonds used in the building of the map, iv) the precise methodology used for the estimation of the yield spreads.

Haldane (2010) uses a proxy of rating sensitive liabilities that excludes retail deposits but includes wholesale borrowing. The estimated subsidy of 16 UK banks is 11, 59 and 107 GBP billion in 2007, 2008 and 2009, respectively. The five largest UK banks benefit primarily from this subsidy, as the estimates for them are 9, 52, and 103 GBP billion in 2007, 2008 and 2009, respectively.

Noss and Sowerbutts (2012) define a proxy of rating sensitive liabilities as the sum of deposits from other banks and financial institutions, some financial liabilities designated at fair value (debt securities, deposits), and certain debt securities in issue (commercial paper, covered bonds, other debt securities and subordinated debt). The implicit subsidy of the major UK banks (Barclays, HSBC, Lloyds Banking Group, and Royal Bank of Scotland) is estimated equal to around 5, 25, 130, and 40 GBP billion in 2007, 2008, 2009, and 2010, respectively.

Schich and Lindh (2012) make use of a proxy for rating sensitive liabilities equal to the “outstanding bonds and loans” issued in the market, which is readily available from Bloomberg. Schich and Lindh (2012) calculate a lower bound of implicit subsidy based on outstanding bonds and loans issued by the rated entity only, and an upper bound where the debt of subsidiaries are also included.

Bijlsma and Mocking (2013) estimate the relationship between bond yields and Moody’s long-term deposit ratings using ordinary least squares regression. The amount of long-term funding is used as a proxy for rating sensitive liabilities. The total implicit subsidy ranges from around EUR 20 billion in early 2008 to peak at almost EUR 150 billion in mid-2011. It falls to around EUR 100 billion in the beginning of 2012.

In conclusion, irrespective of the methodology used, the sample of banks and the period considered, the amount of the implicit subsidies is significant.

3. Estimation of the implicit guarantee

3.1 Data description

Countries and time period

Our sample consists of 112 European banks from 23 countries (BE, BG, CZ, DK, DE, IE, EL, ES, FR,  IT, CY, LU, HU, NL, AT, PL, PT, RO, SI, SK, FI, SE, UK).[77] The sample represents between 60% and 70% of EU banking total assets as of end of 2011[78]. The EU G-SIBs are all represented in the sample with the exception of Groupe BPCE for which data on ratings are not available.[79] The time period under study is from 2007 to 2012, and the data frequency is six months. For comparison purposes, data for the first quarter of 2013 are also included in the analysis.

Ratings description

The following Moody’s ratings are used in the analysis:

a) Bank Financial Strength Rating (BFSR): it represents the ratings agency’s opinion of the banks intrinsic safety and soundness. It does not take into account the probability that the bank will receive external support from its parental company or public authorities. This is an estimate of SACR in equation (1) above.

b) Baseline Credit Assessment (BCA): the BFSR is mapped to the BCA (Moody’s started publishing the BCA in 2011), which express the bank’s intrinsic financial strength using the aaa – c scale (see Table 1). BCA is therefore also an estimate of SACR in Equation (1) above. This approach was also followed by Schich and Lindh (2012).

c) Adjusted Baseline Credit Assessment (adjBCA): in contrast to the BCA, which represents the bank’s stand-alone rating, the Adjusted BCA factors in the effect of parental and cooperative support (SACR* in Equation (2) above).

d) Long Term Issuer Rating (LT) used to estimate the all-in credit rating (AICR) for each bank. In cases where the Long Term Issuer Rating is not available, the Senior Unsecured or the Long Term bank Deposits ratings is used. LT is reported on the  Aaa – C scale.

e) Sovereign Long Term Rating for the countries under study, necessary for the econometric analysis in Section 5.

Table 1: Moody’s ratings: classes and mapping

BFSR || BCA/adjBCA || LT issuer || Notch

A || aaa || Aaa || 20

A- || aa1 || Aa1 || 19

B+ || aa2 || Aa2 || 18

B || aa3 || Aa3 || 17

B- || a1 || A1 || 16

C+ || a2 || A2 || 15

C || a3 || A3 || 14

C- || baa1 || Baa1 || 13

C- || baa2 || Baa2 || 12

D+ || baa3 || Baa3 || 11

D+ || ba1 || Ba1 || 10

D || ba2 || Ba2 || 9

D- || ba3 || Ba3 || 8

E+ || b1 || B1 || 7

E+ || b2 || B2 || 6

E+ || b3 || B3 || 5

E || caa1 || Caa1 || 4

E || caa2 || Caa2 || 3

E || caa3 || Caa3 || 2

E || ca || Ca || 1

E || c || C || 0

Source: Moody’s 2012b.

Figure 1 gives a schematic representation of the various components of the overall bank rating according to Moody’s methodology.

Figure 1: Components of the overall bank rating according to Moody’s methodology

Variables created and the corresponding time periods

a) As a first step all ratings of each individual bank are transformed into a numerical scale varying from zero to 20, with 20 describing the best rating category (see table above).[80]

b) The UPLIFT is calculated as the difference between the LT and the BCA (as they are measured in the same rating scale). However, as Moody’s started publishing BCA only in 2011 for dates prior to 2011 the BFSR is employed. BFSR is then transformed into BCA according to Table 1. In cases where a specific BFSR rating corresponds to more than one BCA rating the average of those is considered (e.g. C- corresponds to the ratings Baa1 and Baa2).

c) The UPLIFT* is calculated as the difference between the LT and the adjBCA for the period 2011-2012 (4 semi-annual observations)[81].

d) The difference between adjBCA and BCA, which can be interpreted as an implicit measure for parental support quality, is also computed.

Appendix B presents detailed statistics for the various ratings considered for various points in time.

3.2 Descriptive statistics on the UPLIFT/UPLIFT*

Figure 2 shows the evolution over time of the BCA and the UPLIFT. Results suggest that UPLIFT is significant and has been around 3 notches throughout the period. Results are in line with previous literature[82] and show that the stand-alone rating has decreased during the last years, while the UPLIFT is more stable. One can thus conclude that the UPLIFT represents an increasing share of the total rating and that the ratio of the UPLIFT over the individual bank strength is increasing over time. The UPLIFT peaked between 2009 and 2011.

Figure 2: Evolution of the average BCA (estimate of Stand Alone Credit Rating) and average UPLIFT over time

Source: Moody’s and own calculations

Note: UPLIFT is estimated as the difference in notches between AICR and SACR. Prior to 2011 the BFSR to BCA mapping procedure is employed.

Table 2 presents the statistics of the UPLIFT* and

Figure 3 shows the evolution of the UPLIFT* for the available points in time. UPLIFT* is significant and around 2 notches in the period between June 2011 and March 2013. Note that as adjusted stand-alone rating (adjBCA) are published only since 2011, the UPLIFT* cannot be computed in earlier periods.

Table 2: Summary statistics for UPLIFT*

|| # Observations || Simple Average || St. dev || min || max

June 2011 || 89 || 2.34 || 1.35 || 0 || 6

December 2011 || 104 || 1.91 || 1.33 || 0 || 7

June 2012 || 110 || 1.88 || 1.33 || 0 || 6

December 2012 || 112 || 1.96 || 1.50 || 0 || 8

March 2013 || 112 || 2.01 || 1.77 || 0 || 11

Figure 3: Evolution of the average adjBCA (estimate of Stand Alone Credit Rating and parental/cooperative support) and average UPLIFT* over time

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*.

The weighted averages of the adjBCA and UPLIFT*, which are obtained by weighing each observation according to the bank’s total assets, is shown in Figure 4. Although direct comparisons across time periods cannot be made (due to small differences in the sample of banks available in the dataset in each period), one can observe that the weighted average UPLIFT* variable is greater than the simple average UPLIFT* indicating that larger banks are benefiting from a higher government support.

Figure 5 presents the contribution of the parental support (defined as the difference between the adjusted stand-alone rating (adjBCA) and the stand-alone rating (BCA)) to the long term rating. One can conclude that in contrast to the average stand-alone rating (BCA), which has decreased between June 2011 and March 2013, the average parental support has slightly increased.

Figure 4: Evolution of the weighted average adjBCA (estimate of Stand Alone Credit Rating and parental/cooperative support) and weighted average UPLIFT* over time [83]. March 2013 is missing since total assets are not available.

Source: SNL, Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*.

Figure 5: Evolution of the average BCA (estimate of Stand Alone Credit Rating), average UPLIFT* and average parental support over time

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*, parental support is calculated as the difference between SACR* and SACR.

The average UPLIFT* varies between countries implying different levels of support, as can be seen in Figure 6. Belgium, France, Austria and Germany are among the countries for which the estimated UPLIFT* is consistently higher, which is in line with the findings of Schich and Lindh (2012). When comparing Member States one should consider that this analysis includes only the banks in the sample and that simple averages are provided.

Figure 7 presents the change of the average UPLIFT* between June 2011 and March 2013.

Figure 6: Simple average UPLIFT* for 23 European MS (March 2013), in notches. In parentheses one can read the number of banks in the sample for each country.

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*

Figure 7: Changes of average UPLIFT* between June 2011 and March 2013 for 23 European MS

 Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*

The UPLIFT* is also higher for larger banks. The relative difference between the average UPLIFT* of the 25% largest banks in the sample and of the remaining ones is between 7% and 26%, depending on the point in time (see Figure 8). For the 10% largest banks the difference is 5% to 23% (see Figure 9). The relationship between size and UPLIFT* is better examined through the regression analysis in Section 5, which also control for other variables that may affect the UPLIFT*.

Figure 8: Average UPLIFT* for the 25% largest banks (total assets) over time

Source: SNL, Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*

Figure 9: Average UPLIFT* for the 10% largest banks (total assets) over time

Source: SNL, Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*

Appendix D presents scatterplots for each point in time of the UPLIFT* versus the total assets at the four different time periods. No clear discernible pattern can be identified.

Figure 10, Figure 11 and Figure 12 are scatters plots of the average UPLIFT* against the countries’ sovereign rating at different points in time. There is a positive relationship between the country’s sovereign rating and the UPLIFT* in all periods of the analysis. The fact that for countries such as CY, IT, ES one can observe a shift left down through time points towards a positive relationship between the two variables (see also Section 5).

Figure 10: Scatter plot of the average UPLIFT* and country’s sovereign rating for June 2011

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*.

Figure 11: Scatter plot of the average UPLIFT* and country’s sovereign rating for December 2012

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*.

Figure 12: Scatter plot of the average UPLIFT* and country’s sovereign rating for March 2013

Source: Moody’s and own calculations

Note: UPLIFT* is estimated as the difference in notches between AICR and SACR*

4. Estimation of the implicit subsidy

We aim at estimating the funding cost advantage of banks due to an implicit support from the sovereign where the reporting entity is headquartered, for the sample of banks used in the above analysis of the UPLIFT* based on balance sheet data for the end-of-year of 2011 and 2012.

4.1 Methodology

The estimation is done in two steps. Firstly, we compare the funding cost implied by the long-term credit rating of a bank with the counterfactual funding cost based on its stand-alone credit rating, following Schich and Lindh (2012) and Noss and Sowerbutts (2012). We assume that the difference of these two values approximates the potential funding cost advantage (in basis points) due to the implicit subsidy from the sovereign. Secondly, we multiply the funding cost advantage with a share of the outstanding debt.

To perform this analysis we use:

- The banks’ adjust stand-alone rating (the adjusted BCA) and the all-in (long-term) rating;

- A rating-yield curve to map the credit ratings to yield spreads;

- A proxy of the share of each banks liabilities that can be assumed to be rating sensitive in the sense that the cost depends on the credit rating.

4.2 The rating-yield curve map

The rating-yield curve is built by taking weekly observations of market yields for a sample of European banks from 18 countries. The sample consists of 77 banks in 2011 and 82 banks in 2012[84]. We have built one rating-yield curve for the second half-year of 2011 (2011H2) and one for the second half-year of 2012 (2012H2). The data used comes from DG ECFIN’s Bank Watch. The yields given in the Bank Watch data are the market yields for bonds with a maturity of 5 years. Note that the sample used to build the rating-yield map and the sample used to estimate the implicit subsidy (and the UPLIFT*) is not the same. Overall 54 banks in 2011H2 and 57 banks in 2012H2 belong to both samples. A description of the sample used to build the mapping between ratings and yield spreads is given in Appendix E.

The average yield spread for each rating bucket is estimated for 2011H2 and 2012H2 and a non-linear curve is fitted to the estimated average yield spreads (see Figure 13). It can be observed that the average yield spreads (marked with diamonds) have decreased over the full spectrum of rating categories from 2011H2 to 2012H2 and that the yield spread curve is flatter for the latter time period.[85]

Figure 13: Rating-yield spread curve for 2011H2 and 2012H2

Source: DG ECFIN’s Bank Watch and own calculations

Note: The diamonds indicate the average 5-year market yield spreads per rating class.

As mentioned previously, the sample used to build the rating-yield curves and the sample used to estimate the implicit subsidy overlaps partly (54 banks in 2011H2 and 57 banks in 2012H2). Thus, for these banks we can compare the average market yield spreads with the yield spreads given by our rating-yield curve mapping as presented in Figure 14.

Figure 14: Estimated yield spreads derived from long-term ratings versus average observed market yield spreads (in basis points)

Source: Bank Watch and own calculations

From Figure 14, one can see that the average yield spreads derived from the mapping exercise and the actual market spreads are more or less in line with each other in both 2011 and 2012. For these banks present in both samples, the average market yield spreads were 21 basis points and 8 basis points lower than the yield spreads estimated from the rating-yield mapping in 2011 and 2012, respectively. This indicates that the market gave on average a higher funding advantage to these banks than our rating-yield spread map indicates, as banks were on average able to secure funding at lower yield spreads than what we predict.[86] It is therefore likely that we are slightly conservative in our estimate of the implicit subsidy with respect to the calculation of the yield curve. There are, however, some cases where the difference between the observed market yield spreads and the estimated yield spreads are significant.[87]

From the above rating-yield spread curves we estimate the difference between the yield spreads given by using the long-term issuer rating and the yield spreads given by the adjusted BCA.[88] Scatterplots of the yield spread difference versus bank size is shown in Figure 15 and

Figure 16. Section 5 presents an econometric analysis investigating the relationship between funding advantage as measured via the yield spread difference and bank size.

Figure 15: Difference between long-term issuer (all-in) rating yield spreads and adjusted BCA (estimate of Stand Alone Credit Rating and parental/cooperative support) yield spreads in 2011H2 versus banks total assets

Source: Bank Watch, SNL, and own calculations.

Figure 16: Difference between long-term issuer (all-in) rating yield spreads and adjusted BCA (estimate of Stand Alone Credit Rating and parental/cooperative support) yield spreads in 2012H2 versus banks total assets

Source: Bank Watch, SNL, and own calculations.

4.3 Proxies of rating sensitive liabilities

Banks fund their activities with various sources of funding (e.g. bonds, deposits and loans) with different costs. The cost of each of these sources of funding is assumed to be influenced by the banks credit rating(s) differently. These various types of liabilities are therefore expected to benefit from any potential implicit guarantee differently. We have chosen to work with two different proxies of rating sensitive liabilities:

· ‘Total Debt’ as reported in SNL[89] (corresponding to around 20% of total liabilities); and

· ‘Total Long-term Funding’ as reported in Bankscope[90] (corresponding to around 15% of total liabilities).

A description of the sample used for estimation of rating sensitive liabilities can be found in Appendix F.

The data we are using for approximating the rating sensitive liabilities is on a consolidated basis.[91] It includes, therefore, all debt issued by the reporting entity (bank holding company, parent bank or subsidiary bank) and all of its subsidiaries which are consolidated on the reporting entity’s balance sheet. This estimation should be viewed as an upper bound of a large (parent) bank’s rating sensitive liabilities.[92],[93]

4.4 Implicit subsidy estimates

The total estimated implicit subsidy for the banks in the sample based on Total Debt is EUR 95 billion and EUR 82.3 billion in 2011 and 2012, respectively. The estimated total amount based on Total Long-term Funding is lower, EUR 71.5 billion and EUR 58.5 billion in 2011 and 2012, respectively, as can be expected since the Total Long-term Funding is for most banks lower than the Total Debt measure. In relative terms, the implicit subsidy estimate is 0.8% and 0.6% of EU-27 GDP for the first estimate and for the second estimate, respectively, for 2011. For 2012 the figures are lower, 0.7% and 0.5%, respectively.

Estimated implicit subsidy for the banks in the sample is presented country by country in Figure 17- Figure 20 and in Table 3. The banks in France, Germany and UK have the largest implicit subsidy estimates in absolute terms, regardless of which estimate is used (Figure 17 and Figure 19), due to the fact that many of the largest banks (measured by total assets) are headquartered in these countries and also because the relatively larger number of banks from these countries included in the sample. For France and Germany, the implicit subsidy estimates seem stable during the two periods. In most countries there is a reduction in the implicit subsidy, driven by a combination of reduction in the difference between (the adjusted) standalone yield spread and all-in yield spread and a reduction in reported total debt and long-term funding for some banks. A notable exception is Portugal, for which relatively large increases of implicit subsidy estimates can be observed (230-250% increase, respectively). [94]

When measuring the implicit subsidy relative to domestic GDP, Sweden and France (together with Portugal in 2012) are in the top, inter alia, due to the significant domestic banks' size compared to GDP in the sample (Figure 18 and Figure 20).

The results are broadly in line with other studies. However, one should be cautious when making a comparison of results across studies as differences may arise due to: differences in definition of rating sensitive liabilities, the estimation of yield spreads and the sample of banks as well as the sample period. A detailed comparison is provided in Appendix C.

Figure 17: Estimated implicit subsidy per country (EUR billion) based on Total Debt (89 banks in 2011 and 90 banks in 2012)

Source: Bank watch, SNL and own calculations

Figure 18: Estimated implicit subsidy per country (relative to domestic GDP) based on Total Debt (89 banks in 2011 and 90 banks in 2012)

Source: Bank watch, SNL and own calculations

Figure 19: Estimated implicit subsidy per country (EUR billion) based on Total Long-term Funding (96 banks in 2011 and 102 banks in 2012)

Source: Bank watch, Bankscope and own calculations

Figure 20: Estimated implicit subsidy per country (relative to domestic GDP) based on Total Long-term Funding (96 banks in 2011 and 102 banks in 2012)

Source: Bank watch, Bankscope and own calculations

Table 3: Implicit subsidy country-by-country

Country || Implicit subsidy (EUR billion) based on Total Debt || Implicit subsidy (EUR billion) based on Total Long-term Funding

|| 2011 || 2012 || 2011 || 2012

AT || 2.8 || 2.4 || 2.5 || 2.0

BE || 4.2 || 2.8 || 2.3 || 1.7

BG || 0.0 || 0.0 || 0.01 || 0.0

CY || 0.06 || 0.01 || 0.01 || 0.01

CZ || 0.07 || 0.09 || 0.04 || 0.04

DE || 18.1 || 17.0 || 12.2 || 10.8

DK || 1.3 || 1.1 || 1.2 || 1.0

EL || 0.2 || 0.1 || 0.2 || 0.1

ES || 4.3 || 2.9 || 5.4 || 4.9

FI || 0.7 || 0.6 || 0.3 || 0.3

FR || 30.1 || 31.0 || 20.0 || 17.2

HU || 0.0 || 0.0 || 0.0 || 0.0

IE || 0.3 || 0.1 || 0.3 || 0.1

IT || 6.3 || 4.4 || 6.0 || 4.2

LU || 0.4 || 0.2 || 0.3 || 0.1

NL || 2.1 || 1.9 || 0.9 || 1.7

PL || 0.1 || 0.1 || 0.0 || 0.1

PT || 0.6 || 2.1 || 0.5 || 1.8

RO || 0.00 || 0.01 || 0.00 || 0.01

SE || 9.3 || 6.7 || 5.7 || 4.4

SI || 0.0 || 0.1 || 0.0 || 0.1

SK || 0.1 || 0.0 || 0.1 || 0.0

UK || 13.9 || 8.6 || 13.5 || 7.9

Source: SNL, Bankscope and own calculations

4.5 Comparing implicit subsidy with pre-impairment operating profit[95], [96]

We also compare the aggregated estimated implicit subsidy with the aggregated pre-impairment operating profit. The total pre-impairment operating profit was EUR 197.5 billion and EUR 175.5 billion, in 2011 and 2012, respectively. The aggregated implicit subsidy of the sample was around 35-50% and 30-50% of pre-impairment operating profit in 2011 and 2012. These figures underline the importance of the estimated implicit subsidies as they account for a significant part of the pre-impairment operating profits of the banks. A more detailed description of the calculations and a country-by-country analysis is given in Appendix G.

4.6 Relation between implicit subsidy and bank characteristics

There is a positive relation between the size of the implicit subsidy and the size of the bank if both variables are measured in absolute terms (see Figure 21 and Figure 22). This could be expected since a large bank would have a larger amount of debt. Comparing the ratio of the implicit subsidy relative to the bank’s total assets, such a positive relation is not discernible (see Appendix H).

We have investigated also the relation between implicit subsidy and various bank characteristics (leverage, share of total assets held for trading, Tier1 capital and RWA) and found no apparent pattern. Graphs are provided in Appendix H.

Figure 21: Implicit subsidy versus total assets (end of 2011)

Source: Bank Watch, SNL and own calculations

Note: The outlier at the top left is Dexia Credit Locale SA.

Figure 22: Implicit subsidy versus total assets (end of 2012)

Source: Bank Watch, SNL and own calculations

5. Determinants of the implicit guarantee

Our objective is to test the following three hypotheses:

· The implicit guarantee is larger in EU countries with high sovereign ratings. Governments with higher ratings should be in a better position for providing support to the banking sector.

· For a given stand-alone rating, the implicit guarantee is greater for large banks. Large banks are often considered TBTF and therefore could be more likely to benefit from an implicit state support.

· The business profile has an impact on the implicit guarantee. Retail-oriented, wholesale and investment banks may have benefitted from a different implicit subsidy.

These hypotheses are checked via significance tests in Probit regressions. We shall verify that the model confirms the expectation that weak banks with low stand-alone rating and weak parental support receive a larger implicit subsidy compared to strong banks.

Section 5.1 describes our dataset, Section 5.2 details the econometric methodology, Section 5.3 present the empirical results for the first two hypotheses about the sovereign rating and the bank size. Section 5.4 reports results about the effect of the business profile.

5.1 Dataset and explanatory variables

Our database is made up of Moody’s ratings for 112 EU banks for four semesters over the years 2011-2012. The implicit guarantee is reflected in the UPLIFT*. This variable is obtained as the difference between the long-term and the stand-alone rating that Moody’s assigns to each bank. The long-term rating takes into consideration the possibility of a government support. The bank stand-alone rating is measured by Moody’s Baseline Credit Assessment (BCA) variable.  Moody’s also makes available a stand-alone rating that includes the possibility of a parental support (adjBCA). We use the difference P between these two variables as an indicator of the parental support quality, i.e. P = adjBCA – BCA. Moody’s rating of the sovereign debt is labelled Sov.

The banks in our dataset have received an UPLIFT* from 0 to 7 notches. The empirical distribution of the UPLIFT* is displayed in Table 4 below whereas the cumulative distribution is shown in Figure 23.

Table 4: Empirical distribution of the UPLIFT* for 112 EU banks over 2011-2012

UPLIFT* || 0 || 1 || 2 || 3 || 4 || 5 || 6 || 7

Frequency || 15% || 24% || 23% || 29% || 4% || 4% || 1% || 1%

Figure 23: Cumulative density function of the UPLIFT* over 2011-2012

Box-plots of BCA, P, and Sov according to each UPLIFT* category are displayed in Figure 24 below. One can observe that banks with relatively low stand-alone and parental support enjoy high levels of UPLIFT*. Similarly banks headquartered in countries with high sovereign ratings enjoy a higher UPLIFT*.

Figure 24: Box-plots for stand-alone rating (BCA), parental support rating (P) and sovereign debt rating (Sov)

Notes: in the Box-plots the central mark corresponds to the median, the edges of the box are the 25th and 75th percentiles, and the red crosses point to outliers.

The banks size and profile are described using publicly available balance sheet information collected from the SNL database. The currency unit is euro.

The bank size is measured by the logarithm of the total assets, logTA = logarithm(Total Assets).

Following existing literature (see Ayadi et al. 2012, Blundell-Wignal and Roulet, 2013 and IMF, 2013), we describe the bank business profile using the following variables:

1. AHTTA: the ratio of assets held for trading to total assets.

2. STFTA: the ratio of short-term funding to total assets.

3. DEPCTA: the ratio of customers deposits to the total assets.

4. NETLOANSCTA: the share of net loans to customers in total assets.

5. DEPBTA: the ratio of bank deposits to the total assets.

6. NETLOANSBTA: the share of net loans to banks in total assets.

7. WR: the wholesale funding ratio calculated by dividing Total Financial Liabilities minus Customers Deposits by Total Financial Liabilities. This measures the share of funding coming from the wholesale market.

8. FeeToOpInc: the ratio of net fees and commissions to operating income.

9. NetToOpInc: the ratio of net interest income to operating income.

10. DEVToFL: the ratio of derivatives liabilities to total financial liabilities.

11. TOTDERTA: Derivatives assets plus derivatives liabilities compared to total assets.

12. EquiTA: the ratio of total equity to total assets.

13. RWATA: the ratio of risk-weighted- assets to total assets.

14. T1RWA: Tier 1 ratio, which is the amount of Tier 1 capital over risk-weighted-assets.

15. T1T2RWA : Tier 1 plus Tier 2 ratio.

Broadly speaking, investment-oriented banks are characterized by high volumes of assets held for trading and gross market value of derivatives compared to their total assets (AHTTA and TOTDERTA). Accordingly, the ratio of derivatives liabilities to total financial liabilities (DEVToFL) is expected to be high. Market operations make net fees and commissions contribute significantly to the operating income (FeeToOpInc).

Wholesale banks are highly inter-connected with other financial institutions. They are thus characterized by a large share of bank loans in their total assets (NETLONSBTA) and of bank deposits in their liabilities (DEPBTA). They are expected to have a large wholesale funding ratio (WR).  They should rely relatively more on short-term funding (STFTA), which has been identified as having played a key role in the financial crisis.

Retail banks have a relatively high share of customer loans in their total assets (NETLOANSCTA) and of customer deposits in their liabilities (DepCTA). Net interest income (NetToOpInc) is expected to contribute significantly to their operating income.

The ratio of total equity to total assets (EquiTA) is a simple proxy for leverage: risk-averse banks are characterized by a high level of total equity compared to their total assets.

Finally, we include two regulatory capital ratios: Tier 1 ratio (T1RWA) and total regulatory capital ratio (T1T2RWA). We also consider the average risk-weighted assets (RWATA) which should reflect the risk profile of the bank as measured by the Basel regulation, although some researchers have questioned the appropriateness of RWATA as risk-indicator (see Blundell-Wignal and Atkinson, 2011, and Ayadi et al., 2012). For instance Ayadi et al. (2012) consider RWATA to be a good indicator of the underlying risk for both wholesale and retail banking but not for investment banks. One possible reason they put forward is that investment banks may perform regulatory arbitrage through RWATA optimization.

These variables, together with logTA, are displayed in Figure 25 and Figure 26 in the next two pages.[97]

Figure 25: Box-plots for various balance sheet indicators

Notes: logTA: logarithm of total assets; AHTTA: assets held for trading as a share of total assets; RWATA: risk-weighted assets as a share of total assets; STFTA: short-term funding as a share of total assets; DepCTA: customer deposits as a share of total assets; NETLOANSCTA: net loans to customers as a share of total asset; DepBTA: bank deposits as a share of total assets; NETLOANSBTA: net loans to banks as a share of total asset; WR: wholesale funding ratio. In the box-plots the central mark corresponds to the median, the edges of the box are the 25th and 75th percentiles, and the red crosses point to outliers.

Figure 26: Box-plots for various balance sheet indicators

Notes: FeeToOpInc: the ratio of net fees and commissions to operating income; NetToOpInc: the ratio of net income to operating income; DEVTOFL: the ratio of derivatives liabilities to total financial liabilities; EquiTA: total equity over total assets; TOTDERTA: total derivatives in proportion of total assets; T1RWA: Tier 1 ratio; T1T2RWA: Tier 1 plus Tier 2 ratio. In the box-plots the central mark corresponds to the median, the edges of the box are the 25th and 75th percentiles, and the red crosses indicate outliers.

The box-plots in the two figures suggest that:

· The larger the log of total assets, the larger the UPLIFT*.

· A high share of asset held for trading to total assets ratio (AHTTA) tends to be associated with an UPLIFT* equal to 2, 3, or 4.

· High values of risk-weighted assets to total assets (RWATA) tend to be associated with a low UPLIFT*.

· The lower the share of customer deposits in total assets (DEPCTA) the larger the UPLIFT*.

· High proportions of net loans to customers in total assets (NETLOANSCTA) tend to be associated with a low UPLIFT*.

· The larger the proportion of net loans to banks in total assets (NETLOANSBTA) the larger the UPLIFT*.

· The larger the wholesale funding ratio (WR) the larger the UPLIFT*.

· Large ratios of derivatives liabilities to total financial liabilities (DEVToFL) seem to be associated with a large UPLIFT*.

· Larger amounts of total equity compared to total assets (EquiTA) are associated with a smaller UPLIFT*.

· Large amounts of derivatives assets plus derivatives liabilities compared to total assets (TOTDERTA) seem to be predominantly associated with an UPLIFT* equal to three notches.

For short-term funding (STFTA), banks deposits (DepBTA), net fees and commissions to total income ratio (FeeToOpInc), net interest income to operating income ratio (NetToOpInc),  and the two regulatory variables (T1RWA and T1T2RWA), no systematic  impact on the UPLIFT* can be read.

Some of these variables are highly correlated (>0.80): asset held for trading to total assets (AHTTA) and derivatives assets plus derivatives liabilities to total assets (TOTDERTA) at 0.93, AHTTA and derivatives liabilities to total financial liabilities (DEVToFL) at 0.88, customer deposits to total assets (DEPCTA) and wholesale funding ratio (WR) at -0.99, TOTDERTA and DEVToFL at 0.96, as well as the two regulatory capital ratio (T1RWA and T1T2RWA) at 0.90. Although not reported in the variable list, also the logarithm of total deposits is correlated with the logarithm of total assets at 0.96. Since regression analysis does not allow to discriminate between two explanatory variables that are so highly correlated, we shall not use simultaneously any two of these variables.

5.2 Econometric methodology

The dependent variable UPLIFTi*, i=1,...,N, belongs to 8 categories that for convenience we index with value j=0,…,m. Since our dependent variable is ordinal with several outcomes, we fit an ordered Probit model. Grouping the explanatory variables into the vector xi, we define the latent variable yi* such as:

UPLIFTi* = 0             if          yi* = xi' β + εi < μ1

UPLIFTi* = j if          μj < yi* = xi' β + εi < μj+1   j=1,…,m-1

UPLIFTi* = m            if          μm < yi* = xi' β + εi

where the constants μ1,…, μm, are the cut-offs. The Probit model assumes that the shocks ε follow a standard normal distribution. Under this assumption the probability that the UPLIFT* for bank i falls into category j is equal to:

P(UPLIFTi*=0|xi)       =  P(yi* < μ1) = Φ(μ1 - xi' β)

P(UPLIFTi*=j|xi)        =  P(μ j < yi* < μj+1)

= P(μj < xi' β  + εi  < μj+1)

= P(μj - xi' β < εi  < μj+1- xi' β)

= Φ(μj+1- xi' β) - Φ(μj- xi' β)

P(UPLIFTi*=m|xi)      = P(μm < yi* ) = 1- Φ (μm - xi' β)

where Φ() represents the standard normal cumulative distribution. The likelihood function is such as:

P(UPLIFT1*,…, UPLIFTN*|x1,…, xN) = Πi=1N Πj=0m  P(UPLIFTi*=j|xi) 1(UPLIFTi*=j)

where 1(UPLIFTi*=j)=1  if UPLIFTi*=j and 0 otherwise. The probability P(UPLIFTi*=j) is given in the previous system of equation. Maximizing the likelihood function gives estimates of the coefficients β as well as of the cut-off values μ1,…, μm.

The hypothesis of independent ε-shocks implies that the UPLIFT* assigned to a bank given its own characteristics and the rating of the country where it is headquartered does not depend neither on the rating of the other countries nor on the profile of the other banks. In other words, the model considers that Moody’s evaluates EU banks one by one and not altogether at the same time.

5.3 Empirical results without business model variables

Our first model relates the UPLIFT* to the stand-alone rating, the parental support, the sovereign debt rating, and the bank size. Business model variables are left out in this preliminary round. Putting together data for the two semesters of 2011 and 2012 gives a maximum number of 448 observations. Because the variables have missing values at different places, in each regression the effective number of observations depends on the subset of variables selected. The Probit regression results are described in Table 5.

 

Table 5: Model 1, N=364 observations

Variable || Coefficient || t-statistics

Stand-alone rating (BCA) || -.382 || -12.12

Parental support (P) || -.549 || -8.69

Sovereign rating (Sov) || .367 || 15.03

Size of bank (logTA) || .230 || 5.37

Log-lik || -426.55

Log-lik/N || -1.17

The cut-offs are estimated as (μ1,…,μ7)=(3.75, 5.22, 6.30, 7.93, 8.34, 9.18, 10.09) with standard error around 0.80. Table 6 also reports the log-likelihood and the average log-likelihood by observation at the aim of model comparison. Model 1 fit is shown in Table 6 below. It reports the empirical distribution of UPLIFT* over the 364 observations used in Model 1 together with the unconditional distribution of UPLIFT* as predicted by the model.[98]

Table 6: Model 1 goodness of fit

UPLIFT* || 0 || 1 || 2 || 3 || 4 || 5 || 6

Frequency || 0.12 || 0.26 || 0.24 || 0.30 || 0.04 || 0.03 || 0.01

Model prediction || 0.08 || 0.27 || 0.29 || 0.30 || 0.03 || 0.03 || 0.01

As can be seen from Table 6, Model 1 fits the data reasonably well: the largest discrepancy between empirical and predicted probability is equal to 5 percentage points (pp) for the class UPLIFT*=2. For all other classes the discrepancy is less than 4 pp. We now turn to the interpretation of the Probit results.

The sign of the coefficients in Table 5 suggests that:

· Banks with higher stand-alone rating (BCA) have a greater probability to get a small UPLIFT*. As BCA also reflects the riskiness of a bank, the negative coefficient implies that a more risky bank is likely to benefit from a greater UPLIFT*.

· A higher parental support (P) is associated with a smaller UPLIFT*. A bank affiliated to a strong group can receive support within the group and is thus less likely to require public support.

· Banks with higher sovereign rating have a greater probability to receive a large UPLIFT*. A higher sovereign rating implies that the government is in a better position to provide such support.

· Big banks enjoy a larger UPLIFT*: for a given stand-alone and parental rating, the larger the bank the greater the UPLIFT*.

To get a more quantitative picture, we calculate the marginal effect, i.e. the impact on the UPLIFT* distribution of a variation around the average of one of the bank characteristics, the other ones being kept constant at their average value. In the linear regression, a positive coefficient implies a positive impact of the variable. Instead in a Probit regression the explanatory variables describe a probability distribution, so the marginal effect of each variable is positive for some categories of the UPLIFT* and negative for the others. Table 7 below reports such marginal effects in Model 1 – see Appendix J for methodological details.

Table 7: Marginal effects in Model 1

|| || UPLIFT*

|| Average (st. error) || 0 || 1 || 2 || 3 || 4 || 5

P(UPLIFT*=j|AX)  |||| .023 || .276 || .409 || .277 || .010 || .005

Stand-alone rating (BCA) || 10.56 || .021* || .112* || -.001 || -.117* || -.009* || -.005*

Parental support  (P) || 0.72 || .030* || .160* || -.002 || -.168* || -.010* || -.005*

Sovereign rating (Sov) || 16.09 || -.020* || -.107* || .001 || .111* || .009* || .005*

Size of bank  (logTA) || 18.65 (1.52) || -.013* || -.067* || .001 || .070* || .005* || .003*

Notes: * indicates numbers that are significantly different from 0 at the 10% level. The standard error of logTA is given between parentheses. The average values reported for BCA, P, and Sov are related to the 0-20 coding (see Section 3.1). Notice that the probability that the average bank gets one particular UPLIFT*, i.e. P(UPLIFT*=j|AX), should not be compared with the unconditional probability reported in Table 6.

We observe that:

· The stronger the stand-alone rating (BCA) the lower the probability of a large UPLIFT*: a BCA above average by one notch increases the probability that the UPLIFT* does not exceed one notch by 13 pp, from 0.30 to 0.43[99]. Conversely, the probability of an UPLIFT* strictly greater than two notches decreases by 13 pp from 0.29 to 0.16.

· A parental support above average by one notch increases the probability that the UPLIFT* does not exceed one notch by 19 pp, from 0.30 to 0.49. Conversely, the probability of receiving an UPLIFT* strictly greater than two notches decreases by 18 pp from 0.29 to 0.11.

· A sovereign rating above average by one notch increases the probability that UPLIFT* is greater or equal to three notches by 13 pp, from 0.29 to 0.42. Conversely, the probability of receiving an UPLIFT* of zero or one notch decreases by 13 pp from 0.30 to 0.17.

· The average logTA is equal to 18.65 with standard error 1.52. There is considerable heterogeneity among the banks in our dataset since this implies a variation by 350% on the original scale. Banks whose logTA variable is above the average bank size by one standard error have a probability of receiving an UPLIFT* greater than two notches by 12 pp (1.52*(.070+0.005+0.003)), from 0.29 to 0.41. Conversely, the probability of receiving an UPLIFT* of no more than one notch shrinks from 0.30 to 0.18, which is obtained as 0.30-(1.52*(.013+0.067)). Hence, banks with larger amount of total assets benefit from a larger UPLIFT*.

These inferences are in broad agreement with the data features shown in Figure 24 to Figure 25. They seem to be robust to several alternatives. For instance, re-weighting the observations using logTA (option “aw” in STATA) does not harm the results. We have checked that the regression is stable over the four semesters by adding time dummies: only the dummy for the first semester of 2011 is significant. As the coefficients of the other variables are almost invariant we do not consider time dummy variables. We also tried country dummies but no improvement was obtained[100].

Some endogeneity may be present in Model 1. First, as acknowledged in Moody’s (2012a) sovereign debt rating methodology, the sovereign debt assessment takes into consideration the risks in the banking sector of the country. Moody’s uses the latest data available, so there is likely simultaneity. Second, the assessment of a bank is based on the bank’s balance sheet and income statement (Moody’s, 2012b). Since the all-in rating[101] influences funding costs, the all-in rating impacts on both the balance sheet and the income statement. The impact on bank bond spreads have been analysed for instance in Morgan and Stiroh (2005) and Resti and Sironi (2005). Next, Noss and Sowerbutts (2012) argue that “The implicit guarantee … distorts banks’ risk-taking incentives as investors no longer fully price the risks they are aware the banks are taking, allowing banks to take more risk. A pernicious spiral can therefore develop, where the existence of an implicit guarantee encourages banks to take more risk” (see also Alessandri and Haldane, 2009).

Such endogenous mechanisms (simultaneity and feedback) imply a bias in the coefficients estimates. To correct for this we substitute the country debt to GDP ratio for the sovereign rating[102] (see also Estrella and Schich, 2012). In order to break any contemporaneous feedback we use a lag for both the debt to GDP ratio and the balance sheet variables. The lag removes Moody’s ratings in the first semester of 2011 from the regression.

Table 8 below shows the results.

Table 8: Model 1B with Debt/GDP ratio substituted for the sovereign rating. Both Debt/GDP and logTA are lagged, 283 observations

Regressors || Coeff. || t-stat.

Stand alone rating (BCA) || -0.158 || -5.72

Parental support (P) || -0.051 || -1.04

Lagged debt to GDP ratio || -0.022 || -7.38

Lagged size of bank (logTA) || 0.522 || 8.73

Log-lik || -392.77

Log-lik/N || -1.39

We observe that:

· The lagged Debt/GDP ratio has negative and highly significant coefficient, showing that banks located in countries with high debt to GDP ratio have received a lower UPLIFT*.

· The BCA coefficient is smaller in absolute value but still significant: the impact of BCA is reduced but the direction is preserved.

· The parental support looses significance.

· The coefficient of the log of total assets lagged by one period is larger with high significance.

· Substituting the lagged Debt/GDP ratio for the sovereign rating has an adverse impact on the model fit: the log-likelihood by observation is 20% smaller. 

Hence, if an endogenous effect is present, it may have harmed the inference about the impact of the parental support but the model-based inference about the impact of the bank size appears rather robust. As a further robustness check, we use the yield benefit enjoyed by the bank as dependent variable instead of UPLIFT*. Since the yield benefit is a continuous variable and not a categorical one, a simple least-square regression is fitted. Table 9 below shows the results.

Table 9: Model 1C with the yield benefit as dependent variable, 183 observations

Regressors || Coeff. || t-stat.

Stand alone rating (BCA) || -.329 || -9.34

Parental support (P) || -.253 || -3.37

Lagged debt to GDP ratio || -.022 || -5.69

Lagged size of bank (logTA) || .460 || 6.10

R2 || 0.34

Compared to Table 8 the coefficients estimates appear to be stable enough, except for the parental support whose significance is restored. Hence the empirical evidence about the impact of the sovereign rating and of the bank size is robust to the use of either UPLIFT* or the yield benefit as dependent variable.

We take Model 1B specification in Table 8 as benchmark for examining the role of the various balance sheet variables.

5.4 Empirical results with business model variables

We evaluate the importance of the business model variables first by checking their significance when added one by one to Model 1B. All balance sheet variables are lagged by one period, still for mitigating endogeneity. The short-term funding variable STFTA has been discarded due to numerous missing observations that reduce the effective sample size to one-third compared to the case where STFTA is left out. Out of the fifteen candidates, this first stage identifies RWATA, DEPCTA, NETLOANSBTA, WR, and EquiTA as variables with potential explanatory power. In a second round we add these five variables altogether to Model 1B and remove the ones that appear to be non-significant, namely RWATA and DEPCTA. We are left with the following model:

Table 10: Model 2 with business variables, 264 observations

Regressors || Coeff. || t-stat.

Stand alone rating (BCA) || -.133 || -4.29

Parental support (P) || -.201 || -3.18

Lagged debt to GDP ratio || -.022 || -6.97

Lagged size of bank (logTA) || .366 || 5.01

Lagged net loans to banks (NETLOANSBTA) || 3.729 || 4.27

Lagged wholesale funding ratio (WR) || .970 || 2.42

Lagged total equity to total assets (EquiTA) || -6.127 || -2.14

Log-lik || -345.54

Log-lik/N || -1.31

The cut-offs are estimated at (2.524, 3.750, 4.517, 6.128, 6.448, 6.840, 7.222). We observe that:

· Model 1B and 2 have a similar fit.

· The coefficient of the stand-alone rating, the lagged debt to GDP ratio, and the lagged log of total assets are stable compared to Model 1B in Table 8. There is some instability in the coefficient of the parental support.

· The coefficient of the lagged ratio of net loans to banks to total assets (NETLOANSBTA) is positive, showing that Moody’s gives a higher UPLIFT* to banks that are more active in bank lending.

· The wholesale funding ratio (WR) has a positive impact on UPLIFT*: the larger WR, the greater UPLIFT*.

· The coefficient of the lagged equity to total assets ratio (EquiTA) is negative: Moody’s gives a lower UPLIFT* to banks that have a high EquiTA ratio. Hence better capitalized banks have a lower UPLIFT*. This leverage variable can be seen as a proxy for the stand-alone rating: for instance we could observe that its significance further increases when the stand-alone rating is not included in the regression.

Table 11 below reports the goodness of fit by UPLIFT* category for Model 2.

Table 11: Goodness of fit of Model 2

UPLIFT* || 0 || 1 || 2 || 3 || 4 || 5 || 6

Frequency || 0.11 || 0.30 || 0.22 || 0.31 || 0.02 || 0.02 || 0.01

Model prediction || 0.08 || 0.30 || 0.27 || 0.31 || 0.02 || 0.01 || 0.00

The empirical frequencies vary slightly because the effective common sample changes according to the missing values in each explanatory variable. As expected, Model 1 and 2 have a similar fit. The maximum error on the empirical frequencies is limited to 5 pp.

Table 12 below displays the marginal effects under Model 2.

Table 12: Marginal effects on UPLIFT* categories for Model 2

|| Average (st. error) || 0 || 1 || 2 || 3 || 4 || 5

P(UPLIFT*=j|AX) |||| .061 || .313 || .298 || .308 || .011 || .006

Stand alone rating (BCA) || 10.10 || .016* || .034* || -.002 || -.041* || -.003* || -.002*

Parental support (P) || 0.76 || .024* || .052* || -.003 || -.062* || -.004* || -.003*

Lagged debt to GDP ratio || 78.19 (30.98) ||  .003* || .006* || -.000 || -.006* || -.001* || -.001*

Lagged size of bank (logTA) || 18.57 (1.49) || -.044* || -.095* || .006 || .115* || .009* || .006*

Lagged net loans to banks (NETLOANSBTA) || 0.10 (0.08) || -.449* || -.963* || .065 || 1.168* || .091* || .056*

Lagged wholesale funding ratio (WR) || 0.48 (0.22) || -.117* || -.251* || .017 || .304* || .023* || .014

Lagged total equity to total assets (EquiTA) || 0.06 (0.03) || .738 || 1.583* || -.108 || -1.919* || -.149 || -.092

Notes: * indicates numbers that are significantly different from 0 at the 10% level. The standard error of continuous variables is given between parentheses.

Comments:

· The marginal effects of stand-alone rating (BCA) and parental support (P) are broadly stable compared to Model 1B. We do not comment further on these variables.

· Also the bank size has a similar impact over Model 1B and 2: both models foresee a positive and significant effect on the probability to get an UPLIFT* greater than 2 when total assets is above average.

· Banks located in a country where the debt to GDP ratio is above average by one standard error have a probability of receiving an UPLIFT* greater than two notches that shrinks by 25pp (30.98*(-.006-.001-.001)), while the probability of receiving an UPLIFT* of no more than one notch increases by 28pp (30.98*(.006+.003)).

· Banks whose net loans to banks to total assets ratio is above average by one standard error have a probability of receiving an UPLIFT* greater than two notches that increases by 11pp (0.08*(1.168+.091+.056)), while the probability of receiving an UPLIFT* of no more than one notch decreases by 11pp (0.08*(-.449-.963)).

· Banks whose wholesale funding ratio is above average by one standard error have a probability of receiving an UPLIFT* greater than two notches that increases by 8pp (0.22*(.304+.023+.014)), while the probability of receiving an UPLIFT* of no more than one notch decreases by about 8pp (0.22*(-.117-.251)).

· Banks whose total equity over total assets ratio is above average by one standard error have a probability of receiving an UPLIFT* greater than two notches that decreases by 6pp (0.03*(-1.919 -.149 -.092)), while the probability of receiving an UPLIFT* of no more than one notch increases by about 7pp (0.03*(.738+1.583)).

This inference is in broad agreement with the Box-plots in Figure 25 and Figure 26.

We have also substituted the sovereign rating for the debt to GDP ratio in Model 2. All coefficients remain significant except the wholesale funding ratio whose impact becomes insignificant. As a further robustness check, we have re-scaled the UPLIFT* variable by stand-alone plus parental rating, i.e. UPLIFT*/(21–BCA-P), as in Bijlsma and Mocking (2013). This re-scaled variable takes into account that the absolute uplift is constrained by the adjusted BCA. Running a least-square regression, we could observe that all variables in Model 2 remain significant as in Table 10.

Furthermore, to check the robustness of Model 2, we substitute the yield benefit for the UPLIFT*. Table 13 below shows the results. All coefficients remain significant except the total equity to total assets ratio which looses significance.

Table 13: Model 2B with the yield benefit as dependent variable. 166 observations

Regressors || Coeff. || t-stat.

Stand alone rating (BCA) || -0.297 || -7.81

Parental support (P) || -0.350 || -4.48

Lagged debt to GDP ratio || -0.020 || -5.30

Lagged size of bank (logTA) || 0.279 || 2.94

Lagged net loans to banks (NETLOANSBTA) || 3.014 || 2.83

Lagged wholesale funding ratio (WR) || 0.919 || 1.95

Lagged total equity to total assets (EquiTA) || -4.914 || -1.34

R2 || 0.393

6. Conclusions

This report investigates the size and determinants of the implicit state guarantee, as measured by UPLIFT*, enjoyed by a sample of EU banks. It also estimates in monetary terms the implicit subsidy due to this guarantee.

The size of the implicit guarantee is measured using the rating-based approach based on Moody’s credit ratings. Long-term credit ratings take into account the bank’s stand-alone financial strength rating, a rating upgrade due to possible support from a parent or a cooperative group, and a rating upgrade due to potential government support. This third component (which in the report is denoted by UPLIFT*) is used to measure the implicit government guarantee. The implicit guarantee is translated into a funding cost advantage by comparing the funding cost inferred from the long-term credit rating of a bank with the funding cost inferred from its stand-alone credit rating. Multiplying the funding cost advantage with a share of the outstanding debt results in a monetary estimate which can be viewed as an implicit subsidy given to banks.

Main conclusions on the size of the implicit government guarantee (measured through the UPLIFT*):

· Statistics on our Moody’s dataset on EU banks ratings, covering 112 banks from 23 EU countries in the period 2011-2013, are in line with the few studies available for the EU and confirm that many EU banks enjoy a government implicit guarantee that increases their ratings.

· In the period 2011-2013 the largest part of the distribution of the UPLIFT* lies between 1 and 3 notches on average.

· The top 25% banks in size (roughly 25 banks, where most of the EU G-SIBs are represented) enjoy an implicit government guarantee larger than the remaining banks in the sample. The relative difference between the average UPLIFT* of largest banks and the remaining ones is between 7% and 26%, depending on the point in time.

· The average upgrade due to implicit guarantee is relatively stable over the period under study, while a decrease in the average long-term credit rating can be observed.

Main conclusions on the implicit subsidy:

· The reduction in banks' funding costs due to the implicit government guarantee obtained via the rating-yield map is largely in line with previous literature studies.

· The total implicit subsidy for the considered sample is estimated in the interval EUR 72 to 95 billion euro in 2011 and EUR 59 to 82 billion euro in 2012.

· In relative terms, the implicit subsidy for ranges between 0.5% and 0.8% of EU-27 GDP.

· The estimated implicit subsidy is 0.4-0.7% GDP for DE, 0.9-1.5% for FR and 0.4% for UK as of end 2012.

· The estimated implicit subsidy is between 30% and 50% of the banks aggregated annual pre-impairment operating profit.

Main conclusions on implicit state guarantee (measured through the UPLIFT*) drivers:

In our sample of EU banks, the UPLIFT* appears to be driven by:

· The bank individual strength: as expected, a higher individual strength rating increases the probability of observing lower UPLIFT*;

· The parental support: as expected, banks with stronger parents receive a lower UPLIFT*.

· The credit rating of the country where a bank is headquartered: being headquartered in a country with a strong sovereign credit rating increases the probability of observing a higher UPLIFT*. This evidence also appears if the debt to GDP ratio is used as a proxy for the sovereign credit worthiness.

· The size of the bank: larger banks are more likely to enjoy a greater UPLIFT*.

Concerning the business model we find the following evidence:

· More inter-connected banks that hold a larger proportion of net loans to banks in their total assets benefit from a higher UPLIFT*.

· Banks that rely relatively more on the wholesale market for funding compared to other banks (i.e. a high wholesale funding ratio) also benefit from a higher UPLIFT*.

· Better capitalized banks are more likely to receive a lower implicit guarantee.

References

Alessandri P. and Haldane A., 2009, Banking on the state, twelfth annual International Banking Conference on ‘The international financial crisis: have the rules of finance changed?’, Federal Reserve Bank of Chicago.

Ayadi R., Arbak E., and de Groen W.P., 2012, Regulation of European Banks and Business Models: Towards a New Paradigm?, Centre for European Policy Studies, Brussels. 

Bijlsma M.J. and Mocking R.J.M., 2013, The private value of too-big-to fail guarantees, discussion paper 240, CPB Netherlands Bureau for Economic Policy Analysis, available at www.cpb.nl/sites/.../dp240-private-value-too-big-fail-guarantees_0.pdf‎.

Blundell-Wignall A. and Roulet C., 2013, Business models of banks, leverage and the distance-to-default, OECD Journal: Financial Market Trends, 103, 1-29.

Blundell-Wignall A. and Atkinson, P., 2011, Global SIFIs, Derivatives and Financial Stability OECD Journal: Financial Market Trends, 1, 1-34.

Estrella A. and Schich S., 2012, Sovereign and Banking Sector Debt: Interconnections through guarantees, OECD Journal: Financial Market Trends, Volume 2011, Issue 2.

Haldane A. G., 2010, The $100 Billion Question, BIS Review 40/2010.

Haldane A. G., 2011, Control Rights (and Wrongs), speech at the Wincott Annual Memorial Lecture.

Haldane A. G., 2012, On being the right size, speech given on 25th October.

International Monetary Fund, Global Financial Stability Report, October 2013.

Moody’s Investor Service, 2012a, Proposed Refinements to the Sovereign Bond Rating Methodology.

Moody’s Investor Service, 2012b, Moody’s Consolidated Global Bank Rating.

Morgan D. and Stiroh K., 2005, Too Big to Fail after All These Years, Federal Reserve Bank of New York Staff Reports.

Noss J. and Sowerbutts R., 2012, The implicit subsidy of banks, Financial Stability Paper No. 15, Bank of England.

Resti A. and Sironi A., 2005, The Basel Committee Approach To Risk-Weights And External Ratings: What Do We Learn From Bond Spreads?, Economic Working Paper nr. 548, Bank of Italy.

Schich S. and Lindh, 2012, Implicit Guarantees for Bank Debt: Where Do We Stand?, OECD Journal: Financial Market Trends, Volume 2012, Issue 1, downloadable at

 www.oecd.org/finance/financial-markets/Implicit-Guarantees-for-bank-debt.pdf

Schich S. and Kim B.H., 2012, Developments in the Value of Implicit Guarantees for Bank Debt: The Role of Resolution Regimes and Practices, OECD Journal: Financial Market Trends, Volume 2012, Issue 2, www.oecd.org/finance/Value_Implicit_Guarantees_Bank_Debt.pdf.

Ueda K., and Di Mauro B.W., 2012, Quantifying Structural Subsidy Values for Systemically Important Financial Institutions, IMF Working Paper, WP/12/128.

Appendix A: List of Banks

Country || Company Name, Short

AT || UniCredit Bank Austria AG

Erste Group Bank AG

Raiffeisen Bank International AG

Österreichische Volksbanken-AG

Bank für Arbeit und Wirtschaft und Österreichische Postsparkasse AG

BE || Belfius Banque SA

BNP Paribas Fortis SA

KBC Bank NV

ING Belgium SA/NV

BG || DSK Bank EAD

Raiffeisenbank (Bulgaria) EAD

CY || Cyprus Popular Bank Public Co. Ltd.

Bank of Cyprus Public Company Limited

Hellenic Bank Public Company Limited

CZ || Česká spořitelna, a.s.

Ceskoslovenska obchodni banka, a.s.

Komercní banka, a.s.

DK || Danske Bank A/S

Nordea Bank Danmark A/S

FI || Nordea Bank Finland Plc

Pohjola Bank Plc

Danske Bank Oyj

FR || BNP Paribas SA

Crédit Agricole SA

Crédit Agricole Corporate and Investment Bank

Natixis

Crédit Foncier de France SA

Société Générale SA

Banque Fédérative du Crédit Mutuel SA

Dexia Crédit Local SA

DE || Deutsche Bank AG

Commerzbank AG

Hypothekenbank Frankfurt AG

UniCredit Bank AG

LBBW Landesbank Baden-Württemberg

Deutsche Zentral-Genossenschaftsbank AG

Bayerische Landesbank

NORD/LB Norddeutsche Landesbank Girozentrale

Deutsche Postbank AG

HSH Nordbank AG

Landesbank Hessen-Thüringen Girozentrale

DekaBank Deutsche Girozentrale

Westdeutsche Genossenschafts-Zentralbank AG

GR || National Bank of Greece SA

Eurobank Ergasias SA

Alpha Bank AE

Piraeus Bank SA

HU || K&H Bank Zrt.

MKB Bank Zrt.

Erste Bank Hungary Zrt.

IE || Depfa ACS Bank

Bank of Ireland

Allied Irish Banks, Plc

Ulster Bank Limited

IT || UniCredit SpA

Intesa Sanpaolo SpA

Banca IMI SpA

Banca Monte dei Paschi di Siena SpA

Unione di Banche Italiane SCpA

Banco Popolare Società Cooperativa

LU || Banque Internationale à Luxembourg SA

BGL BNP Paribas SA

Banque et Caisse d'Epargne de l'Etat, Luxembourg

UniCredit Luxembourg SA

NL || ING Bank NV

Royal Bank of Scotland N

ABN AMRO Bank NV

Rabobank Nederland

PL || Powszechna Kasa Oszczędności Bank Polski SA

BRE Bank SA

ING Bank Śląski SA

Bank Zachodni WBK SA

Bank Millennium SA

Bank Handlowy w Warszawie SA

Bank BPH SA

PT || Caixa Geral de Depósitos SA

Banco Comercial Português SA

Banco Espírito Santo SA

Banco Santander Totta SA

Banco BPI SA

RO || Banca Comerciala Romana SA

Raiffeisen Bank SA

SK || Všeobecná úverová banka, a.s.

Tatra banka, a.s.

Ceskoslovenska Obchodni

SI || Nova Ljubljanska Banka d.d.

Nova Kreditna banka Maribor d.d.

Abanka Vipa d.d.

ES || Banco Santander SA

Banco Español de Crédito SA

Banco Bilbao Vizcaya Argentaria, SA

Bankia, SA

CaixaBank, SA

Banco Popular Español SA

Banco de Sabadell, SA

Bankinter SA

Ibercaja Banco SAU

SE || Nordea Bank AB

Skandinaviska Enskilda Banken AB

Svenska Handelsbanken AB

Swedbank AB

UK || Royal Bank of Scotland Plc

HSBC Bank Plc

Bank of Scotland Plc

Lloyds TSB Bank Plc

National Westminster Bank Plc

Santander UK Plc

Nationwide Building Society

Barclays Bank PLC

Standard Chartered Bank

Appendix B: Statistics on Moody’s rating data

Table B.1 to Table B.4 summarizes statistics on LT, BCA, BCAadj, Sov (country sovereign), and BCAadj-BCA (p) for each time period (the categories from C to AAA are translated into an index over the range 0-20).

Table B.1: Summary statistics for June 2011

Table B.2: Summary statistics for December 2011

Table B.3: Summary statistics June 2012

Table B.4. Summary statistics December 2012

Appendix C: Comparison with the existing literature

This Appendix attempts to compare summary statistics for the UPLIFT and UPLIFT* presented in the literature with the results obtained in Section 3. We are also comparing our implicit subsidy estimate with those available in the literature. We mainly focus on Schich and Lindh (2012), Haldane (2010, 2011, and 2012) and Bijlsma and Mocking (2013). Only a qualitative comparison can be performed since in the available studies the samples as well as the ratings employed differ.

C.1 Comparison of the UPLIFT and UPLIFT*

C.1.1 Comparison with: Schich and Lindh (2012)

Schich and Lindh (2012) employ a dataset of 118 large European banks for calculating the UPLIFT using Moody’s data. Their calculations are done on a yearly basis (from 2007 to 2011). The last trimester of 2012 is also taken into consideration.

Comparing the results from Schich and Lindh (2012) and Figure 2 one can conclude that the UPLIFT is essentially the same (in this study we have on average 0.5 notches higher UPLIFT for each time period). In more detail:

|| Average UPLIFT Schich and Lindh (2012) || Average UPLIFT (this study)

2007 || 2.21 || 2.69

2008 || 2.40 || 2.90

2009 || 3.14 || 3.73

2010 || 3.12 || 3.59

2011 || 2.18 || 2.63

2012 || 2.20 (Mar-2012) || 2.73 (Dec-2012)

In the same study the authors calculate the average UPLIFT* for each country employed in their sample for March-2012. For comparison purposes Figure C.1 presents the average UPLIFT* for each county for June 2012.

Figure C.1: Average UPLIFT* as of June 2012 from this study

Source: Moody’s and own calculations

Direct comparisons for the calculated UPLIFT* between Schich and Lindh (2012) and this study cannot be made as:

1) For most of the countries a different number of banks is considered.

2) The banks included in Schich and Lindh (2012) cannot be identified.

3) The authors of Schich and Lindh (2012)  calculate the UPLIFT* as of March-2012, while in Figure C.3 the UPLIFT* is calculated as of June-2012. For many banks in the sample, ratings changed between March-2012 and June-2012 (see Table C.1).

Table C.1: Comparison of UPLIFT* samples

Country[103] || UPLIFT* (this study) as of June 2012 || UPLIFT*from Schich and Lindh (2012) as of March 2012 || #Banks in this study || #Banks in Schich and Lindh (2012) || #of banks for which ratings changed after 31/03/2012

BE || 2.75 || ≈4 || 4 || 2 || 4

DK || 2.00 || ≈0.5 || 2 || 5 || 2

DE || 2.42 || ≈2.8 || 13 || 17 || 11

IE || 1.50 || ≈0.3 || 4 || 3 || 1

El || 1.00 || ≈1 || 4 || 5 || 0

ES || 0.78 || ≈1 || 9 || 10 || 9

FR || 3.10 || ≈2.8 || 10 || 7 || 9

IT || 1.60 || ≈1 || 6 || 13 || 4

LU || 2.50 || ≈4.5 || 4 || 2 || 3

NL || 3.00 || ≈1.4 || 4 || 8 || 3

AT || 2.60 || ≈3.7 || 5 || 6 || 4

PT || 1.00 || ≈0.4 || 5 || 6 || 2

FI || 2.33 || ≈1.9 || 2 || 2 || 3

SE || 3.00 || ≈1.9 || 4 || 6 || 2

UK || 2.11 || ≈1.5 || 9 || 14 || 6

Average || 2.11 || ≈1.91 || || ||

From the results presented in Table C.1 it can be concluded that for the countries for which a large number of banks is included in both samples (e.g. Germany, France and Spain) the average UPLIFT* is similar. Moreover, for Greece we observe that the average UPLIFT* is the same, while for Ireland, if we exclude one bank (which had an UPLIFT* equal to 5) we obtain the same UPLIFT*.  For the countries for which there is large difference in the number of banks (e.g. Denmark, Netherlands and Italy) no conclusions can be drawn.

C.1.2 Comparison with Haldane (2012)

There are two main issues when comparing our study with Haldane (2012):

1) Haldane (2012) reports the UPLIFT for 29 institutions deemed by the Financial Stability Board (FSB) to be the world’s most systemically-important. Thus non-EU banks are in the sample.

2) The ratings used in the present study and in Haldane (2012) are different. Haldane (2012) must rely on the Bank Financial Strength rating (BFSR), as UPLIFTs prior to 2011 are reported (the first announcement of the BCA, at least for the European banks, was in 2011). In contrast, this study relies on the BCA Moody’s rating and the rating scales of the two differ.

As discussed above, no direct comparisons can be made, as our sample includes only 13 EU G-SIBs banks. However, in order to have an intuition of the direction in which their average UPLIFT moves over time, we summarize in Table C.2 the evolution of the average UPLIFT.

Table C.2: Statistics of the sample used in the present study

|| UPLIFT

2011st || 2.62

2011nd || 2.08

2012st || 2.38

2012nd || 2.23

Comparing Table C.2 and the results presented in Haldane (2012) one can conclude that in both studies between 2011 and 2012 the UPLIFT for the G-SIBs is between 2 and 3 notches.

C.1.3 Comparison with Bijlsma and Mocking (2013)

Bijlsma and Mocking (2013) is based on a sample of 151 European banks which partly overlaps with the banks in our sample. As in Haldane (2012), Bijlsma and Mocking (2013) use the BFSR to calculate the UPLIFT. The authors conclude that banks enjoy an average UPLIFT of 2.9 for 2011 and 2.5 for 2012 respectively.

In Table C.3 summary statistics for the average UPLIFT calculated for the sample used in the present study are presented.

Table C.3: UPLIFT statistics (present study)

Dec-2011 || Dec-2012

Mean || 2.63 || Mean || 2.73

Minimum || 0 || Minimum || 0

Maximum || 8 || Maximum || 8

Count || 108 || Count || 112

Although averages for the UPLIFT points in time are similar, one must be cautions on making exact comparisons as the rating scales for BFSR and BCA differ.

C.2 Comparison of implicit subsidy estimates

One must be cautious when making a comparison of our results with the results presented in other studies on the implicit subsidy. Due to differences in definition of rating sensitive liabilities, the estimation of the yield spreads and the sample of banks the estimates will most likely not perfectly match. Schich and Lindh (2012), for example, report banks’ implicit subsidies for France, UK and Germany to be 1%, 0.4% and 1.4% of domestic GDP, respectively, for March 2012 (when using their upper bound estimates). Our estimates of the implicit subsidy for banks in France and UK are around 0.9-1.5% and 0.4% in 2012H2, which is in line with their results. For Germany we report an implicit subsidy of 0.5-0.7% (of domestic GDP) in 2012H2, which is half the Schich and Lindh (2012) estimate. Bijlsma and Mocking (2013) estimate the implicit subsidy to be in the range of 0.5-1.5%, 0.5-1.2% and 0.5-1.2% of domestic GDP for France, UK and Germany, respectively, in 2011. We estimate the implicit subsidy to be around 1-1.5%, 0.8% and 0.5-0.7% for the banks in these countries in 2011H2.

In Haldane (2010), the estimation of the average total annual subsidy of the five largest UK banks equals GBP 55 billion for the period 2007-2009. The proxy used in this study for rating-sensitive liabilities excludes retail deposits but includes wholesale borrowing. Haldane (2011) reports an average total annual subsidy of USD 74 billion for the period 2007-2010 for the four largest UK banks. In 2010, the total annual subsidy is estimated to USD 58 billion. The proxy for rating-sensitive liabilities is not explicitly defined.

Noss and Sowerbutts (2012) reports implicit subsidy estimates (with the ratings based approach) to around 5, 25-30, around 130, and around 40 GBP billion in 2007, 2008, 2009, and 2010, respectively.

In our study, we estimate the implicit subsidy for the four largest UK banks (RBS, HSBC, Lloyds and Barclays) to be around EUR 12 and EUR 7 billion in 2011 and 2012, respectively, when using the long-term funding proxy for rating-sensitive liabilities. Adding Santader UK to the others, we get EUR 13.6 and EUR 7.7 billion in 2011 and 2012, respectively. For the implicit subsidy estimates based on the total debt proxy, we do not have any data for Barclays. However for the other four banks, we estimate the implicit subsidy to be EUR 13.4 and EUR 8.4 billion in 2011 and 2012, respectively.

Appendix D: UPLIFT* for selected large EU groups

Source: SNL, Moody’s and own calculations

Note: BNP Paribas (BNP), Deutsche Bank AG (DB), Credit Agricole SA (CA), Royal Bank of Scotland (RBS), Banco Santander S.A.  (Ban.San), Societe Generale SA (SG), Lloyds TSB Bank (Ll), HSBC Bank PLC (HSBC), ING Bank NV (ING), UniCredit SpA (UC), Nordea Bank AB (Nord),  Bank of Scotland PLC (BS), Barclays (Bar), Standard Chartered (St.Ch.)

Appendix E: Descriptive statistics of the sample used for building the rating-yield map

The rating-yield curve is built by taking weekly observations of market yields for a sample of European banks from 18 countries. We have built one rating-yield curve for 2011H2 and one for 2012H2.

The data on bond yield spreads and long-term credit ratings are taken from DG ECFIN’s Bank Watch.[104]  The yield spreads are with respect to German 5 year bonds (Bobls).[105]  The maturity of the bank bonds may vary between 4.5 and 5.5 years.[106]

Descriptive statistics for the two semesters (2011H2 and 2012H2) are presented in Table E.1. For both semesters, the number of weeks is 22. The 2011H2 sample contains 74 banks, while the 2012H2 sample covers 80 banks. The total number of observations is 1590 and 1760 in 2011H2 and 2012H2, respectively.

The average yield spreads has fallen for each rating bucket in 2012H2 compared to 2011H2. In Table E.1, one can also observe that the majority of observations (covering 85% of the observations) in 2011H2 was in rating classes Aa2 to Baa2, while in 2012H2 it is Aa3 to Ba1 that have the majority of observations, illustrating that banks have been downgraded.

Table E.1. Descriptive statistics of sample for the rating - yield spread map

|| 2011H2 || 2012H2

Rating || Average yield spread || Average number of banks / week || Total number of observations || Average yield spread || Average number of banks / week || Total number of observations

Aaa || 66.65 || 1.0 || 22 || 7.9 || 1.0 || 22

Aa1 || 124.83 || 1.4 || 30 || 56.5 || 1.3 || 29

Aa2 || 184.20 || 11.2 || 246 || #N/A || 0.0 || 0

Aa3 || 238.33 || 14.0 || 309 || 79.4 || 5.2 || 115

A1 || 225.57 || 12.1 || 267 || 129.4 || 4.0 || 88

A2 || 370.34 || 13.0 || 287 || 145.8 || 14.5 || 318

A3 || 258.94 || 7.1 || 157 || 151.6 || 13.7 || 301

Baa1 || 343.27 || 0.9 || 20 || 231.8 || 4.0 || 88

Baa2 || 528.24 || 3.3 || 72 || 384.2 || 11.5 || 252

Baa3 || 526.34 || 1.9 || 42 || 389.6 || 9.9 || 218

Ba1 || 1098.61 || 2.9 || 64 || 801.3 || 5.7 || 126

Ba2 || 1061.65 || 1.5 || 32 || 500.9 || 2.4 || 53

Ba3 || 1222.25 || 0.5 || 10 || 544.4 || 3.9 || 85

B1 || #N/A || 0.0 || 0 || 698.1 || 1.1 || 25

B2 || #N/A || 0.0 || 0 || #N/A || 0.0 || 0

B3 || 1438.38 || 0.5 || 12 || #N/A || 0.0 || 0

Caa1 || #N/A || 0.0 || 0 || #N/A || 0.0 || 0

Caa2 || 2114.92 || 0.9 || 20 || 1506.7 || 1.8 || 40

Caa3 || #N/A || 0.0 || 0 || #N/A || 0.0 || 0

Ca || #N/A || 0.0 || 0 || #N/A || 0.0 || 0

C || #N/A || 0.0 || 0 || #N/A || 0.0 || 0

Total || || 72.3 || 1590 || || 80.0 || 1760

Appendix F: Description of the proxies for the rating-sensitive liabilities

The estimation of implicit subsidy is based on two alternative approximations of rating sensitive liabilities: Total Debt from SNL and Total Long-term Funding from Bankscope.

Estimate 1: Total Debt

Total Debt is part of the total financial liabilities, as reported in SNL. ‘Total Financial Liabilities’ consist of:

· Total Deposits: Total Deposits from Banks + Total Customer Deposits;

· Total Debt: Total Subordinated debt + Senior Debt;

· Derivative Liabilities[107]; and

· Other Financial Liabilities (incl. Securities Sold, not yet purchased).

In our sample, SNL reports Total Debt for 90 banks for both 2011 and 2012.

Estimate 2: Total Long-term Funding

Total long-term funding is part of the total funding, as reported in Bankscope. ‘Total Funding’ consists of:

· Total Deposits, Money Market and short-term funding = Total Customer Deposits + Deposits from Bank + Other Deposits and Short-term Borrowings

· Total Long-term funding = Senior Debts Maturing after one Year[108] + Subordinated Borrowing[109] + Other Funding;

· Derivatives[110]; and

· Trading Liabilities[111].

In our sample, Bankscope reports Total Long-term Funding for 96 banks for 2011 and for 97 banks for 2012.

Comparing the two approximations

In total, we have values for both Total Debt and Total Long-term Funding for 87 banks.

On average Total Long-term Funding represents 80% of Total Debt.[112] However, there is quite some variability in the sample and for several banks the ratio is above 100%. Based on this comparison we would expect the implicit subsidy estimate based on Total Long-term Funding to be lower than the estimate based on Total Debt.

Appendix G: Implicit subsidy versus pre-impairment operating profit

The aggregated pre-impairment operating profit by country is presented in Figure G.1. The total pre-impairment operating profit for the whole sample was EUR 197.5 billion and EUR 175.5 billion, in 2011 and 2012, respectively. The implicit subsidy in 2011 is estimated to be EUR 68 – 94 billion for the banks in our sample with profit data available in SNL. For 2012, the estimate is EUR 56 – 82 billion. So the implicit subsidy was 35-50% and 30-45% of pre-impairment operating profit in 2011 and 2012, respectively.

Figure G.1: Aggregated pre-impairment operating profit (EUR billion) country-by-country

Source: SNL and own calculations

Country-by-country the implicit subsidy as a percentage of pre-impairment operating profit is highest for Sweden, Germany, Belgium, France and UK, see Figure G.2 and Figure G.3. The numbers indicate that the implicit subsidy can be quite significant for the banking system. One should of course be careful when analysing these results and ideally the implicit subsidy as well as the profit should be estimated over a longer period of time.

Figure G.2: Aggregated implicit subsidy (% of Pre-impairment Operating Profit) based on Total Debt

Source: SNL and own calculations

Figure G.3: Aggregated implicit subsidy (% of Pre-impairment Operating Profit) based on Total Long-term funding

Source: SNL and own calculations

Appendix H: Implicit subsidy versus bank characteristics

The present Appendix summarizes the comparison of the relative size of the implicit subsidy with several balance sheet variables. We report here the graphs for the implicit subsidy estimate based on ‘Total Debt’ as reported by SNL for 2012. The general conclusions remain valid also for 2011 and the other implicit subsidy estimate.

From Figure 21 we have observed that there is a positive relation between the size of the bank (total assets) and the size of the implicit subsidy. This result comes from the fact that the larger the bank is the larger is the amount of debt on its balance sheet. Taking the size of the balance sheet into account, we rescale the implicit subsidy by total assets, to facilitate comparison between the banks in our sample.

Comparing the ratio of implicit subsidy to total assets with the size of the bank, the positive relation reported in Figure 21, is not that profound anymore (see Figure H.1).

There seems to be no or a week positive relation between the ratio of implicit subsidy to total assets and several risk measures (leverage, Tier 1 Capital (as % of total assets) and risk weighted assets (as % of total assets)), as reported in Figure H.2-H.4.

Finally, the share of trading (Assets Held for Trading as a % of Total Assets) seems not to influence the relative size of implicit subsidy (see Figure H.5).

Figure H.1: Implicit subsidy (as a % of Total Assets) versus Total Assets (end of 2012)

Figure H.2: Implicit subsidy (as a % of Total Assets) versus leverage (Total Assets over Total Equity) (end of 2012)

Figure H.3: Implicit subsidy (as a % of Total Assets) versus Tier 1 Capital (as a % of Total Assets) (end of 2012)

Figure H.4: Implicit subsidy (as a % of Total Assets) versus Risk Weighted Assets (as a % of Total Assets) (end of 2012)

Figure H.5: Implicit subsidy (as a % of Total Assets) versus Assets Held for Trading (end of 2012)

Appendix I: SNL business model variables description

· Total assets: All assets owned by the company as of the date indicated, as carried on the balance sheet and defined under the indicated accounting principles.

· Assets Held for Trading: Trading portfolio assets are: assets acquired principally for the purpose of selling in the near term, assets that on initial recognition are part of a portfolio of identified financial instruments that are managed together and for which there is evidence of a recent actual pattern of short-term profit-taking, or derivative assets (except for a derivative that is designated as an effective hedging instrument).

· Total Deposits from Customers: Amounts in customers' banking deposits; any accounts subject to federal banking deposit insurance, including any portions in jumbo deposits that are not insured but subject to the FDIC deposit regulations.

· Total Deposits from Banks: Total deposits from banks.

· Total Equity: Equity as defined under the indicated accounting principles. Includes par value, paid in capital, retained earnings, and other adjustments to equity. Minority interest may be included, per relevant accounting standards.

· Risk Weighted Assets: Total risk-adjusted assets as reported by the company. For European banks, this includes transitional capital adjustments when available.

· Tier1 Capital: For OTS-regulated institutions it represents the amount of core capital as defined under the latest OTS guidelines at period-end. For FDIC-regulated institutions it represents the amount of Tier 1 capital as defined by the latest regulatory agency guidelines.

· Tier2 Capital: Tier 2 eligible capital as defined by the bank's domestic central bank.

· Net Loans to Banks: Net loans and advances made to banks after deducting any allowance for impairment.

· Net Loans to Customers: Total loans to customers, net of reserves for loan losses.  Includes any loans held at amortised cost, available for sale, fair value through profit and loss and trading.  For U.S. GAAP companies, this is total loans and finance leases outstanding, including those held for sale.

· Total Financial Liabilities: Total interest accruing liabilities

· Operating Income: Total operating income from banking, insurance and asset management

· Net Interest Income: Interest income less interest expense before the provision for loan losses.

· Net Fee & Commission Income: Revenue from services to customers, net of expense from third parties related to services provided to the company.

· Derivative Liabilities: Total negative replacement values of hedging and non-hedging derivatives. A derivative is a financial instrument with all of the following three characteristics: its value changes in response to the change in an underlying variable; it requires no initial net investment or an initial net investment that is smaller than would be required for other contracts that would be expected to have a similar response to changes in market factors; it is settled at a future date. For European Insurers, this also includes liabilities held at fair value through profit and loss.

· Short Term Funding: Principal amount of debt payable within the next calendar year, including the current portion of long-term debt.

· Derivatives Assets: Derivatives Held for Trading (Derivatives with positive replacement values not identified as hedging or embedded derivatives. A derivative is a financial instrument with all of the following three characteristics: its value changes in response to the change in an Underlying variable; it requires no initial net investment or an initial net investment that is smaller than would be required for other contracts that would be expected to have a similar response to changes in market factors; it is settled at a future date.) and Derivatives Identified as Positive Hedges (Derivatives with positive replacement values accounted for under fair value hedge accounting, cash flow hedge accounting, or hedging of a net investment in a foreign operation. This includes fair value changes associated with hedging instruments.)

Appendix J: Marginal effects in Probit regression

Let AX represents the average bank characteristics with ℓ-th average characteristic denoted as AXℓ. We wish to measure the impact that a change of the ℓ-th characteristic by dAXℓ (or ΔAXℓ if the variable is discrete) has on the probability that the average bank receives an UPLIFT* equal to  j=0,…,m, the other variables being held constant. A first-order approximation around AX gives:

P(UPLIFT*=j|AX + dAXℓ) = P(UPLIFT*=j|AX)  +  dAXℓ [dP(UPLIFT*=j|AX)/dAXℓ]

where dP(UPLIFTi*=j|AX)/dAXℓ represents the derivative of the probability that the average bank UPLIFT* belongs to category j when the characteristic ℓ changes by an amount equal to dAXℓ (the so called marginal effect). In the Probit framework, this derivative verifies:

dP(UPLIFT*=0|AX)/dAXℓ         =   - βℓ Φ(μ1 - AX' β)

dP(UPLIFTi*=j|AX)/dAXℓ         =   - βℓ [Φ (μj+1- AX' β) - Φ (μj- AX' β)] , j  = 1,…,m-1

dP(UPLIFTi*=m|AX)/dAXℓ       =     βℓ Φ (μm - AX' β)

where βℓ is the coefficient associated to the ℓ-th characteristic and Φ is the standard normal density. The discrete case is trivially obtained. Hence, when the explanatory variables depart from average values, negative values of βℓ emphasize the lower tail and positive values the upper one. It can be verified that the marginal effects of dAXℓ on the probability to belong to each category j=0,…,m  sum to 0 over all categories, which does not occur under the linear probabilistic model.

[1]       For further details about the historical evolution of national banking systems, see e.g. D. Casserley et al (2010), “Should commercial and investment banking be separated: the historical background to the current debate”, McKinsey&Company; A.D. Morrison, “Universal Banking”, R. DeYoung “Banking in the United States”, and J. Goddard et al, “Banking in the European Union” in A.N. Berger et al, The Oxford Handbook of Banking (2010); and C.A.E. Goodhart (2013), “The optimal financial structure”, LSE Financial Markets Group Paper Series, Special paper No. 220, March 2013. Additional specific references are provided as appropriate.

[2]       Note that the German Banking Act contains a provision on the limitation of qualified holdings in non-financial enterprises for deposit-taking credit institutions (Section 12 of the German Banking Act (Kreditwesengesetz)).

[3]       For further detail, see National Bank of Belgium (2012), Interim report: Structural banking reforms in Belgium, June 2012.

[4]       Examples of other types of transactions include the sale of securities or other assets by a bank to an affiliate, payment of money or furnishing of services by a bank to an affiliate, transactions in which an affiliate acts as an agent or broker for a bank (or for any other person if the bank is a participant in the transaction), and any transaction by a bank with a third party if an affiliate has a financial interest in the third party or if an affiliate is a participant in the transaction.

[5]       This means that each covered transaction must be conducted on terms and under circumstances, “including credit standards, that are substantially the same, or at least as favourable to such bank or its subsidiary, as those prevailing at the time for comparable transactions with or involving other non-affiliated companies.”  If comparable transactions do not exist, then the transaction must be conducted on terms and under circumstances “that in good faith would be offered to, or would apply to, non-affiliated companies.”

[6]       Federal Reserve System (2002), Regulation W: Transactions between Member Banks and their Affiliates, 67 FR 76560, December 12, 2002.

[7]       See e.g. Omarova (2011), “From Gramm-Leach-Bliley to Dodd-Frank: the Unfulfilled promise of Section 23A of the Federal Reserve Act”, on which the above description builds heavily.

[8]       US Government Accountability Office (2011), “Proprietary trading: Regulators Will Need More Comprehensive Information to Fully Monitor Compliance with New Restrictions When Implemented”, GAO-11-529.

[9]       The GAO concluded that banks have conducted proprietary trading at stand-alone proprietary trading desks and may have done it elsewhere in the firm, given the difficulty to delineate proprietary trading and market-making (e.g. traders may accumulate positions in a particular assets at levels that exceed the amount of the firm’s typical or necessary inventory in that asset used to facilitate customer trades). As US banks do not maintain records on proprietary trading done elsewhere, the GAO study focused on the activities of stand-alone desks only.

[10]     Ibid, p. 23.

[11]     This included more than 18,400 ‘form letters’, i.e. letters signed by different associations or individuals that follow an identical template.

[12]     According to Fitch Ratings, the total notional value of derivatives for the four commercial banks that would be most affected – JPMorgan Chase, Bank of America, Citigroup and Wells Fargo – was USD193bn at the end of 2011. Of these 95% would be permissible derivatives and hence would not need to be pushed out.

[13]     H.R. 992, the Swaps Regulatory Improvement Act, as introduced by a bipartisan set of members of the House of Representatives. If enacted, it would effectively only push out those ‘structured finance swaps’, i.e. a swap or security-based swap based on an asset-backed security, that are not of credit quality and that are not used for hedging or risk management purposes.

[14]     S. 798, TBTF Act, introduced by Senators Brown (D) and Vitter (R) on 24 April 2013.

[15]     H.R. 5714, SAFE Banking Act of 2012, introduced by Congressman Miller (D) and Sherman (D) on 10 May 2012.

[16]     S. 1282, A bill to reduce risks to the financial system by limiting banks’ ability to engage in certain risky activities and limiting conflicts of  interest, to reinstate certain Glass-Steagall Act protections that were repealed by the Gramm-Leach-Bliley Act, and for other purposes, introduced by Senators Warren (D), McCain (R), and others on 11 July 2013.

[17]     HM Treasury (2012), Sound banking: delivering reform, October 2012.

[18]     Financial Services (Banking Reform) Bill, No. 130, 4 February 2013. An overview of the UK legislative process, including proposed amendments, can be found on the Parliament’s website: http://services.parliament.uk/bills/2013-14/financialservicesbankingreform.html

[19]     UK Government – White Paper on banking reform, June 2012.

[20]     UK Parliamentary Commission on Banking Standards, First and Second Report, December 2012 and February 2013.

[21]     Independent Commission on Banking (2011), Final report: recommendation, September 2011, p. 53.

[22]     PCBS (2012), First report, p. 69 onwards.

[23]     PCBS (2013), Second report, p. 5-12.

[24]     PCBS (2012), First report, p. 95.

[25]     PCBS (2013), Second report, p. 20, as well as Amendment R, p. 65.

[26]     However, the 1986 Building Societies Act already restricts the activities building societies can provide and the government has indicated that a forthcoming review will further bring the Act in line with the ring-fencing requirement.

[27]     In December 2011 the UK Government indicated that it would consider exempting non-EEA operations from the PLAC requirement if banks could show that those operations posed no risk to the EEA. It subsequently reversed the burden of proof, with such exposures being automatically exempt unless authorities can show that they pose a risk. This has been contested by the PCBS, arguing that banks should carry the burden of proof. This has in principle been accepted by the government and will be reflected in secondary legislation.

[28]     PCBS (2013), Second report, p. 28.

[29]     « Projet de loi de séparation et de régulation des activités bancaire », 19 December 2012.

[30]     Information about the debate and amendments can be found on the website of the Assemblée nationale: http://www.assemblee-nationale.fr/14/dossiers/separation_regulation_activites_bancaires.asp

[31]     Note that following discussions in the Assemblée nationale, the draft law features three new paragraphs on this distinction:

« L'Autorité de contrôle prudentiel et de résolution contrôle que la distinction de l'activité de tenue de marché, mentionnée aux 1° et 2°, par rapport aux autres activités est bien établie en se fondant, pour les activités mentionnées au 1°, notamment sur des indicateurs précisant les conditions de présence régulière sur le marché, l'activité minimale sur le marché, les exigences en termes d'écarts de cotation proposés et les règles d'organisation internes incluant des limites de risques. Les indicateurs sont adaptés en fonction du type d'instrument financier négocié et des lieux de négociation sur lesquels s'effectue l'activité de tenue de marché. Le teneur de marché fournit sur base régulière les indicateurs à l'Autorité de contrôle prudentiel et de résolution et à l'Autorité des marchés financiers.

« Pour les activités visées au 2°, l'établissement doit pouvoir justifier d'un lien entre le besoin des clients et les opérations réalisées pour compte propre. L'Autorité de contrôle prudentiel et de résolution apprécie cette activité au regard notamment de la fréquence des opérations réalisées.

« Un arrêté du ministre de l'économie, après avis de l'Autorité des marchés financiers et de l'Autorité de contrôle prudentiel et de résolution, fixe la liste des indicateurs visés au I du présent article. »

[32]     Ministère de l’Economie et des Finances (2012), Projet de loi de séparation et de régulation des activités bancaires : Étude d’Impact, p. 17.

[33]     See footnote 29.

[34]     For example, BNP Paribas is estimated to have less than 2 % of its corporate- and investment-banking revenues affected by the government plans. See Chevreux Daily B.A.N.K. Keynote on bank reform, 3 December 2012.

[35]     See e.g. BofA Merrill Lynch Global Research (2012), “French banks: French bank regulatory reform evolutionary not revolutionary”, December 2012.

[36]     The impact assessment accompanying the proposal indicates that the threshold will be set so as to capture the largest French banks. Accordingly four banks (BNPP, SG, CA, BPCE) with trading activities ranging from 20-40% of total assets are likely to be subject to the requirements.

[37]     Entwurf eines Gesetzes zur Abschirmung von Risiken und zur Planung der Sanierung und Abwicklung von Kreditinstituten und Finanzgruppen, 6 February 2013. The draft law can be found on the website of the Parliamentary Material Information System http://dipbt.bundestag.de/dip21/brd/2013/0094-13.pdf.

[38]     Information about the debate and amendments can be found on the website of the Parliamentary Material Information System http://dipbt.bundestag.de/extrakt/ba/WP17/508/50871.html.

[39]     Gesetz zur Abschirmung von Risiken und zur Planung der Sanierung und Abwicklung von Kreditinstituten und Finanzgruppen of 7 August 2013 (BGBl. I, page 3090).

[40]     I.e. the networks of credit cooperatives and savings banks.

[41]     According to the HLEG, these permitted activities would “include, but need not be limited to, lending to large as well as small and medium-sized companies; trade finance; consumer lending; mortgage lending; interbank lending; participation in loan syndications; plain vanilla securitisation for funding purposes; private wealth management and asset management; and, exposures to regulated money market (UCITS) funds.”

[42]     Throughout, figures in parentheses following the chosen option indicate how many positive responses this option received; not necessarily the number of individual stakeholders who would choose the option.  That is, respondents were not restricted to choosing a single preferred option for each question.  It is for this reason that the sum of the figures in the graphs below may equate to a number greater than the total number of responses received.

[43]     For an overview of these measures and their current state of play, see European Commission (2013), “Towards a stronger financial sector to support growth” (http://ec.europa.eu/internal_market/publications/docs/financial-reform-for-growth_en.pdf).

[44]     Directives 2009/111/EC and 2010/76/EC.

[45]     Proposals for a Directive on access to the activity of credit institutions and the prudential supervision of credit institutions and investment firms, and a Regulation on prudential requirements for credit institutions and investment firms, COM(2011)452 and 453 of 20 July 2012 – also referred to as the CRD IV package. Political agreement between the European Parliament and the Council was reached in April 2013.

[46]     Proposal for a Directive establishing a framework for the recovery and resolution of credit institutions and investment firms and amending Council Directives 77/91/EEC and 82/891/EC, Directives 2001/24/EC, 2002/47/EC, 2004/25/EC, 2005/56/EC, 2007/36/EC and 2011/35/EC and Regulation (EU) No 1093/2010, COM/2012/0280 final - 2012/0150 (COD). Political agreement was reached in July 2013.

[47]     The exact rules in some areas, notably liquidity, remain to be fully specified.

[48]     “The ratio of risk-weighted assets to total assets differs significantly between banks. It is remarkable that the banks with the highest amount of trading assets, notional derivatives, etc. (i.e. banks that are least "traditional") tend to have the lowest ratio.” Report of the HLEG, p. 43.

[49]     “[…] for a sample of 16 large EU banks, the capital requirements for market risks vary between close to 0% to just over 2% of the total value of trading assets, the average being close to 1%.” Report of the HLEG, p. 48.

[50]     E.g. the use of stressed VaR as part of Basel 2.5’s revisions to the market risk framework.

[51]     European Banking Authority (2013), “Interim results of the EBA review of the consistency of risk-weighted assets. Top-down assessment of the banking book”, February 2013; Basel Committee on Banking Supervision (2013), “Regulatory consistency assessment programme (RCAP) – Analysis of risk-weighted assets for market risk”, January 2013.

[52]     Ibid.

[53]     See Annex A5, “Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements.

[54]     Schich, S. and S. Lindh (2012), “Implicit Guarantees for Bank Debt: Where Do We Stand?”, OECD Journal: Financial Market Trends, were on the basis of empirical work unable to generate strong support for the hypothesis that the availability or introduction of special bank failure resolution regimes has been successful in reducing the incidence of implicit guarantees.

[55]     G20 (2013), G20 Leaders Declaration, St Petersburg, September 2013.

[56]     See e.g. FSB (2013), Progress and Next Steps Towards Ending “Too-Big-to-Fail” (TBTF), Report of the Financial Stability Board to the G-20, 30 August 2013, highlighting the potential of structural reform to put constraints on excessive risk-taking and contribute to improve resolvability at jurisdictional level. FSB also stresses the complexities in the legal, financial and operational structures of banks.

[57]     FSA (2011), “Recovery and resolution plans”, CP11/16.

[58]     According to the FSB, the extent of CCP clearing varies according to segment. For example, for interest rate swaps, where CCPs have existed since quite some time, estimates of activity by the G15 dealers indicate that, as of end-February 2013 around 50% of these dealers’ gross notional outstandings where CCPs active had been centrally cleared. For OTC credit derivative products the corresponding figure for all market participants was around 30% of the total notional outstandings. Central clearing of OTC commodity, equity and FX derivatives is yet to be well established at a global level. FSB (2013), OTC Derivatives Market Reform: Fifth Progress Report, April 2013.

[59]     Belgium, Germany, Estonia, Greece, Spain, France, Italy, Austria, Portugal, Slovenia and Slovakia

[60]     0.1% for shares and bonds, units of collective investment funds, money market instruments, repurchase agreements and securities lending agreements, and 0.01% for derivative products.

[61]     The FTT proposal contains two safeguards against relocation of financial transactions. The first is the "residence principle", where who is party to the transaction is what counts, not where it takes place. If a financial institution involved in the transaction is established in the FTT zone, or is acting on behalf of a party established in this zone, then the transaction will be taxed, regardless of where it takes place in the world. To further prevent avoidance of the tax, the Commission has added to this proposal the "issuance principle". This means that a transaction will also be taxed, whenever and wherever it takes place, if it involves financial instruments issued in one of the participating Member States.

[62]     A parallel with the leverage ratios of hedge funds vs. banks is interesting in that respect. Banks typically operate with a higher leverage than hedge funds, which is only tolerated by  investors due to the fact that the bank has a core of safe, stable deposit funding that is subject to explicit government guarantees.

[63]     As put by the Federal Reserve Bank of New York in a 2010 shadow banking paper: "Time and again, history shows that activities regulated out of banks, or financial innovations conducted and embodied by non-bank financial specialists (such as money-market mutual funds or finance companies) that pose a threat to the profitability of essential bank functions, are later acquired by banks – what's "regulated and innovated out" is usually "acquired back in". ("Shadow Banking", July 2010).

[64]     http://ec.europa.eu/internal_market/bank/docs/shadow/green-paper_en.pdf

[65]     G20 communiqué following the Cannes summit, November 2011.

[66] One should distinguish the systemic importance of an institution with its size. While many systemically important financial institutions (SIFIs) are systemically important by virtue of size, size is neither a sufficient nor a necessary conditions for a firm to be SIFI.

[67] Bankruptcy costs are very high for banks. Typically in bankruptcies the liabilities of the company are freezed while it business is operating in a way to maximize the resources available to creditors. However, a bank's liabilities are central to the value it creates and therefore this process is almost impossible for banks (see ICB (2011)).

[68] Such assets valued at fair value are likely to be extremely volatile in a crisis and can eventually wipe out the capital of the whole group. Securities business sharing capital with commercial banking creates a contagion risk as the highly volatile securities business shares capital with the less volatile commercial bank where cost amortisation accounting applies. They conclude that removing TBTF problems will not be credible if it is clear to investors that contagion risk is present.

[69] They also find that there is an increase in the yield spread of saving banks' bonds after the announcement of the court decision to remove guarantees (from 45 to 51 basis points) and savings banks adjusted their liabilities away from risk-sensitive debt instruments, and towards insured deposits and equity.

[70] See Moody's Investors service (2011) "Status Report on Systemic Support Incorporated in Moody's Bank Debt Ratings Globally", Special Comment, page 2

[71] For example Washington Mutual and Lehman Brothers with more than USD 300 billion and 600 billion assets respectively were not bailed out.

[72] For example Moody's (2011) explains the methodology underpinning the assessment of support which is based on the willingness to provide support, the capacity of a sovereign to do so and the workability of overall resolution regimes.

[73] They also find that both target and acquiring bank bond returns record a positive cumulative adjusted return (of 4.3 and 1.2% respectively).

[74] They find that non-bank financial s enjoyed a much higher benefit of 77 bps for the same period.

[75] The authors would like to thank S. Maes and M. Marchesi for useful comments and discussion.

[76] One must be cautious in interpreting the results of Schich and Lindh (2012) as they fit a linear probability model to numerically-translated ratings. Indeed, handling categorical variables as if they were continuous leads to the well-known flaws of the linear probability model such as heteroskedasticity and non-normality of residuals. These flaws may make inference invalid.

[77] Appendix A provides a list of the banks used

[78] Source: ECB statistics.

[79] For some G-SIBs no stand-alone ratings are available at group/holding level, thus the focus is on the main bank (e.g. Barclays and Standard Chartered are included in the sample as Barclays Bank PLC and Standard Chartered Bank).

[80] BFSR, which is only used to describe the evolution of the UPLIFT is mapped into 13 classes.

[81] It must be noted that when constructing the UPLIFT* variable the following simplified assumption is made: An UPLIFT* of 1 notch is quantitatively the same irrespective of the underlying adjusted BCA and stand-alone ratings.

[82] A detailed comparison of the obtained results with the existing literature can be found in Appendix C.

[83] Each observation of the two ratings is weighted according to the bank’s total assets.

[84] On average 72 and 80 banks per week in 2011 and 2012, respectively, due to missing values for some banks in some weeks.

[85] The very high yield spreads for the lower ratings might be driven by the fact that there are few observations of bonds for the low notches (see Appendix E). Therefore the possibility that our estimates are not accurate is greater for such observations. However, there is also a small number of banks with low notches in the sample of the banks used for the computation of the UPLIFT* and therefore this is unlikely to lead to very significant errors in the monetary quantification of the implicit subsidies over the whole sample.

[86] Points below the straight line represent banks for which the estimated yield spreads is lower than the average market yield spreads. Points above the straight line represent cases for which the estimated yield spreads are higher than the average market yield spreads. The difference between a bank’s average market yield spread and its estimated yield spread is the vertical distance between the point and the straight line.

[87] We have examined in more detail three cases where the market yield spreads are higher than the estimated yield spreads in Figure 15. Santander UK, for example, had an average market yield spread of 624 bp in 2012H2, which corresponds approximately to a Ba2 rating. The estimated yield spread based on its long-term rating (A2) is 142 bp, a difference of 482 bp compared to the average market spread. For Dexia Credit Local SA (FR) and Banco Popular Español SA (ES) the average yield spreads are 532 and 770 bp, respectively, higher than the estimated yield spreads in 2012H2. There are also cases where the average market yield spread is lower than the estimated yield spread. See, for example, National Bank of Greece SA in 2011H2, which had an average market yield spread of 1527 while the estimated yield spread is 2057.

[88] The weighted average funding cost advantage (weighted on bank’s total assets) is 171 and 149 basis points for 2011 and 2012 respectively. This corresponds to an increase in the WACC of 25-35 bps, since the proxies for credit sensitive liabilities represent 15% to 20% of total liabilities (see Section 4.3).

[89] In SNL, Total Debt is the sum of Total subordinated debt and Senior debt.

[90] In Bankscope, Total Long-term Funding is the sum of Senior Debts Maturing after one Year, Subordinated Borrowing and Other Funding.

[91] For each large banking group we have used the highest possible consolidation level, i.e., the group head if possible. Only if the group head does not have both the long-term issuer rating and the adjusted BCA, we have included the main subsidiary in the country where the group's headquarter is located. We have, for example, included HSBC Bank PLC (UK) and not HSBC Holdings PLC in the sample because the latter lacks an adjusted BCA. This will of course influence the amount of rating sensitive liabilities.

[92] For a large bank with relatively large subsidiaries in the country where it is headquartered we have chosen not to include these subsidiaries separately when calculating the total countrywide implicit subsidy even though we have the necessary information (ratings and balance sheets). This was done in the cases of Crédit Agricole CIB (subsidiary of Crédit Agricole SA), Hypothekenbank Frankfurt AG (Commerzbank AG), Deutsche Postbank (Deutsche Bank), Bank of Scotland (Lloyds TSB Bank), and National Westminster (RBS Plc).

[93] The use of consolidated balance sheet data might have a second caveat for cross-border banking groups. We aim at estimating the implicit subsidy due to an implicit guarantee from the sovereign where the reporting entity is headquartered, excluding any potential support from the parent bank. Consequently, for cross-border banking groups with bank subsidiaries outside their respective domestic country (i.e. where it is headquartered) the use of consolidated balance sheet data might lead to an overestimation of the implicit subsidy from the sovereign where the parent bank is headquartered.

[94] These changes in Portugal are driven by banks stand-alone and long-term ratings being downgraded, with a larger downgrade of the stand-alone rating, and, therefore, increasing the UPLIFT* from 0 to 1 notch to 0 to 4 notches. This increase in the UPLIFT*, together with the downgrade of the adjusted BCA, is translated into higher funding advantage.

[95] The sample used here is slightly smaller than the full sample, since some banks are not in SNL which implies that we do not have data for the pre-impairment profit. Most notably we have excluded Barclays Bank PLC and Standard Chartered Bank in the UK.

[96] Pre-impairment profit equals operating income less operating expenses. Operating income is the total operating income from banking, insurance and asset management. It is the sum of net interest income, net fee and commissions, net insurance income, net trading income and other operating incomes. Operating expenses is the total operating expenses from banking, insurance and asset management. It is mainly personnel expenses.

[97] These variables have been computed using SNL data. Appendix A gives a detailed description on the employed variables.

[98] The model prediction is computed as follows. Let us denote Xk the average of the explanatory variables of all banks for which UPLIFT*=k is observed and let P(Xk) be the probability that a bank has an UPLIFT* equal to k. Trivially P(Xk) corresponds to the empirical probability of the UPLIFT* in Table 4. The model prediction is calculated using the relationship:

P(UPLIFT* = j) = Σk=0m  P(Xk) P(UPLIFT* = j | Xk)

The model enters through the term P(UPLIFT* = j | Xk) which is measured as:

P(UPLIFTi*= 0|Xk)              =  Φ(μ1 - Xk' β)

P(UPLIFTi*= j|Xk)                               =  Φ(μj+1- Xi' β) - Φ(μj- Xi' β) , j  = 1,…,m-1

P(UPLIFTi*= m|Xk)             =  1- Φ (μm - Xi' β)

where the β and μ coefficients are given in Table 5.

[99] These figures are obtained from Table 7 as follows: the probability of having an UPLIFT* equal to 0 or 1 notch is 0.023 + 0.276 = 0.30 (third and fourth column). When increasing BCA by one notch, one should add to this quantity the marginal effect of BCA (third and fourth column): 0.30 + 0.021 + 0.112 = 0.43.

[100] Some country dummies were found to be significant at the 10% level, i.e. for AT, IE, NL, SI and UK. They leave unaltered the sign, the order of magnitude, and the significance of the coefficient estimates. More country dummies appear to be significant when the sovereign rating is excluded from the regression, but the coefficients of the other variables remain stable. Given the high correlation between country dummies and sovereign rating, we do not consider it worth adding them.

[101] The all-in rating, we remind, is determined together with the UPLIFT*.

[102] The source of data for the debt to GDP ratio is DG ECFIN AMECO database.

[103] In parentheses the respective number of banks

[104] The source of the Bank Watch data is Bloomberg.

[105] We have used yield spreads to isolate the credit risk component of each bank bond and to facilitate a comparison between bonds.

[106] A mixture of bonds is used to estimate the yields, mainly senior unsecured and covered bonds.

[107] Including both hedging and non-hedging derivatives.

[108] Item includes loans from banks; debts securities in issue, the liability component of convertible bonds, and other borrowed funds.

[109] Subordinated loans and debt including any dated hybrid instruments.

[110] Including both trading and hedging derivatives.

[111] Short positions, repos; short-term notes and other financial liabilities classified as held for trading.

[112] The ratio of Total Funding to Total Financial Liabilities is around 100% for all the banks.

LIST OF ANNEXES

ANNEX A5. Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements - PAGE 3

ANNEX A6. Qualitative assessment of benefits and costs of separating banking activities from deposit-taking entities - PAGE 53

ANNEX A7. Strength of separation - PAGE 88

ANNEX A8. Trading activities and bank structural separation: possible definitions and calibration of screening - PAGE 119

ANNEX A9. Summary of the main findings in literature on economies of scale and scope in the banking sector - PAGE 175

ANNEX A10. Quantitative estimation of a part of the costs and benefits of bank structural separation - PAGE 203

ANNEX A11. Impact on private costs – bank responses - PAGE 245

ANNEX A12. Economy-wide impact of structural separation - PAGE 249

ANNEX A13. Shadow banking – Securities finance transactions and transparency - PAGE 254

ANNEX A14. Glossary - PAGE 287

Annex A5 : Analysis of Possible Incentives towards Trading Activities implied by the Structure of Banks’s Minimum Capital Requirements

Forename(s) Surname(s)

M. Petracco-Giudici, M. Marchesi, A. Rossi, N. Ndacyayisenga, J. Cariboni

2013

European Commission

Joint Research Centre

Institute for the Protection and Security of the Citizen

Contact information

Marco Petracco Giudici

Address: Joint Research Centre, Via Enrico Fermi 2749, TP 361, 21027 Ispra (VA), Italy

E-mail: marco.petracco@jrc.ec.europa.eu

Tel.: +39 0332 78 3511

Fax: +39 0332 78 5733

http://ipsc.jrc.ec.europa.eu/

http://www.jrc.ec.europa.eu/

Legal Notice

Neither the European Commission nor any person acting on behalf of the Commission

is responsible for the use which might be made of this publication.

Europe Direct is a service to help you find answers to your questions about the European Union

Freephone number (*): 00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.

A great deal of additional information on the European Union is available on the Internet.

It can be accessed through the Europa server http://europa.eu/.

JRC88529

© European Union, 2013

Executive Summary

This report focuses on the analysis of structure of banks’ Minimum Capital Requirements (MCRs) within the EU. In particular, the aim is to investigate the possibility of the current regulatory framework making trading activities more attractive than traditional deposit taking bank activities (e.g. loans and credits), and more generally its implications for the adequacy of MCRs on the two activities.

Possible incentive toward trading activities and the effectiveness of MCRs are assessed by looking at a measure of returns per unit of MCR in each line of activity.

As publicly available data in the annual reports and in commercial databases only report the overall amount of MCRs but do not report their attribution to different activities, the estimate of the MCRs referring to each activity line are obtained via an econometric panel analysis.

The MCRs and income attributable to each of the two activity lines are estimated under various regulatory scenarios and the net income per unit of regulatory capital generated by the two activities over the period 2006-2011 are analysed.

Based on available data, and subject to the caveat that the division of assets and of risk weighted assets between trading and deposit taking activities is subject to a degree of uncertainty, results of this analysis show:

ü Possible existence of an incentive towards trading activities

Current regulation appears to provide incentives to banks to prefer trading activities to deposit taking activities due to differences in returns on minimum capital requirements on the two activities. Results indicate that even moving to Basel III MCRs could still not eliminate this effect.

ü MCR adequacy

Based on the definition of trading activities adopted, estimated MCRs for trading activities under current rules do not seem to allow absorbing net losses stemming from trading in crisis periods (2008), while estimated MCRs for deposit taking activities appear to allow covering net losses stemming from them more fully The analysis confirms that the introduction of Basel 2.5[1] rules in 2011 at least partially achieved the goal of substantially increasing the MCRs for some activities.

Table of Contents

Executive Summary.. 5

1..... Introduction and background.. 7

2..... The datasets for the panel regression and for the presentation of results.. 9

3..... Investigating the relationship between balance sheet composition and MCRs: estimation of risk weights via panel data analysis.. 11

4..... Determining the assets and income items to be assigned to each activity line.. 14

4.1.. Balance sheet split between activity lines and determination of the MCRs.. 14

4.2.. Income statement split between activity lines.. 17

5. Results.. 19

5.1 Total Assets, RWAs, RWA densities and Returns on Assets per line of activity.. 19

5.3 Income per unit of MCRs or Regulatory Capital per line of activity   26

6. Conclusions.. 31

APPENDIX A: Panel analysis to estimate risk weighted assets for deposit taking and trading activities.. 32

APPENDIX B: Estimation of the returns for the two business lines   34

APPENDIX C: Candidates sample.. 40

APPENDIX D: Analysis of the MCR structure for the full sample of banks   45

APPENDIX E: Correction of RWAs.. 50

APPENDIX F: ROA by line of activity and relationship with ROMCR.. 52

1. Introduction and background

The European Commission is putting forward a proposal to reform the structure of the EU banking sector which could result in separation of trading activities from deposit taking activities.[2] This analysis is being developed as part of the background material of the Impact Assessment for this legislative initiative.

In particular, the present note aims at assessing the presence of an implicit incentive toward trading activities embedded in current regulations[3] by estimating the difference in the returns per unit of Minimum Capital Requirements (MCR) or of Regulatory Capital for trading and deposit taking activities.

Since banks are not obliged to publish the allocation of their MCRs between these activities,[4] the estimation of the average amount of MCRs associated to categories of assets and liabilities in the balance sheet of banks is obtained via a three step procedure: first, an estimation of the average risk weights associated to different categories of assets and liabilities is performed via a panel data analysis.[5] Second, assets and liabilities in the balance sheet are allocated to the two activities; finally, the MCRs and Regulatory Capitals for each line of activity are obtained based on the allocation of assets and liabilities and the estimated average risk weights, under the current regulatory framework and a set of counterfactual scenarios (e.g. Basel III …).

Income for each line of activity is estimated by allocating income statement items to each of the two activities and by using, where necessary, a set of proportionality assumptions.

Finally, returns on capital requirements are obtained as the ratio of income to MCRs for each line of activity under each regulatory scenario in order to assess the possible presence of an incentive towards trading activities.[6]

As one the objectives of the analysis is to provide background material for the structural reform, results are presented separately for a sub-sample of 29 banks which might be candidates for structural separation (see next section and Annex A8 for details) and for the rest of the sample used for the analysis of risk weights and MCRs.

The remaining of the report is structured as follows: Section 2 describes the dataset used in the analysis; Section 3 illustrates the results of the panel analysis used to estimate the average risk weight of different balance sheet items; Section 4 describes the assumptions used to split the balance sheet and the income statement between trading and deposit taking activities and the regulatory scenarios used to calculate the MCRs referred to each activity; Section 5 presents the results of the analysis of the relationship between income and MCRs for the two lines for banks in the sample of banks potentially subject to structural reform and Section 6 concludes. Several technical annexes provide details on: the panel regression specification (Appendix A); the way income is assigned to trading and deposit taking activities (Appendix B) and the dataset (Appendix C). Appendix D presents the returns per unit of MCRs for the full sample of banks. Appendix E reports the assumptions made for the Basel 3 scenario and in particular details the corrections applied to RWAs and Regulatory Capital. Appendix F presents a comparison of estimated returns per unit of capital with balance sheet figures as well as a decomposition of the differences.

2. The datasets for the panel regression and for the presentation of results

The dataset for the panel analysis used to allocate RWAs between trading and deposit taking activities includes 255[7] banks located in the EU27 area having total assets available for 2011.[8] Their total assets as of end 2011 are 34’645 bn EUR for EU27[9] (77% of the EU27 banking sector[10]).

Results in the main text are presented for an “candidates sample” formed of 29 banks which are possible candidates for separation.[11],[12] The “candidates sample” also includes Norwegian DNB bank, with 274 bn EUR of total assets. All results are calculated for the “candidates sample” and for the rest of the EU27 sample used for the regression.[13]

Table 1: Distribution of the banks of the whole sample based on their average total assets over the 2006-2011 period. 500 bn EUR roughly corresponds to the 75th percentile of the total assets in the sample of banks considered by the European Banking Authority in its capital exercise.[14] 30 bn EUR is the size above which banks will be supervised directly by the SSM.

Banks’ Buckets || Banks’ size (Total Assets) || Average size of a bank in the bucket (average over years) – bn Eur || Number of banks in the sample || Percentage of banks in the sample || Total Assets in the sample (average over years) – bn Eur || Share of total assets in the sample (average over years) ||

Small || Up to 30 bn EUR || 9.8 || 152 || 59% || 1'477 || 4% ||

Medium || 30 to 500 bn EUR || 128.9 || 85 || 33% || 10'959 || 32% ||

Large || Over 500 bn EUR || 1'233.8 || 18 || 7% || 22'209 || 64% ||

Total || 135.8 || 255 || 100% || 34'645 || 100% ||

Source: SNL database and JRC estimates

NOTE: Large-size banks are located in seven countries: DE, FR, GB, IT, NL ES and SE.

3. Investigating the relationship between balance sheet composition and MCRs: estimation of risk weights via panel data analysis

Data on the allocation of MCRs by types of activities are not published by banks in the majority of cases.[15] We therefore need to establish a relationship between the nominal values of asset and liabilities attributable to different activities and the corresponding MCRs in order to proceed.

Since data on MCRs are not publicly available, while data on Risk Weighted Assets (RWAs) are, and since MCRs are mandated in regulation to be a fixed percentage of RWAs, we concentrate our attention on the latter quantity. Given (i) an allocation of nominal assets and liabilities to different activities, (ii) an average ratio of RWAs to nominal values for different balance sheet items and (iii) a fixed ratio of MCRs to RWAs; we will then be able to estimate MCRs for each line of activity.

In order to obtain the average ratio of RWAs to nominal amounts (i.e. the average risk weights or “RWA density” as termed in related literature[16]) for different balance sheet items a regression panel analysis is performed. Technical details of the analysis are illustrated in Appendix A, while only results which are used in the rest of this report are presented here.

Table 2 shows the results of the preferred regression model to predict RWAs based on nominal balance sheet data.[17] The values of the coefficients correspond to the equivalent weights of the various classes of assets to be used for the purposes of calculating total RWAs.[18]

The following classes of assets and liabilities available in SNL are included in the preferred model:[19]

1. Net loans to banks (LB)

2. Net loans to customers (NCL)

3. Activities held at amortized cost excluding loans (AMZ)

4. Securities held to maturity (HTM)

5. Available for sale assets excluding loans (AFS)

6. Assets held at fair value excluding loans (FV)

7. Securities held for trading excluding derivatives (volume in asset and liability sides) (TSA+TSL)/2

8. Derivatives held for trading (volume in asset and liability sides) (DA+DL)/2

9. Derivatives held for hedging purposes (volume in asset and liability sides) (DH).

Results show that net loans to customers (1), activities at amortized cost (3) and assets held at fair value (6) explain most of the RWAs (and hence of the MCRs).

Results are controlled for time effects and for individual bank effects, making use of robust estimation methods. The effects of the introduction of Basel 2.5 regulation were also tested by introducing time dummies for the period before 2011. The only significant change which can be detected based on available data is an increase in the coefficient for available for sale assets.

Table 2: Resulting coefficients as estimated under current regulation (Basel 2 till 2010, Basel 2.5 in 2011) for the panel analysis linking RWAs to balance sheet composition. All EU27 countries. [20]

|| LB || NCL || AMZ || HTM || AFS

|| Net loans to banks || Net loans to customers || Assets held at amortized cost || Sec. held to maturity || AFS assets[21]

|| (1) || (2) || (3) || (4) || (5)

RWA coefficient || .18 || .52 || .31 || .25[22] || .39

t-statistic[23] || 1.44 || 8.04*** || 2.98*** || 0.28 || 2.75***

Source: SNL database and JRC estimates

|| FV || (TSA+TSL) / 2 || (DA+DL) / 2 || DHV

continue || Assets held at fair value || Sec. held for trading || Derivatives held for trading || Derivatives held for hedging purposes

|| (6) || (7) || (8) || (9)

RWA coefficient || .23 || .198 || .048 || -1.95

t-statistic || 1.06 || 1.86** || 2.75*** || -3.64***

Note: 183 units[24], 923 observations. St. Errors adjusted for 183 clusters.

4. Determining the assets and income items to be assigned to each activity line 4.1 Balance sheet split between activity lines and determination of the MCRs

Having obtained the average weight or density of each balance sheet item, in order to obtain the MCRs corresponding to deposit taking and trading activities, it is necessary to (i) attribute each element of the balance sheet to one of the two lines of activity, (ii) predict the RWAs of each activity based on the weights above and then (iii) estimate the MCRs based on a regulatory scenario.[25]

The first step to estimate the MCRs is therefore to classify balance sheet items as belonging to one activity line or the other. As there is no clear-cut definition for these two lines of activities either in literature or in current regulatory practice, we propose the following methodology:

1) securities held for trading and derivatives held for trading, (i.e. classes (7) and (8) in Table 2) are classified as trading activities;

2) classes (1) to (6) are classified as commercial activities[26]

3) class (4) coefficient is not significant so this class is not considered (i.e. the coefficient is set to zero);

4) derivatives held for hedging purposes (class (9)) are allocated between the two lines of activities proportionally to total RWAs computed based on classes (1) to (8).

Based on this allocation, predicted RWAs are calculated for each year and for each bank according to the coefficients obtained in the econometric model.

Predicted RWAs are not however directly used in the rest of the analyses. Instead, predicted RWAs for each of the activities are re-normalized to sum to the RWAs as reported in the balance sheet.[27]

These predicted and re-normalized RWAs reflect an allocation of RWAs between activities under Basel 2 and Basel 2.5 rules. The ratio of predicted RWAs for Deposit Taking Banking activities to total predicted RWAs is hereafter termed βBasel II. The value (1- βBasel II) instead represents the ratio of predicted RWAs for the trading activity.

In addition to this re-normalization, RWAs can also be corrected to obtain counterfactual scenarios reproducing the impacts of introducing Basel III rules, which allows to better understand whether Basel III could contribute to shift any incentive towards trading activities which might be found under current regulation. These correction factors are obtained from the Basel III Quantitative Impact Study exercise conducted by EBA (see Appendix E for details).[28]

MCRs are obtained based on these normalized RWAs by multiplying them by the Capital Adequacy Ratio appropriate for each regulatory scenario (e.g. 8% for Basel II and 10.5% for Basel III).

Finally, as the definition of eligible capital will also be changing under Basel III, also MCRs can be corrected under some regulatory scenarios, to reflect that current rules admit as capital a set of instruments which will not be recognized as eligible capital under future regulation. In this way a uniform baseline is assured to compare incentives and capital adequacy across regulatory regimes. Also these correction factors are obtained from the EBA Basel III Quantitative Impact Study (see Appendix E for details).[29]

An overview of the regulatory scenario used in the rest of the analysis are presented in the following tables:

Table 3: Scenarios for assessing the effectiveness of minimum capital requirement allocating the minimum capital requirement to deposit taking or trading activity. α, βbasel II and βbasel III represent receptively the share of assets or RWA computed under Basel2 or Basel 3 scenario.

|| || MCR

Basel 2 scenario || Deposit taking (DTB) ||

Trading (TE) ||

Basel 3 scenario || Deposit taking (DTB) ||  

Trading (TE) ||

Basel 3 scenario || Deposit taking (DTB) ||  

Trading (TE) ||

Table 4: Scenarios for assessing the effectiveness of equity allocating the minimum capital requirement to deposit taking or trading activity. α, βbasel II and βbasel III represent receptively the share of assets or RWA computed under Basel2 or Basel 3 scenario.

|| || Equity (approximated by actual total regulatory capital, B3 def)

Basel 2 scenario || Deposit taking (DTB) ||

Trading (TE) ||

Basel 3 scenario || Deposit taking (DTB) ||

Trading (TE) ||

Basel 3 scenario || Deposit taking (DTB) ||

Trading (TE) ||

Table 5: Total assets in the various regulatory scenarios.

Activity || Regulatory scenario || Total Assets (TA)

Deposit taking (DTB) || Basel 2 || α TA

Deposit taking (DTB) || Basel 3 || α TA

Trading (TE) || Basel 2 || (1-α) TA

Trading (TE) || Basel 3 || (1-α) TA

Note: α represents the share of assets that are attributed to deposit taking activities. See details in Appendix B.

Source: SNL and JRC estimates.

4.2  Income statement split between activity lines

SNL income statement data are used to estimate the net income per line of activity. Based on the definitions of trading activities introduced in the previous section, each income statement item is allocated to each activity either fully or according to a set of proportionality rules, as fully detailed in Appendix B.

The following rules are used in the income allocation procedure:[30]

The revenue from all loans, securities Held To Maturity, Assets held at Fair Value, securities Available For Sale are allocated to deposit taking activities; Commissions on Loans and credit cards are allocated to deposit taking activities; The revenue attributed to securities and derivatives Held For Trading are completely allocated to trading activities; Taxes are not considered (i.e. pre-tax income is considered), as well as non-recurring expenses, non-recurring revenues and insurance incomes Some comprehensive income elements are included,[31] especially unrealized variations of value of Available For Sale securities (which are allocated to deposit taking activities together with realized gains, as detailed above) The part of the revenues or the expenses that are not linked to any of these two activities is split proportionally to the respective shares of assets or funding (see Appendix B for details on how funding costs are allocated).[32]

This set of rules allows us to completely allocate all gross recurring income excluding insurance income, as well as some comprehensive income elements, to the two lines of activity.

Missing data on sub-items of the income statement hierarchy were considered as zeros.

Appendix B offers a decomposition of the differences between the sum of the incomes allocated to the two activities and the net profit presented in the balance sheets.

Due to the impossibility of precisely allocating operational expenses and funding costs, this methodology could underestimate net incomes from deposit taking activities in favour of income from trading activities. The robustness of the core results has therefore been tested by repeating the analysis without considering operational expenses: this leads to a limited convergence of the two series.[33]

5. Results

5.1 Total Assets, RWAs, RWA densities and Returns on Assets per line of activity

Based on the definitions presented in the previous section and on the results of the panel analysis, Total Assets, Risk Weighted Assets and MCRs for trading and deposit taking activities can now be computed under each regulatory scenario.

Estimates of Total Assets, RWA densities (i.e. the ratio RWA/TA) and MCRs for the two lines of activities are obtained for each bank and each year between 2006 and 2011 and for each scenario.

Results in this section refer to both the “candidates sample” (termed the “Output sample” in the first parts of this section) for the part illustrating the allocation of assets and Risk Weighted Assets, and only to the “candidates sample” for the part looking at Returns on Minimum Capital Requirements.

Figures 1 and 2 provide a graphical representation on how the shares on total assets of the trading and deposit taking activities and of their associated RWAs change when moving from the Candidates sample (right plots) to Rest of the sample (left plots).[34] In the plots, the central line of the boxes are the median value, the edges are the 25th and 75th percentiles, the whiskers extend to the most extreme data points not considered outliers.[35]

Figure 1 shows that banks that are selected as candidate for structural separation in the calibration exercise tend to have a lower amount of deposit taking activities assets and/or RWAs. This figure also underlines the fact that the shares of RWAs (red boxplots) which are assigned to the deposit taking activity are always higher than the share of underlying assets which can be classified as “deposit taking activities” by the corresponding definition (blue boxplots).

Figure 1: Variation of the shares of assets or risk-weighted-assets for the deposit taking activity

Source: SNL database and JRC estimates

Figure 2: Variation of the shares of assets or risk-weighted-assets for the trading activity

Source: SNL database and JRC estimates

Figures 3 and 4 show how the dispersion of RWA densities differs for deposit taking and trading activities, and how it varies across years and how the implementation of Basel 3 should allow to lower this difference. These graphs again illustrate how the risk weighted assets allocated to trading per unit of assets are consistently lower than the risk weighted assets allocated to deposit taking activities’ per unit of assets.

We observe a decrease in the overall RWA density starting in 2006 for banks which get selected as candidates for separation in both deposit taking and trading activities. For banks that are candidate for separation the median RWA density of deposit taking activities lies between 40 and 50% for all years. For trading activities, the level of RWA on total assets remains low, under 20%, for almost all banks selected for structural separation and for all years.[36]

Figure 3: share of RWA on TA for trading activities, deposit taking activities and for the whole bank, RWA is computed based on Basel II definition

 

Source: SNL database and JRC estimates

Figure 4: share of RWA on TA for trading activities, deposit taking activities and for the whole bank, RWA is computed based on Basel 3 definition

Source: SNL database and JRC estimates

Figure 5 shows how Basel 3 regulation, represented by our scenario Basel 3, has allowed increasing the RWAs associated to trading activities relative to underlying assets. It is possible to observe the shift of scatterplot towards the x=y line, going from the Basel 2 scenario (left plot) to the Basel 3 scenario (right plot), which shows that RWA density of trading activities increased more than the RWA density of deposit taking activities: the cloud of dots (either black from the Candidates sample or blue dots from the rest of the sample) shifts to the right on the right figure.

Figure 5: Plot of the share of RWA over TA when changing the definition of RWA, which can be computed based on Basel II (left) or Basel III definition (right)

Source: SNL database and JRC estimates

Figure 6 presents the evolution of the ROA for trading and for deposit taking activities for the 29 banks of the “candidates sample”. [37] By observing the figure it seems possible to conclude that:

Deposit taking activities seem less profitable per unit of nominal value compared to trading, In 2008, when extreme losses were incurred, deposit taking activities performed similarly to trading activities for banks proposed for structural reform, However ROA on trading appears to be both much larger on average and much more volatile.

It should be kept in mind that, as the income split between trading and deposit taking activity doesn’t include non-recurring expenses, non-recurring revenues and insurance, the item ‘ROA for the whole bank’ and ‘ROA from Balance Sheet’ are different as the difference between the orange line (representing the whole bank, as the sum of both trading and deposit taking activity) and light blue line (ROA from Balance Sheet) show. A decomposition of the differences due to the definitions used to calculate the balance sheet income and the income used in this analysis is presented in Appendix B.

Figure 6: Average income / total assets for trading activity, deposit taking activity, for the whole bank (summing trading and deposit taking activity as estimated according to the methodology used for this report) and based on public data information for the 29 banks of Candidates sample. The average is weighted on banks' total assets.

Source: SNL database and JRC estimates

5.3 Income per unit of MCRs or Regulatory Capital per line of activity

To assess if existing regulation on MCRs encourages banks to prefer trading activities to credit activities, the Return On Minimum Capital Requirement (ROMCR) of the two classes are compared:

 vs

,

where t labels the various years.

Income for the two classes is estimated using data from the financial statements of banks reclassified according to the SNL template, as detailed in Appendix B. MCRs are estimated as detailed in sections 2 to 4. The comparison is performed under each regulatory scenario introduced in Table 3.[38] Results are always expressed as a weighted average, weighted over the total assets of the banks. Figure 7 and Figure 8 show the evolution over time of the average ROMCR for each activity under each scenario for the banks in the candidates sample.

From the figure it seems possible to observe that:

The recent and current regulatory framework seems to provide incentives to banks to prefer trading with respect to deposit taking activities, as the ratio of income per unit of MCR in trading seems always much higher than in deposit taking activities (the exception is 2008 which could be considered as a very critical year given the economic outlook). Under Basel 3 the difference would be reduced but it could still on average be not negligible. It can also be observed that, especially under the Basel II scenario, the average ratios for trading activities exhibit a much higher volatility than deposit taking activities.

For comparison purposes, a proxy of a Return on Equity measure is also calculated in Figure 9 and Figure 10. This measure is calculated by using actual balance sheet regulatory capital (eventually including a correction for changes in the definitions of capital when moving from Basel II to Basel III) as a proxy for equity, and allocating it to the two lines of business proportionally to estimated RWAs.[39]

Figure 7: Average income /MCR by type of activity and by regulatory scenario for the candidates sample of 29 banks. MCR refer to estimated minimum capital requirements. The average is weighted on banks' total assets. , while RegCap is a proxy for return on equity based on total balance sheet regulatory capital (corrected for changes in capital definition)

Source: SNL database and JRC estimates

Figure 8: Zoom on Figure 7: average income / MCR by type of activity and by regulatory scenario based on balance sheet information) for the Candidates sample. The average is weighted on banks' EU total assets.

Figure 9: Average income / Equity by type of activity and by regulatory scenario for the candidates sample of 29 banks. RegCap is a proxy for return on equity based on total balance sheet regulatory capital (corrected for changes in capital definition). . The average is weighted on banks' EU total assets.

Source: SNL database and JRC estimates

Figure 10: Zoom on figure 9: average income / Equity by type of activity and by 0regulatory scenario for the Candidates sample which corresponds to the 29 banks proposed for further investigation for structural reform. The average is weighted on banks' EU total assets.

6. Conclusions

Based on the available data, and subject to the caveat that the division of activities and of risk weighted assets between trading and deposit taking activities is subject to a degree of uncertainty, results of JRC preliminary analyses show for the banks that are part of the EBA list for capital exercise:

ü Possible existence of an incentive towards trading activities

Current regulation appears to provide incentives to banks to prefer trading activities to deposit taking activities due to differences in returns on minimum capital requirements on the two activities. Results indicate that even moving to Basel III MCRs could still not eliminate this effect.

ü MCR adequacy

Based on the definition of trading activities adopted, estimated MCRs for trading activities under current rules do not seem to allow absorbing net losses stemming from trading in crisis periods (2008), while estimated MCRs for deposit taking activities appear to allow covering net losses stemming from them more fully The analysis confirms that the introduction of Basel 2.5[40] rules in 2011 at least partially achieved the goal of substantially increasing the MCRs for some activities.

Appendix A: Panel analysis to estimate risk weighted assets for deposit taking and trading activities

Given both the cross-sectional and temporal dimensions of the problem we implement a set a panel regressions with the aim of identifying the model that best and parsimoniously describes the data.

Given both the cross-sectional and temporal dimensions of the problem we implement a set a panel regressions with the aim of identifying the determinants of Risk Weighted Asset (RWA).

RWA for a generic bank i at given time t is explained by the following set of covariates: Total Loans to Banks (LB), Net Customer Loans (NCL), Securities (excluding loans) held at amortized value (AMZ), Securities Held to Maturity (HTM), Available For Sale securities (excluding loans) (AFS), Securities held at Fair Value (FV), Securities (excluding loans and derivatives) Held For Trading as Assets or Liabilities (TSA+TSV)/2, Derivatives Held For Trading as Assets or Liabilities (TDV), Derivatives held for Hedging purposes as Assets or Liabilities (DHV) and an interaction dummy for the period before 2011 (B2dum) representing the use of Basel 2 rules, as opposed to Basel 2.5 rules from 2011. The sample covers a total of 215 European banks of various size and typology from 2006 to 2011. The panel is balanced. The type of regressions we use is panel data model of the form:

(1)

where the  represents the unobserved effect that is peculiar of the i-th bank, and are the error terms that we assume to be independent and identically distributed Gaussian random variables uncorrelated with the regressors at any lead and lag and across units. Model parameters are estimated by the fixed-effect estimator assuming that all regressors are uncorrelated with the errors and the fixed effects, i.e. . As there is some evidence of heteroskedasticity, we employ robust standard errors (cluster method) in the estimation.

Alternative regressions have been tested to check for the presence of effects of the introduction of Basel 2.5 rules on other coefficients, and to test for alternative variable definitions. The only coefficient that seems significant and shows the correct (i.e. expected on the basis of economic theory) is the coefficient for AFS, which is retained. In some of these models, we could note the presence of barely significant coefficients with positive sign on assets held at fair value and with negative sign on derivatives held for hedging, representing an indication that weights assigned to these categories in the risk management process could have been lowered (i.e. more negative for hedging) after the crisis, either due to asset reorganization or due to risk weights optimization. These results point in an interesting direction but the effects would need to be confirmed with further, more sophisticated analyses. It should also be noted that in some versions of the model also the coefficients for trading securities excluding loans and derivatives appears to increase after 2011, however in order to keep that coefficient as significant, it would be necessary to allow the post-2011 coefficients for derivatives to drop considerably and the coefficients on FVA to become not significant. We therefore repute these models not fully supported by both theory and data and discard them.

Alternative designs included considering the breakdown of loans by asset holding classification (i.e. at amortized cost, at fair value, for trading …), the use of net exposures in trading activities, with or without sign, and the inclusion of additional elements from the liability side. None of the models which could be justified based on economic grounds proved to be satisfactory in statistical terms.

All regressions include year dummies to control for fixed time effects. Models with interactions of single year time effects and the regressors were tested and rejected.

Results for the preferred model are reported in Table 6. The t-statistic is defined as the ratio between the coefficient and its standard error, the p-value is the probability of observing by chance a test statistic which is at least as extreme as the one observed, under the null-hypothesis that the coefficient is equal to zero.

Table 6: Estimates for the unrestricted fixed-effect model (1) - 50 parameters

|| || || || || || || || || ||

Coeff. || .18 || .52 || .31 || .25 || .39 || .-19 || .23 || 0.198 || 0.048 || -1.95

t-stat || 1.44 || 8.04 || 2.96 || 0.28 || 2.75 || -2.52 || 1.06 || 1.86 || 2.75 || -3.64

Number of observations: 923 in 183 groups Std. Error adjusted for 183 clusters

R-squared: within = 0.72 Between = 0.96 Overall = 0.95 || Corr(ui, Xb) = 0.76

Rho (fraction of variance due to ui) = .91

Notice that the effect of HTM is completely not significant according to the t-statistic.

The magnitude of fixed time effects is extremely small compared to fitted values. Individual fixed effects are very small to small in the vast majority of cases.

The model has an extremely high R-squared, this is in part to be expected due to the fact that we know that the dependent variable is obtained based on calculations including the regressors or some of their combinations.

It should also be noted that the model is not intended to check the relationship between the risk weights and underlying risk factors, but has the more limited aim of predicting (in-sample) the value of RWA based on the composition of the balance sheet.

Appendix B: Estimation of the returns for the two business lines

This appendix reports the criteria used to allocate the revenue between deposit taking and trading activities.

The allocation of the revenue is based on income statement public data as reported in the commercial database SNL[41], based on template data.

The definition of income used in this report differs from net profits as reported in the SNL balance sheet template in the following ways:[42]

1. Income is based on pre-tax income definition

2. non-recurring expenses and nor non-recurring revenues are excluded

3. insurance incomes are excluded

4. unrealized gains and losses (an element of comprehensive income) are included

The main criteria used in the allocation are that:

1. the revenues from loans and HTM securities are always attributed to the deposit taking activity.

2. Securities held at fair value and securities available for sale are always counted to the deposit taking activities income as well.[43]

3. Trading revenue of the bank mainly comes from securities and derivatives held for trading.

4. The part of the revenue or expense that is not directly attributed to one of these activities is proportionally attributed based on the allocation of assets (for income) or of liabilities (for expenses) of each type of activity (trading or deposit taking). Applying this proportionality to net-interest income and to the fees and commissions, we implicitly assume that the interest rate expense and revenue on deposit taking or trading activities’ assets are equivalent. [44]

Table 7 below illustrates the procedure in more detail

Table 7: Split of net income between trading and deposit taking activities based on SNL classification for definition 1 and 2. Income not directly assigned to one type of activity is split based on the share of assets (α) or liabilities (Φ) assigned to each activity.

Level || Income Statement || Rule for repartition between activity

1 Operating income

1.1 || Net interest income ||

1.1.a || Interest income ||

1.1.a1 || Interest Earned on Customer Loans || Comm. Bk

1.1.a2 || Interest Earned on Trading Assets || Trading

1.1.a3 || Rest of Interest Income[45] = 1.1.a – 1.1.a1 – 1.1.a2 || α

1.1.b || Interest Expenses ||

1.1.b1 || Interest Expenses on Customer Deposits || Comm. Bk

1.1.b2 || Interest Expenses on Trading Assets || Trading

1.1.b3 || Rest of Interest Exp. = 1.1.b – 1.1.b1 – 1.1.b2 || Φ

1.2 || Net fee & commission income ||

1.2.a || Deposits and Loans fees || Comm. Bk

1.2.b || Credit card income || Comm. Bk

1.2.c || Investment banking fee || Trading

1.2.d || Rest of Net fee & Com.= 1.2 – 1.2.a – 1.2.b – 1.2.c || α

1.3 || Net trading income ||

1.3.a || Net gain on securities held for trading || Trading

1.3.b || Net gain on securities held at fair value || Comm. Bk

1.3.c || Realised gain on securities || Comm. Bk

1.3.d || Other net gain on securities || α

1.4 || Other operating income[46] || α

2 Operating expense || α

3 || Assets write-downs ||

3.1 || Loans and credit commitment impairments || Comm. Bk

Level || Income Statement || Rule for repartition between activity

3.2 || Impairment on securities ||

3.2.a || AFS financial assets impairment || Comm. Bk

3.2.b || Held to maturity financial assets imp. || Comm. Bk

3.2.c || Other financial assets impairment || α

3.3 || Impairment on non-financial assets || α

4 || Total other comprehensive income ||

4.1 || Change in Unrealised Gain || Comm. Bk

4.2 || Change in FV of Effective Hedge || α

4.3 || Change in Foreign currecny ex || α

4.4 || Other comprehensive income || α

The revenue of deposit taking activities has thus been estimated as:

Interest Earned on Customer Loans                                                               (1)

- Interest Expenses on Customer Deposits

+Deposits and Loans fees

+Credit card income

+ Net gain on securities held at fair value

+Realised gain on securities

- Loans and credit commitment impairments

- AFS financial assets impairment - held-to-maturity financial assets imp.

+ change in unrealized gain

-Φ *Sum of expenses to be split along phi as defined in eq. (3)

+α *Sum of income to be split along alpha as defined in eq. (6)

 where

α is the proportion of the retail assets over total assets as detailed in equation (4) ,

Φ is the proportion of the retail funding over total liabilities as detailed in equation (7)

Income issued from AFS and held to maturity securities are always completely in the income of  deposit taking activities.

The revenue of trading activities has been estimated as:

Interest Earned on Trading Assets                                                     (2)

- Interest Expenses on Trading Assets

+ Net gain on securities held for trading

- (1-Φ) *Sum of expenses to be split along phi as defined in eq. (3)

            + (1-α) *Sum of income to be split along alpha as defined in eq. (6)

 where

(1-α) is the proportion of the trading assets over total assets as detailed in equation (4),

(1-Φ) is the proportion of the trading funding over total liabilities as detailed in equation (7)

The net gain issued from securities held for trading is completely included as trading income.

All remaining voices correspond to income that shall be distributed between trading and deposit taking activities based on the repartition of the underlying assets, RWA or liabilities.

The remaining voice linked to expenses will be split between trading and deposit taking activities based on the repartition of the related funding Φ. It includes only

Interest Expenses excluding the part already computed in eq(1) and (2).                             (3)

These expenses are split between deposit taking (Φ) or trading activities (1- Φ) based on the share of funding / liabilities repartition between these two activities as followed.

γ corresponds to the part of liabilities assigned to deposit taking activities and equals to

                                       a * total assets – customer deposits                                        .     

total debt + banks deposits + non-financial liabilities+ derivitatives id. as +hedges+ equity            (4)

We can compute the share of funding Φ for deposit taking activities

Φ = (cust deposits + γ (debt + bank deposits + derivatives id. as negative hedges)

 financial liabilities                                                      (5)

The remaining voices not linked to expenses will be split between trading and deposit taking activities based on the repartition of the related assetsa. They include:

Interest Income excluding Interest Earned on Customer Loans or on Trading Assets          (6)

+ Net fee & commission income not yet considered in (1) or (2)

- Operating expense

-Other financial assets impairment - Impairment on non-financial assets

+Change in FV of Effective Hedge

+Change in Foreign currency ex

+Other comprehensive income

where

a = net loans to customers/banks + sec. held at maturity or at fair value+ amortized cost sec.

            financial assets- cash at central bank – derivatives identified as positive hedges      (7)

It should be noted that the implementation of some of the formulas that use more detail from balance sheet present problems due to data missigness even for some large banks. Missing data are imputed as zeros for the purposes of the calculations presented here.

In order to make comparison with performance indicators from the balance sheet such as ROAE or ROAA, figure 11 presents a decomposition of the differences between Net profit as presented in the balance sheet and the sum of the incomes allocated to trading and deposit taking activities used for the purposes of this study. In this decomposition, items which are are not included in net profit (as defined in the SNL template) but are included in our custom definition are termed “extra income items”; items which are included in the calculation of net profit, but are ignored for the purposes of our definition are termed “missing income items”.

Also they have been no imputation of missing values, the respective amount of income is not considered if the value is missing. The voices of income that we have used are sometimes deep in the income hierarchy detail and thus for some bank some missing values can be observed. In this case, we assume missing values are 0.

Figure 11: Representation of the difference between the sum of net profit summing TE and CBT income and net profit for 2011 only from the income statement based on the 29 banks of our sample. Incomes are profited in Bio €.  Area in Blue represent positive value and in orange are provided negative values.

In absolute value over the sample, the net income diminished from 3’477 Bio EUR which correspond to 8% of the summed income from TE and CBT. This is an overall figure and some individual banks can present larger differences.

Appendix C: Candidates sample[47]

This appendix describes the Candidates sample: the 29 banks have been selected as potential candidates for structural reform as presented in the Commission Staff paper “Trading activities and functional structural separation: possible definitions and calibration of a de minimis exemption rule”[48]   

The list of banks is presented in the following table.

Table 8: List of banks considered in the Output Sample sample used for this section.

Deutsche Bank || Groupe BPCE || LB Baden-Württemberg || Swedbank

HSBC Holdings || ING Bank || Bayerische LB || Portigon

BNP Paribas || UniCredit || KBC Group || HELABA

Crédit Agricole Group || Nordea || Handelsbanken || DekaBank

Barclays || Commerzbank || DNB || Mediobanca

RBS Group || Danske Bank || SEB ||

Santander || Standard Chartered || Monte dei Paschi Siena ||

Société Générale || DZ-bank || Belfius ||

Table 9: Summary by bank of the allocation of assets between trading (TE) and commercial banking (DTB) activities under Basel II or Basel III scenario for 2011

Institution Name || || Share Business in EU || Total EU Assets (€000) || RWA / TA || a = TADTB/ TA || GrpQIS[49] || RWADTB / RWA (B2) || RWADTB/ RWA(B3) || RWADTB(B2) / TADTB || RWATE(B2)/ TATE || RWADTB/ TADTB (B3) || RWATE/ TATE (B3)

Deutsche Bank || DE || 79.31% ||  1,716,350 || 18% || 45% || 1 || 83% || 73% || 33% || 5% || 35% || 10%

HSBC Holdings || GB || 41.05% ||  807,780 || 47% || 77% || 1 || 92% || 83% || 57% || 16% || 62% || 42%

BNP Paribas || FR || 87.57% ||  1,720,998 || 31% || 56% || 1 || 85% || 75% || 47% || 11% || 51% || 21%

Crédit Agricole Group || FR || 97.74% ||  1,837,058 || 28% || 73% || 1 || 94% || 84% || 36% || 6% || 39% || 20%

Barclays || GB || 56.34% ||  1,053,412 || 25% || 52% || 1 || 88% || 78% || 42% || 6% || 45% || 14%

RBS Group || GB || 77.96% ||  1,406,125 || 29% || 53% || 1 || 86% || 76% || 47% || 9% || 50% || 18%

Santander || ES || 77.81% ||  973,410 || 45% || 85% || 1 || 97% || 87% || 52% || 10% || 56% || 46%

Société Générale || FR || 90.11% ||  1,064,534 || 30% || 62% || 1 || 87% || 78% || 42% || 10% || 45% || 21%

Groupe BPCE || FR || 97.10% ||  1,105,382 || 34% || 85% || 1 || 97% || 87% || 39% || 7% || 42% || 36%

ING Bank || NL || 85.38% ||  820,643 || 34% || 90% || 1 || 97% || 88% || 37% || 9% || 41% || 50%

UniCredit || IT || 93.54% ||  854,551 || 50% || 86% || 1 || 97% || 88% || 57% || 9% || 62% || 54%

Nordea || SE || 87.70% ||  628,111 || 26% || 67% || 1 || 92% || 82% || 36% || 6% || 39% || 17%

Commerzbank || DE || 97% ||  642,506 || 36% || 75% || 1 || 95% || 85% || 45% || 7% || 49% || 26%

Danske Bank || DK || 99% ||  458,252 || 26% || 71% || 1 || 93% || 83% || 35% || 6% || 38% || 19%

Standard Chartered || GB || 100% ||  456,368 || 46% || 86% || 1 || 97% || 88% || 52% || 9% || 57% || 47%

DZ-bank || DE || 100% ||  405,926 || 25% || 84% || 1 || NA || NA || NA || NA || NA || NA

LB Baden-Württemb. || DE || 100% ||  373,069 || 29% || 71% || 1 || 90% || 81% || 37% || 10% || 40% || 23%

Bayerische LB || DE || 100% ||  309,172 || 38% || 84% || 1 || 97% || 88% || 45% || 6% || 49% || 35%

Institution Name || || Share Business in EU || Total EU Assets (€000) || RWA / TA || a = TADTB/ TA || GrpQIS[50] || RWADTB / RWA (B2) || RWADTB/ RWA(B3) || RWADTB(B2) / TADTB || RWATE(B2)/ TATE || RWADTB/ TADTB (B3) || RWATE/ TATE (B3)

KBC Group || BE || 100% ||  285,382 || 44% || 91% || 1 || 98% || 89% || 48% || 10% || 52% || 67%

Handelsbanken || SE || 100% ||  275,514 || 21% || 92% || 1 || 98% || 89% || 22% || 4% || 24% || 34%

DNB || NO || 100% ||  274,216 || 50% || 93% || 1 || 93% || 85% || 50% || 48% || 55% || 128%

SEB || SE || 100% ||  264,852 || 29% || 80% || 1 || 95% || 85% || 34% || 7% || 37% || 26%

Monte Paschi di Siena || IT || 100% ||  240,794 || 44% || 89% || 1 || 97% || 88% || 48% || 11% || 52% || 57%

Belfius || BE || 100% ||  232,509 || 23% || 84% || 1 || 98% || 88% || 26% || 3% || 29% || 21%

Swedbank || SE || 100% ||  208,464 || 27% || 86% || 1 || 96% || 87% || 30% || 7% || 33% || 30%

Portigon || DE || 100% ||  167,910 || 29% || 65% || 1 || 94% || 83% || 42% || 5% || 45% || 16%

HELABA || DE || 100% ||  163,985 || 35% || 75% || 1 || 90% || 81% || 42% || 15% || 45% || 33%

DekaBank || DE || 100% ||  133,738 || 19% || 70% || 2 || 89% || 87% || 24% || 7% || 25% || 8%

Mediobanca || IT || 100% ||  72,934 || 75% || 79% || 1 || 95% || 86% || 91% || 17% || 99% || 63%

Source : SNL and JRC estimates

Table 10: Summary by bank of the return on assets (ROA) or return on MCR (ROMCR) obtained after the split between trading (TE) and commercial banking (DTB) entities for 2011

|| Income || ROA || ROMCR

Institution Name || Trading Income || Banking Income || DTB || TE || DTB + TE || ROAA (BalSht) || TE, B2 || TE, B3 || DTB, B2 || DTB, B3

Deutsche Bank ||  7,432 || -275 || 0.62% || -0.03% || 0.33% || 0.21% || 218.2% || 57.2% || -1.7% || -0.8%

HSBC Holdings ||  10,672 ||  5,686 || 2.33% || 0.38% || 0.83% || 0.65% || 286.6% || 53.3% || 12.6% || 5.8%

BNP Paribas || -138 ||  7,364 || -0.02% || 0.67% || 0.37% || 0.35% || -2.8% || -0.7% || 26.9% || 12.5%

Crédit Agricole Group ||  1,119 || -1,827 || 0.22% || -0.13% || -0.04% || 0.06% || 66.3% || 10.6% || -7.1% || -3.3%

Barclays ||  2,878 || -9,811 || 0.32% || -1.01% || -0.37% || 0.25% || 95.9% || 21.6% || -45.5% || -21.2%

RBS Group ||  5,038 || -4,099 || 0.60% || -0.43% || 0.05% || -0.13% || 129.1% || 31.6% || -17.3% || -8.0%

Santander ||  4,037 ||  546 || 2.13% || 0.05% || 0.37% || 0.50% || 408.6% || 43.7% || 1.9% || 0.9%

Société Générale ||  3,878 ||  597 || 0.87% || 0.08% || 0.38% || 0.24% || 168.8% || 39.8% || 3.7% || 1.7%

Groupe BPCE ||  2,532 ||  1,286 || 1.51% || 0.13% || 0.34% || 0.28% || 388.2% || 40.4% || 6.5% || 3.0%

ING Bank || -23,789 ||  27,652 || -24.74% || 3.20% || 0.40% || 0.43% || -5014.3% || -471.8% || 163.8% || 74.7%

UniCredit ||  148 || -10,026 || 0.12% || -1.27% || -1.08% || -0.96% || 24.5% || 2.1% || -42.5% || -19.5%

Nordea || -1,254 ||  3,541 || -0.53% || 0.74% || 0.32% || 0.43% || -166.4% || -30.1% || 39.5% || 18.2%

Commerzbank ||  2,304 || -2,546 || 1.42% || -0.51% || -0.04% || 0.11% || 380.2% || 52.0% || -21.5% || -9.9%

Danske Bank ||  1,120 || -888 || 0.84% || -0.27% || 0.05% || 0.05% || 252.1% || 43.2% || -14.9% || -6.9%

Standard Chartered ||  1,740 ||  1,903 || 2.66% || 0.49% || 0.80% || 0.87% || 574.7% || 53.3% || 17.9% || 8.2%

DZ-bank || -237 || -308 || -0.36% || -0.09% || -0.13% || 0.15% || -4.5% || -2.0% || || -33.0%

LB Baden-Württemb. ||  1,308 || -1,514 || 1.21% || -0.57% || -0.06% || 0.02% || 225.1% || 49.0% || -29.8% || -13.7%

Bayerische LB ||  563 ||  4 || 1.10% || 0.00% || 0.18% || 0.03% || 334.5% || 30.1% || 0.1% || 0.0%

KBC Group ||  53 ||  2 || 0.20% || 0.00% || 0.02% || 0.02% || 37.2% || 2.9% || 0.0% || 0.0%

Handelsbanken || -226 ||  1,904 || -1.02% || 0.75% || 0.61% || 0.53% || -502.4% || -28.7% || 64.5% || 29.4%

|| Income || ROA || ROMCR

Institution Name || Trading Income || Banking Income || DTB || TE || DTB + TE || ROAA (BalSht) || TE, B2 || TE, B3 || DTB, B2 || DTB, B3

DNB ||  505 ||  1,799 || 2.55% || 0.71% || 0.84% || 0.60% || 100.8% || 18.9% || 26.9% || 12.2%

SEB ||  498 ||  970 || 0.95% || 0.46% || 0.55% || 0.49% || 240.9% || 35.1% || 25.6% || 11.7%

Monte Paschi di Siena ||  764 ||  746 || 2.82% || 0.35% || 0.63% || -1.95% || 483.1% || 47.0% || 13.9% || 6.4%

Belfius ||  126 || -1,108 || 0.35% || -0.56% || -0.42% || -0.57% || 193.5% || 15.5% || -40.8% || -18.7%

Swedbank ||  403 ||  1,194 || 1.34% || 0.67% || 0.77% || 0.66% || 351.2% || 43.2% || 42.8% || 19.6%

Portigon ||  958 || -350 || 1.62% || -0.32% || 0.36% || -0.03% || 591.2% || 93.5% || -14.8% || -6.8%

HELABA ||  498 ||  13 || 1.24% || 0.01% || 0.31% || 0.24% || 162.1% || 35.7% || 0.5% || 0.2%

DekaBank ||  587 || -177 || 1.46% || -0.19% || 0.31% || 0.20% || 351.2% || 164.3% || -13.4% || -7.2%

Mediobanca ||  121 ||  386 || 0.81% || 0.67% || 0.70% || 0.49% || 90.1% || 12.1% || 14.0% || 6.4%

Source : SNL and JRC estimates  - extreme values obtained for ROMCR on trading ING for 2006-2011 and Belfius for 2006-2010 made these banks dropped from the figures of this document.

Appendix D: Analysis of the MCR structure for the full sample of banks

The complete sample used for panel regression contains 215 banks. The number of banks for which data are available for the computation of the ROMCR is much lower and varies across the years, as detailed in Table 11.

Table 11: Evolution of the sample size depending on data availability

Year || 2006 || 2007 || 2008 || 2009 || 2010 || 2011

Number of banks used in the figures || 144 || 183 || 187 || 196 || 204 || 209

The below figures are using the whole sample of banks and confirm the conclusions discussed in the main text that focusing only on the EBA sample.

Figure 12: Average income / Total Assets by type of activity: trading, deposit taking or whole bank as computed and based on balance sheet information) for the complete sample[51]. The average is weighted on banks' total assets.

Source : SNL and JRC computation

Figure 13: Average income /MCR by type of activity and by regulatory scenario for the candidates sample of 29 banks. MCR refer to estimated minimum capital requirements. The average is weighted on banks' total assets. , while RegCap is a proxy for return on equity based on total balance sheet regulatory capital (corrected for changes in capital definition)

Source : SNL database and JRC estimates

Figure 14: Zoom on Figure 7 : average income / MCR by type of activity and by regulatory scenario based on balance sheet information) for the Candidates sample. The average is weighted on banks' EU total assets.

Figure 15: Average income / Equity by type of activity and by regulatory scenario for the candidates sample of 29 banks. RegCap is a proxy for return on equity based on total balance sheet regulatory capital (corrected for changes in capital definition). . The average is weighted on banks'  EU total assets.

Source : SNL database and JRC estimates

Figure 16: Zoom on figure 9 : average income / Equity by type of activity and by 0regulatory scenario for the Candidates sample which corresponds to the 29 banks proposed for further investigation for structural reform. The average is weighted on banks' total assets.

Appendix E: Correction of RWAs

To reflect the change in RWA following the introduction of Basel III, QIS correction factors as of June 2011 are employed[52]. These changes are taking into account the strengthening of regulatory capital both in quantity and in quality imposed by Basel III.

Adjustments to take into account the impact due to the introduction of Basel 3 (CRDIV) on RWA, regulatory capital and minimum capital requirements are implemented. These adjustments imply increased RWA, a more strict definition of regulatory capital, and the introduction of the Capital Conservation Buffer. Average EU results of the 2011 Quantitative Impact Study (QIS) are employed for the adjustments, as detailed in the table below. The changes are allocated between the DTB and the TE as described in the next section.

Table 10 - Average EU (weighted on total assets) corrections for RWA and regulatory capital from EBA as of 30/06/2011

|| G1 Banks || G2 Banks

Relative Change in RWA for the whole bank (%) || 21.20 || 6.90

Relative Change in Regulatory capital for the whole bank (%) || -34.35 || -7.76

Source: EBA

Note: In this exercise G1 - Tier 1 Capital > 3 bn€, G2 - Tier 1 Capital <3 bn€

The split Basel 2 RWAs are adjusted to take into account future changes introduced by Basel 3 to RWA definitions and requirements. The Basel 3 increase in the RWA is allocated based on a breakdown of the changes in RWA published by EBA[53], reported here in Table 10. In particular, the Table shows the part of the total percentage increase in RWA due to:

a) the change in the ‘definition of capital’[54], which is split proportionally to the share of total assets allocated to the TE and the DTB.

b) counterparty credit risk, which is allocated to TE for the share due to Credit Valuation adjustment (CVA) and to the DTB for the part due to the higher asset correlation parameter included in the IRB formula.

c) securitization in the banking book, which is fully allocated to the DTB.

d) to market risk (including securitisation in the trading book) is fully allocated to the TE.

Table 11: EBA split of the increase in RWA due to Basel 3 (average %-increase)

Type || Total relative increase in RWA || Part due to definition of capital || CCR banking book || CCR trading book || Securitisation banking book || Trading book

G1 || 21.2 || 7.9 || 1.2 || 6.9 || 1.0 || 4.2

G2 || 6.9 || 3.4 || 2.9 || 0.2 || 0.4

Source: EBA

Results are presented in Table 12 both for Basel 2 and for Basel 3.

Table 12: Allocation of total RWA between the TEs and the DTBs under Basel 2 and Basel 3

|| Basel 2 || Basel 3

DTB || 91% || 79%

TE || 9% || 21%

The new Basel 3 definition of the quality of capital affect both entities, thus the decrease in the regulatory capital is split proportionally to the capital allocated to the two entities. For banks with adjusted regulatory capital below 10.5% of RWA, the capital is topped up to meet the Basel 3 minimum required capital including the capital conservation buffer (10.5% RWA).

Appendix F: ROA by line of activity and relationship with ROMCR

This appendix includes some further formulas illustrating the relationships between ROA, ROE, ROMCR, RWA density, leverage and the capital adequacy ratio.

ROA and ROE are related as follows:

                                 

Where:

ROE – Return on Equity;

E – Equity

TA – Total assets

 - Leverage.

ROA and ROMCR are tied by the following relationship: [55]

Where:

 – Return on Assets

 – Income as defined for the purposes of this report (see Appendix B)

 – Total Assets

 – RWA density

 – Capital Adequacy Ratio (i.e. 8% currently or 10.5% under Basel III regulatory scenarios)

From the two relationships above, ROE is tied to ROMCR as follows:

                                                       

Where all symbols have the same meaning as above.

 

Annex A6: Qualitative assessment of benefits and costs of separating banking activities from deposit taking entities

1. Introduction

Separating banking activities from deposit taking entities should address the problems highlighted in Chapter 2 and should deliver the following social benefits:

· Facilitate bank resolution and recovery;

· Facilitate management, monitoring and supervision;

· Reduce moral hazard;

· Reduce conflicts of interest;

· Reduce capital and resource misallocation; and

· Improve competition.

At the same time, separating banking activities from deposit taking banks may give rise to social costs:

·  Foregone economies of scale and scope; and

· Operational costs.

This annex analyses in more detail the social benefits and costs of separating specific banking activities from a deposit entity. The assessment is largely qualitative, as deciding which activities need to be separated will ultimately be a social and economic issue that cannot be justified on the basis of calibrated and stylised models.

In determining the specific set of activities that should be examined with a view to assess whether they should be subject to separation, the Commission services have considered (i) the extent to which losses related to an activity would impact a bank’s balance sheet; (ii) the extent to which an activity gives rise to market or counterparty risk; (iii) the importance and potential impact of the activity on systemic risk; (iv) the customer-oriented nature and usefulness of an activity for financing the real economy, and (v) the extent to which the banking activity resolves a market failure (such as asymmetric information) in the economy. The application of these criteria suggests a relatively narrow range of corporate and investment banking activities that require further analysis: proprietary trading including bank-owned hedge funds (PT/HF), market making (MM), underwriting (UW), securitisation related activities (SEC), derivatives transactions, exposure to private equity or venture capital funds (PE/VC), and lending to large corporates (LLC).

2. Proprietary trading

Proprietary trading is the purchase and sale of financial instruments for own account with the intent to profit from subsequent price changes. Banks submitted evidence following the Commission public consultation confirming that (dedicated desk) proprietary trading is currently a banking activity of minor importance for many large EU banking groups (see Annex A11). This is consistent with evidence reported for a number of EU Member States.[56]

2.1. Social benefits of separating proprietary trading

Would separating proprietary trading facilitate recovery and resolution? Proprietary trading potentially gives rise to large open positions and counterparty risk (risk that the counterparty to the investment will fail to pay), as well as interconnectedness between institutions. Correspondingly, separating proprietary trading from the deposit entity will facilitate the recovery and resolution of the separate entities. The potential opaqueness, complexity, and interconnectivity of proprietary trading represent important impediments to orderly and swift resolution.

Would separating proprietary trading reduce moral hazard? Proprietary trading is an inherently risky banking activity that is by definition not customer-oriented. It has the ability to produce “tail risk” or systemic risk and is easily scalable (in comparison to more relationship-based activities such as lending). Traders have the ability and incentive to take significant risks, even without having access to liquidity (through short-selling positions). Separating proprietary trading from the deposit entity allows shielding depositors from this type of risk-taking. Reducing the cross-subsidies would also help to re-align private and social interests. The increased funding cost would reflect the inherent riskiness of the activity (although systemic risk may still not be adequately reflected in the institution-specific funding cost). As a result, moral hazard on behalf of the trading entity will be reduced.

Would separating proprietary trading facilitate monitoring, management, and supervision? Increased market discipline on the trading entity will help the supervision of the trading entity, even without factoring in the likely reduction in proprietary trading that would result from its safety net alienation and the enhanced market monitoring that should result from it. The nature of proprietary trading hinders the ability of regulators, supervisors and bank managers to properly understand and thereby calibrate the risks taken, in particular tail risks. It is equally complex to apply the correct capital treatment so that banks have sufficient resources to absorb losses if these occur. Proprietary trading can also be a high-frequency activity that may result in thousands of daily transactions. As a result, snapshots of the positions of these activities may have limited predictive value for future positions. Understanding and monitoring the risks is difficult, in particular when management itself has difficulties in understanding and monitoring the risks. Some of the “rogue trader” losses only became apparent at a late stage, when they eventually could no longer keep accumulated losses hidden from their internal control mechanisms.[57]

Would separating proprietary trading reduce conflicts of interest? Proprietary trading is particularly prone to conflicts of interests because the bank in its role of proprietary trader no longer is a service provider to its client, but becomes a potential competitor and hence faces interests that are no longer aligned with those of its clients. The bank can make improper use of client-related information to increase its own profits. The commercial bank department may have private information about the likely bankruptcy of a firm it has granted a loan and may buy credit protection against the default of the firm from the unsuspecting public, thereby reducing its own credit risk.

Would separating proprietary trading reduce capital and resources misallocation? The traditional raison d’être of deposit-taking banks is to be a financial intermediary between savers and investors (and thereby competing with capital markets that play a similar role). In comparison to capital markets, who intermediate more directly between savers and investors, deposit-taking banks are relatively good at (i) monitoring and knowing their customers, i.e. resolving information asymmetries, (ii) providing insurance against idiosyncratic liquidity risks faced by households and firms, (iii) pooling risks efficiently, and (iv) performing risk-return tranching services to customers. None of these roles is fulfilled by proprietary trading. Therefore, as proprietary trading activities benefit from the implicit subsidy even though they do not fit in the traditional role of banks, capital and human resources are being misallocated to the extent that they are put at work in proprietary trading rather than in engaging in loan making and other core banking services (or even other activities beyond the banking sector). Banks would no longer have an incentive to over- expand their proprietary trading activities.

Would separating proprietary trading impact on competition? Given that implicit subsidies to proprietary trading activities would be reduced with separation, there would be a beneficial impact on competition amongst banks. However, the impact will be proportional to the relatively limited importance of proprietary trading.

2.2. Social costs of separating proprietary trading

Would separating proprietary trading lead to a loss of efficiencies? Separating proprietary trading is unlikely to lead to significant social costs. In fact, as argued above, proprietary trading can be characterised by diseconomies of scope such as excessive complexity, conflicts of interest, excessive risk taking and interconnectedness which can lead to higher systemic risk. Genuine economies of scope related to risk diversification cannot be excluded but are likely to be small, given the limited importance of the activity. Also there may be some cost economies of scope if proprietary trading is subsidiarised as a stand-alone activity, as proprietary trading employs the trading infrastructure used in other activities such as market making. Those would not be lost if trading activities are subsidiarised altogether or if proprietary trading is prohibited from the banking group. In sum, significant economies of scope are unlikely to be foregone following a separation of proprietary trading.

The impact of separating proprietary trading on economic growth is likely to be small. The funding cost related to the activity is likely to go up when the inherent riskiness of the activity needs to be reflected, but the costs cannot be passed on to households and SMEs, given that proprietary trading is an own account activity. Costs are likely to be borne by traders (lower wages) or shareholders (lower net worth). Aggregate consumption may go down as a result, but again not to a material extent. In any case, the increased financial stability and elimination of diseconomies of scope will dwarf any quite hypothetical social cost.

Industry claims that prohibiting proprietary trading would negatively and significantly influence market liquidity and price-discovery relate more to market making. Also, as banks claim that they do no longer engage in proprietary trading activities to a material extent, such negative consequences appear overstated.[58]

What is the impact on stakeholder groups of separating proprietary trading? Depending on the strength of separation, the reduced scope for conflicts of interest should have a positive effect on most bank clients (households and corporates). The impact on the banking industry should also be limited, given claims that banks no longer engage in this activity to a material extent. Traders will face increased competition from the hedge fund industry.

Are there doubts about the effectiveness of separating proprietary trading in achieving its objectives? It is difficult for people not involved in the actual transactions to distinguish proprietary trading (say, buying and holding a highly illiquid asset to benefit from the expected price dynamics) from customer-driven trading (say, buying and holding a highly illiquid asset because you expect a customer demand to arise for the asset in the near future). Indeed, a market maker might legitimately choose to take a long position in an asset either in anticipation of client demand to allow the order to be fulfilled quickly or to facilitate a quick sale by a client of an illiquid asset.

Moreover, proprietary trading can be conducted in other divisions of the banking group alongside the permitted activity. Proprietary trading is for example difficult to distinguish from Treasury management operations.[59]

For both these reasons, it will be challenging to avoid that proprietary trading takes place within a deposit taking entity after merely having formally separated proprietary trading into a separated trading entity. 

3. Market making

Market making makes up for a significant part of large banking group’s trading revenues.[60]

In general terms, market making is the purchase and sale of financial instruments (government bonds, corporate bonds, equities, derivatives, etc.) for own account at prices defined by the market maker, on the basis of a commitment to provide market liquidity on a regular and on-going basis.[61] Consequently, market makers provide "immediacy" to clients and investors by facilitating their requests to buy and sell quickly and, arguably, in a cost-effective way for them. For example, an investor anxious to sell an asset relies on a market maker's standing ability to buy the asset for itself, immediately. Likewise, an investor who wishes to buy an asset often can call on a market maker to sell the asset out of its inventory. By doing so, market makers can instil greater investor confidence in the functioning of financial markets and encourage investors to trade confidently. Without market makers, customers would face higher transaction costs and security prices would be more volatile. A market maker acquires a position at one price and then lays off the position over time at an uncertain average price by providing liquidity to customers. The ultimate goal is to "buy low, sell high". In order to accomplish this goal on average over many trades, with an acceptable level of risk for the expected profit, a market maker relies on its expectation of the investors’ needs and the future path of market prices. In general, market makers provide liquidity and produce positive externalities.

Although traders involved in the actual trade are able to identify any given transaction as being of a market making or proprietary trading nature, such a distinction no longer is simple from the perspective of an outsider such as a manager, regulator, supervisor, creditor, or judge. From a legal and economic point of view, market making (and the securities inventory used to facilitate customer trading) is difficult to distinguish from proprietary trading, in particular for “outsiders”. Indeed, a market maker might legitimately choose to take a long position in an asset either in anticipation of client demand to allow the order to be fulfilled quickly or to facilitate a quick sale by a client of an illiquid asset.

While it is possible for institutions other than banks (such as funds) to take on a similar role to market makers, banks do have a natural advantage in acting as market makers because of the fact that banks have a variety of other relationships with the clients who want to make trades and the fact that acting as a market-maker for a security is often a natural follow-on activity for securities underwritten by the banking group.

The most active market makers in financial markets today are High Frequency Traders (HFT), many of whom trade as voluntary market makers with no obligations to maintain markets. According to several academic studies, high frequency market making is a profitable enterprise and, more importantly, market quality has improved alongside the growth in algorithmic trading. These results are frequently interpreted as support for a structure where participants supply liquidity because it is a profitable and viable activity on its own (see Anand and Vankatamaran (2013) for a more in-depth analysis). Several important market makers are not taking any deposits, also suggesting that market making is a viable activity on its own.

Market-making is entwined with underwriting. In every initial public offering (IPO), the lead underwriter always acts as a market maker. Market makers build up significant inventories following underwriting transactions. Ellis et al. (2000) report for Nasdaq IPOs that the lead underwriter has accumulated as market maker approximately 8% as inventory position after 20 trading days. Inventory accumulation by the underwriter gives a direct measure of price stabilisation activities. While simply buying and selling securities over the course of a trading day is not unusual for any market maker on any stock, accumulation of a significant inventory position is.

Many IPOs experience large price gains however, negating any need for market maker purposes. The inventory position of the lead underwriter will depend on the subsequent return of the IPO. The lead manager may assume approximately 60% of the trading volume on the first day. The lead manager’s share of the trading volume slowly declines, but it typically still remains greater than 40% even three months after the IPO has begun trading. Of course, the trading volume in numbers of shares decreases dramatically over time. Most co-managers make a market in the issue as well, but to a fairly limited extent.

The market making profits of the underwriter include both the trading profits due to buying and selling at his quotes, and the profits and losses of his inventory position. In general, market making is profitable, particularly on the offer day. Ellis et al. (2000) find that there is no significant difference between the inventory profits of underwriters of successful and unsuccessful IPOs. This suggests that the overallotment option (see section 4 on underwriting below) is successful in reducing inventory risks for underwriters. The trading profits seem to relate to the IPO being successful or not.

Overall, it is found that market making is not a cost to underwriters, that total market making profits are positive on the first day, and that they remain positive throughout the first month of trading (Ellis et al. (2000)).

3.1. Social benefits of separating market making

Would separating market making facilitate recovery and resolution? Would separating market making facilitate monitoring, management, and supervision? A separation of market making activities would have social benefits in terms of facilitating resolvability. The resolvability of a bank is impeded by the presence of trading and inventory within a large banking group (again, in particular due to market making as follow-on activity of securities underwriting or proprietary trading, but possibly also when significant customer orders are expected to arise in the future). Individual trading positions are treated the same way in resolution whether they result from client activity driven market making or speculation, and market making affects the quantity of positions needing to be resolved. Impediments to orderly and swift resolution, monitoring, regulation, supervisability of the activity are the opaqueness, complexity, and interconnectivity of market making. Market makers are interconnected with other large banking groups.

Would separating market making reduce moral hazard? When facilitating client business, a bank is likely to try and hedge most of its risks. Hence, genuine market making should entail limited market risk. However, the actual exposure to risk may vary across time depending on the liquidity of the instruments, on changes in market volatility and on significant variation in the sizes of positions that market making clients may wish to acquire or liquidate. Moreover, there may be a mismatch between the position and the hedge (basis risk) and the hedge will need to be rebalanced over time as market moves alter risk profiles. Furthermore, market makers are still exposed to high counterparty risk and the concrete functioning of market making can vary in relation to different financial instruments and market models.

Market making as a follow-on activity of underwriting does imply that significant securities and derivatives inventories are being built up (see above), and hence that risks are potentially significant, although hedging instruments exist.

Given its importance as a share of trading revenues, market making entails significant risk and separating it from the deposit entity will significantly reduce moral hazard, excessive risk taking, and artificial balance sheet expansion.

Making a distinction between genuine market making and proprietary trading is inherently difficult for outsiders, however. Banks remain highly leveraged and highly expert organisations that aim to make profits from managing their balance sheet. To the extent that banks conceal proprietary trades as market making transactions, the arguments raised above for proprietary trading continue to hold. Traders have the ability and the incentive to take significant risks, and interests are not always aligned with those of its customers.

Would separating market making reduce conflicts of interest? In theory, genuine market making is aiming to facilitate client business and hence the bank interests are supposed to be aligned with customer interests. However, principal agent problems need not to be confined to proprietary trading given that market making and proprietary trading activity are difficult to disentangle for outsiders to the actual transactions.

In general, if markets are opaque, such as is the case in over-the-counter markets, and if market makers have superior access to information, collusion and exploitation of conflicts of interests may occur. The origin of the problem is an inherent conflict of interest. Banks possess (asymmetric) information in the form of customer trade details, including the number and size of trades to be executed. And they have knowledge that their own proprietary positions could be harmed without or could benefit with trader intervention. The banks allegedly act on that knowledge, against their customer’s best interests and in favour of their own, as evidenced in recent banking scandals, related to front running, FX bid rigging, Libor benchmark rate setting, etc.[62]

Would separating market making reduce capital and resources misallocation? The inherent riskiness of trading attracts and requires people who are good at taking short-term risks rather than lenders with a long-term perspective. Absent separation, a short-term returns culture may arise within the entire banking group, given the high profitability associated with trading.

Academics have argued that market makers hamper the development of securities markets.[63] Large universal banks are currently accused of having protected their indispensable position in the global CDS market through control of a trading body and information provider, which vetted whether new exchanges should be licensed. The alleged harm consists of exchanges being blocked from bringing part of the over-the-counter CDS transactions onto public exchanges, which would have resulted in lower transaction costs for their investor-customers, as well as in less financial instability as OTC markets are more opaque and involve more counterparty risk.

As market making activities would no longer benefit from the same level of implicit subsidies, banks would not have an incentive to over-expand their activities in the field. The level of market marking activities would reflect the market pricing and therefore would not attract resources from other banking activities (or activities beyond banking).

Would separating market making impact on competition? The competition benefits relate to the removal of the implicit subsidy, post separation, as it would allow banks to compete on a level playing field. Given that market making is a significant part of banks' trading activities, the impact on competition would be relatively important. Also market makers that do not currently benefit from implicit subsidies (non-bank broker dealers) would be in a position to compete on a level playing field with the separated entities undertaking market making activities.

3.2. Social costs of separating market making

Would separating market making lead to a loss of efficiencies? Given that market making comprises significant segment of trading activities, there would be some economies of scale, and cost and diversification economies of scope lost. However, the former would only apply to small banks, while the latter would be limited and are likely to be dominated by diseconomies of scope such as increased systemic risk, excessive risk taking, increased complexity and conflicts of interests. For example, while diversification may make individual bank default less likely it actually may increase the likelihood of systemic risk (as they become more similar and they have increased interconnectedness).

As explained in chapter 4, integrated banks do have a natural advantage in acting as market makers because of the fact that banks have a variety of other relationships with the clients who want to make trades and the fact that acting as a market maker for a security is often a natural follow-on activity for securities underwritten by the banking group. This may allow integrated banks to perform this activity more efficiently than other market players, thus better serving clients and/or contributing to enhancing market liquidity. Such effects should be weighed against potential benefits that could flow from structural separation, in particular for market making, as discussed elsewhere in this Impact Assessment.

It is frequently argued that separating market making will harm market liquidity and hence will be socially costly. Bid-ask spreads may increase, increasing the costs to trade at any scale. Likewise, the set of options to investors will be reduced, as they can no longer trade as much and as easily as before. Price discovery is made more difficult. And price volatility may be reduced, if professional position takers spot price divergences from rational levels and correct them through speculation and trading.

This argument (i) neglects the fact that structural separation aims to reduce the implicit subsidies that distort the proper market functioning and bank activities, (ii) builds on the presumption that more liquidity is inherently positive, which may not always be the case[64], and (iii) should be put into perspective. Indeed, market prices are distorted when contaminated with implicit public subsidies and may in fact produce excess liquidity. One could argue that it is preferable to allow the discipline of the market to choose the pricing of these securities and the amount of liquidity. If liquidity cannot be reached then it may suggest more about the underlying securities' viability (see Richardson, 2013).

Richardson (2013) notes that the issue of liquidity is more relevant in times of crisis than in normal times when liquidity is typically not a pressing concern. Private banks, however, have not performed a significant liquidity role during crisis period and central banks have stepped in to assume the role of Market Maker of Last Resort (in covered bond markets, government bond markets, etc.). Charts 3.11 and 3.2 plot the yields and corresponding bid-ask spreads of 10 year Spanish government bonds. Chart 3 shows that yields have more than doubled and increased from less than 3.5% in June 2006 to more than 7.5% in July 2012. Chart 4 plots bid-ask spreads. Whereas their pattern is equally volatile, it is clear that they matter much less in comparison to the changes in the interest rate level. Bid-ask spreads in the period June 2006 to August 2013 on average are 2bp (0.02%) and spiked at 12bp (0.12%) in June 2012. This suggests that the ability of (private sector) market makers to influence the interest rate level is relatively limited.

Chart 3.1: Yield to maturity of 10 year Spanish government bonds || Chart 3.2: Bid-Ask spreads of 10 year Spanish government bonds

||

Source: Commission Services - Bloomberg || Source: Commission Services - Bloomberg

To function properly markets need a large number of independent traders. A separation between deposit entities and trading entities deprives the latter of access to cheap funds (in the form of deposits), forcing them to limit their size and the size of their bets. These limitations may increase the number of market participants, making markets more liquid.[65]

The increased funding cost for the trading entity that acts as market maker is unlikely to be passed on to the real economy and therefore harm economic growth.

– First, households and SMEs that are clients of a banking group that needs to separate certain capital market activities are typically and mainly clients of the deposit entity. Hence, the increased funding cost for the entity not taking deposits would not necessarily affect borrowing conditions for households and SMEs. In fact, market making entails significant risk.[66] These risks are important, given the size and importance of market making as a share of large banks’ trading activities. Separating market making from the deposit entity will reduce excessive risk taking and artificial balance sheet expansion and hence may lower the funding cost for the deposit entity.

– Second, medium-sized competitors or new entrants that are not subject to mandatory separation may gain market share from large banking groups if artificial competition distortions in favour of too-big-to fail banking groups are being reduced. Hence, whereas some banking groups may face increased costs and may no longer serve certain customers, those activities may be picked up by smaller competitors that do not face structural separation requirements. Customers are accordingly not likely to be left unserved.

– Third, under a subsidiarisation model, market making is not prohibited within a banking group. It just needs to be performed by a legally separate trading entity. Estimates for UK banks (that are amongst the most important players in the targeted trading activities) suggest that funding costs may go up for the trading entity in a range between [0bp and 75bp] (HM Treasury, 2012). As said before, the increased funding cost for the trading entity is part of the desired effects of the separation.

– Finally, market making is a financially viable activity on its own, as shown by Anand and Vankataraman (2012) and as illustrated by the fact that several important market makers are not taking any deposits.

What is the impact on stakeholder groups of separating market making? Again, the ultimate impact will also depend on the strength of the separation. Distortionary implicit subsidies are being eliminated. The scope for excessive risk-taking should decrease and conflicts of interest should be reduced. The impact on bank creditors would differ between the trading entity (more exposed to risk, hence requiring higher returns) and the deposit entity (required returns may decrease).

4. Underwriting

Securities underwriting is a typical investment banking activity in which banks raise investment capital from investors on behalf of corporations and governments that are issuing securities (both equity and debt securities) in return for a fee. It is a way of selling newly issued securities, such as stocks or bonds, to investors.

There are two types of underwriting. "Firm commitment" underwriting is one in which the underwriter guarantees the sale of the issued stock at the agreed-upon price. Hence the underwriter takes the risk of an unsuccessful sale. In practice, a syndicate of banks underwrites the transaction, which means they take on the risk of distributing the securities. Should they not be able to find enough investors, they will have to hold unsold securities themselves. Alternatively, in a "best efforts" contract, the underwriter agrees to sell as many shares as possible at the agreed-upon price. Such a contract generates more limited risk for the underwriter, in comparison. Typically, the process is led by a lead underwriter or book manager, sometimes one or more co-managers and a large syndicate of investment banks that aid in the distribution of shares.

Underwriters make their income from the difference between the price they pay to the issuer and what they collect from investors or from broker-dealers who buy portions of the offering ("underwriting or gross spread" or “underwriting fees”). While underwriters are typically involved in immediate IPO aftermarket trading and this market making is a stand-alone profit centre, underwriters generate most of their profits by the fees they demand for their services in issuing securities.[67]

In most cases, the managing underwriter “overallots” the issue, creating a short position by accepting more orders than there are shares to be sold. The overallotment option grants an option to the underwriter to purchase from the company an additional portion of the shares sold in the IPO at the offer price. With this option, an underwriter can and virtually always does sell more than 100% of the offered shares. The motivation for this option is to provide buying support for the shares without exposing the underwriter to excessive risk. The underwriter buys back the extra portion of shares in the market at the offer price, they can do this without the market risk of being "long" this extra portion in their own account, as they are simply "covering" (closing out) their short position.

Put differently, if the offering is strong and the price goes up, the underwriter covers his short position by exercising the overallotment option at the offering price. The underwriter is able to close its short position by purchasing shares at the same price for which they sold-short the shares, so the underwriter does not lose money. If the offering is weak and the price goes down, the underwriter does not exercise the option, and instead buys back all or part of the extra portion of the shares in the market at the price below which it can sell it to its customers, thereby supporting the stock price. The overallotment option provides the underwriter with buying power in the aftermarket, enabling him to support the price of the newly traded security. The underwriter typically has 30 days to decide to exercise all or part of his option.

Underwriting and advisory services require relationship-building with clients. It is closely connected to corporate banking. From the corporate client’s perspective, issuing a bond is an alternative way of financing to taking a bank loan.

Deutsche Bank, HSBC, BNP Paribas, Barclays and Unicredit make up the top 5 of the league tables for Eurobonds (2012), whereas Goldman Sachs, Deutsche Bank, Morgan Stanley, Credit Suisse, and Bank of America Merrill Lynch make up the top 5 in the league table on stocks (2011, Europe, Middle-East, and Africa). The top 10 debt underwriters account for 43% of Euromarket corporate debt issuance, and even for 51% of the government debt market. Each of the top 5 debt underwriters has underwritten in excess of 100 billion EUR in 2012. See Chart 4.1.

Chart 4.1 Euromarket debt underwriting (2012)

4.1. Social benefits of separating underwriting

Would separating underwriting facilitate recovery and resolution? Given that underwriters typically retain a significant fraction in their inventories and play an active market making role after the issuance, resolution may be improved by separating underwriting from deposit taking, as the inventory of relatively illiquid assets will be smaller. Otherwise, the underwriting as such does not give rise to similar interconnectedness across financial institutions. 

Would separating underwriting reduce moral hazard? Firm-commitment underwriting is an inherently risky banking activity, although underwriters can hedge themselves to reduce the corresponding risks (Ellis et al. (2000)). Separating the activity from the deposit entity will shield depositors from such risks. Allowing it to be performed by a legal entity that enjoys an explicit safety net will encourage the activity, as the safety net presence will reduce the risk-sensitivity of its funding sources. Given that underwriting is not as easily scalable as pure market making, the scope for moral hazard reduction is significant, but smaller than for market making.   

Would separating underwriting facilitate monitoring, management, and supervision? Given that underwriting is prone to conflicts of interests, the management of a group would be facilitated if performed in a more structured way in which underwriting is being separated from lending and other commercial banking activities. It does not create challenges as high as for market making, given that the activity is not short term and easily scalable.

Would separating underwriting reduce conflicts of interest? Separating underwriting will reduce the scope for conflicts of interests, as the interests of the bank as underwriter and as loan provider are typically not aligned.[68] Within a large and diversified banking group, the commercial bank department may have private information about the likely bankruptcy of the firm it has granted a loan and may hence encourage the underwriting department to sell bonds or issue shares to the unsuspecting public, thereby reducing its own credit risk whilst earning a fee. Banks have an incentive to hedge their risk as underwriters, guaranteeing the proceeds of the share issue, but this may potentially have an adverse impact on their clients’ share price. Alternatively, a bank’s lending division may feel pressured to provide bank loans to a firm whose shares have been issued by the bank’s underwriting division, even though such loans would not be granted absent any such in-house pressure. According to certain studies, earnings forecasts and stock recommendations provided by an analyst working with the lead-underwriter are on average inaccurate and positively biased, and unaffiliated analysts perform better and provide higher long-run value to their customers. The main concern is that the bank uses the informational advantage it gains from conducting different activities to its own advantage, thereby misleading customers and investors.

Having said that, the evidence does not suggest that conflicts of interests are obvious between underwriting and loan making.[69] In fact, it suggests that bonds underwritten by commercial banks default less often than bonds underwritten by investment banks. However, the conflicts of interest seem more severe and more likely to exist in a universal bank that has an underwriting division together with an asset management division. These studies seem to support the view that asset management divisions may feel pressured by the bank’s underwriting division to buy and hold poorly performing issues to make a customer satisfied, even though this may be unwise.

Next to internal monitoring and controlling procedures, there is outside regulation (for example with respect to insider trading) and the rule of law to contain the exploitation of possible conflicts of interests. In principle, the market can also respond to apparent conflicts of interests, thereby constraining their scope. The market can penalize the service provider if they exploit conflicts of interest, in the form of a higher funding cost or lower demand for its services even to the point of forcing the firm into bankruptcy. The market can also promote new institutional means to contain conflicts of interest, by generating a demand for information from non-conflicted specialized organizations.

However, the market is likely to be unable to contain the incentives to exploit the conflicts of interests. For the market to be able to do this, it needs to have information on whether exploitation might take place. Sometimes, such information is simply not available or would require the revealing of proprietary information that would benefit a firm's competitors, thus reducing the incentives to reveal this information. Sometimes, when corporate governance is poor, even the top management of the firm is not aware of the conflicts of interest and mala fide opportunistic individuals are able to capture the firm's reputational rents.

Would separating underwriting reduce capital and resources misallocation? Conflicts of interest can substantially reduce the quality of information in financial markets, thereby increasing asymmetric information problems. In turn, asymmetric information prevents financial markets from channeling funds into the most productive investment opportunities and causes financial markets and the economy to become less efficient. Conflicts of interest become a problem for the financial system when they lead to a decrease in the flow of reliable information, either because information is concealed or because misleading information is spread. The decline in the flow of reliable information makes it harder for the financial system to solve adverse selection and moral hazard problems, which can slow the flow of credit to parties with productive investment opportunities.

On the other hand, being a client-driven and core relationship-oriented investment banking activity that was not at the root of the financial crisis, it is not obvious that banks have misdirected significant amounts of human resources and capital to underwriting activities. Moreover, underwriters fulfil a certification role to monitor public firms and their governance structures.[70]

Would separating underwriting impact on competition? To the extent that underwriting activities benefit from an implicit public subsidy, there would be benefits in terms of competition. However, given that underwriting is less scalable than market making or proprietary trading, it is not able to rapidly exploit significant implicit subsidies.

4.2. Social costs of separating underwriting

Would separating underwriting lead to a loss of efficiencies? The literature suggests that the processing of information about making loans to clients facilitates the efficient provision of other financial services, including securities underwriting. This positive information-sharing process can also work in the opposite direction, whereby underwriting and other activities may improve loan-making procedures, but this relationship remains underexplored in the literature. [71] Hence, a strong separation of underwriting from deposit issuing activities would reduce efficiencies (genuine cost and revenue economies of scope) because of the social costs associated with losing knowledge and information advantages associated with combining lending and underwriting.[72]

These efficiency gains and linked savings could be particularly pronounced for the issuers who are noninvestment grade-rated (Drucker and Puri (2005)). There may be economies of scope to be enjoyed from spreading fixed costs of acquiring information over multiple outputs; more specifically, concurrent lending and underwriting could be beneficial. Furthermore, the primary market activity of underwriting is connected to the secondary market activity of market making. In addition, underwriting is naturally followed by market making of the security that has been underwritten by the underwriting bank and therefore significant economies of scope arise between these two activities.

Hence, some economies of scope may arise, in particular if the separation goes beyond a subsidiarisation.

What is the impact on stakeholder groups of separating underwriting? Again, the ultimate impact will also depend on the strength of the separation. SMEs and large corporates may benefit from the reduced scope of conflicts of interests, but they may face a higher cost following the separation of underwriting from the DE.

5. Securitisation

Securitisation as an activity refers to the investing, sponsoring and structuring activities related to certain securitisation instruments.

Asset backed securities (ABS) are securities backed by a pool of receivables. Investors only bear the risk arising from these receivables and are generally insulated from the credit risk of the respective (former) owner of the assets (originator/seller). The receivables of the underlying portfolio that is securitised generate interest and principal payments. These payments as well as potential losses that may occur in case the underlying obligors of the securitised assets do not serve their obligations, are distributed to investors according to certain rules (“the structure”). Hence, the investors in ABS have to focus on both the underlying risk of the securitised portfolio and the rules that determine which consequences investors have to face in case a certain event occurs. Typically, the securitised assets are referenced by various notes with different risk profiles, and hence, ratings. The fact that different notes have different risk profiles, though they all reference the same underlying portfolio, is based on the respective aforementioned transaction structure. This in principle can enable investors to satisfy their individual risk appetite and needs. ABS allows for a broad band of flexibility in terms of asset classes being securitised and structures being applied.

Chart 5.1: Stylised illustration of the transfer of cash flows and risks in a securitisation deal

Chart 5.1 is a simple illustration of a securitisation deal with the main three parties involved: originator, the Special Purpose Vehicle («SPV») and investors. The seller or originator is the original owner of the assets. The Special Purpose Vehicle («SPV») is an independent entity, created for the purpose of the one specific securitisation transaction. It is typically bankruptcy remote and often located off-shore for tax reasons. Investors are typically asset managers. They are allowed to buy and sell assets during the time period subject to rules (in case of managed, i.e. non-static, pools of underlying securities). However, in practice, there are more parties involved in the transaction. The arranger determines the underlying pool of assets and sets up the transaction structure on behalf of the originator. The servicer collects and distributes the cash flows that arise from the assets (interest and/or principal), arrears management, collateral management. The provider of external credit enhancement/liquidity provides subordinated loans, guarantees, insurances (as protection for investors against credit risk) as well as liquidity facilities (“sponsor”). The rating agencies assess both the credit risk of the underlying pool of assets and the structural features of the transaction, and consequently determine the size of the required credit enhancement to achieve a certain rating level. The trustee is the “agent” of the investors that controls and checks cash flows, loss allocations, adherence to provisions, etc. The swap counterparty provides the interest rate swap to hedge market risk, as the SPV must not bear any interest rate risk due to the bankruptcy remoteness requirement, and as the securitised assets very often generate fixed-rate interest income, while the notes are mostly floaters (possibly also currency swaps).

The rationale of securitisation from the viewpoint of the originator/seller-perspective is that it allows for (i) balance sheet management, (ii) portfolio and risk management, (iii) funding management, (iv) price discovery by selling and “liquifying” illiquid assets and, (v) most importantly, lower capital requirements. The rationale of securitisation from the investor-perspective is that investors benefit from (i) the diversifications’ effects due to investments in alternative asset classes (risk diversification across geographies and asset types), (ii) the flexibility of securitisation (instruments tailored to the needs, preferences and profile with the opportunity to invest in high quality asset classes or earn a high rate of return), and (iii) being insulated from the originator’s/seller’s credit risk. In an increasingly fragmented European banking market where interbank markets gradually evaporate. Securitisation allows to channel savings from parts of Europe that have a surplus to those parts of Europe with the greatest needs and growth prospects, which is especially beneficial in an increasingly fragmented European banking market where interbank markets gradually evaporate. Whereas the funding can come from investors and depositors across Europe, the loan origination can remain with the existing local providers.

Different types of securitisations are classified, depending on the nature of assets underlying the pool. The underlying assets can be credit cards, consumer or auto credits as well leases. However, the majority of securitisation transactions in Europe, as Chart 5.2 illustrates, are Residential Mortgage-backed Securities (RMBS)[73], followed by Collateralized Debt Obligations[74] (CDO), Collateralized Loan Obligations[75] (CLO) and Commercial Mortgage-backed Securities (CMBS). After the peak of securitisation issuance in 2008, securitisation in Europe shows a steep downwards trend, with a significant reduction in the issuance of CDO/CLO and RMBS, as displayed in chart 5.3.

Chart 5.2. Source: JP Morgan research June 2013

 

Chart 5.3. Source: JP Morgan research June 2013

Prior to the crisis, securitisation was regarded as contributing to financial stability, as it disperses credit risk to those that are best suited to bear it and as long intermediation chains may promote more efficient maturity transformation. However, the financial crisis has led to a rethink. Ex post, the boom of securitisation allowed to build up risks quickly, it enabled significant growth in short-term debt between financial intermediaries and led to financial intermediaries becoming more intertwined.

The valuation losses and difficulties with securitisation that have materialised since 2007 can be linked to excessive (i) originate-to-distribute activity, (ii) leverage, (iii) embedded maturity transformation, and (iv) lack of transparency.

Accordingly, for the purposes of this impact assessment, “complex” securitisation is defined as all securitisation activity that is characterised by any of the following four undesirable features:

(1) Securitisation with risk transfer - having insufficient “skin in the game”.

(2) Re-securitisation – Securitisation of a synthetic nature and/or containing mixed and complex pools of underlying assets. Repackaging or leveraging of existing securitisation instruments becomes increasingly complex, requires sophisticated modelling approach and is prone to model risk and valuation uncertainty.

(3) Securitisation with maturity transformation: vehicles such as SIVs and ABCPs perform maturity transformation, on top of the liquidity risk and credit risk transformation and can be classified as unsafe types of securitisation.

(4) Intransparent securitisation: due to the inherent complexity of the underlying pool of assets or due to available data to assess the riskiness of the instrument.

Alternatively, “simple and transparent” securitisation is characterised by the following joint characteristics:

(1) Retention of sufficient “skin in the game” requirements (absence of full risk transfer), ideally of a vertical slice, as allowed by CRDIV-CRR rules.

(2) Primary securitisation: structures other than of a synthetic nature; no repackaging or leveraging of existing securitisation instruments, as they become increasingly complex and prone to modelling risk; no mixed pools of underlying assets..

(3) Securitisation without maturity transformation, the so called “pass-through” securitisation, links the rights of investors to receive interest and principal to the securitised assets’ generation of cash, subject only to hedging instruments or the availability of a liquidity facility designed to smooth out irregularities in payments. Vehicles such as SIVs and ABCPs perform maturity transformation and are considered “unsafe” securitisation.

Separation of securitisation activities would allow deposit entities to conduct and invest only in “simple and transparent” and not in “complex” securitisation.[76] In the following sections the social benefits and costs of this separation are discussed.

5.1. Social benefits of separating securitisation and debt origination

Would separating securitisation facilitate recovery and resolution? Would separating securitisation facilitate monitoring, management, and supervision? Separating “complex” securitisation will facilitate resolvability of large banking groups and facilitate management and supervision, as it concerns complex, opaque and illiquid securitised instruments. If the securitised assets become complex, they become hard to value and sell off without incurring large losses. Likewise, investing in securitised products may give rise to interconnectedness of financial institutions which impedes orderly and swift resolution. The ability to adequately supervise, regulate, and manage ABS structures is likely to be inversely related to its complexity and degree of transparency

Would separating securitisation reduce moral hazard? “Complex” securitization has allowed banks to grow their bank balance sheet aggressively, build up risks quickly, concentrate risks within the leveraged sector, grow notably short-term debt reliance between financial intermediaries, and make financial intermediaries significantly more intertwined. Without sufficient skin in the game, securitisation becomes a transaction-oriented and fee-driven activity, which can be scaled up easily, as clearly demonstrated in the financial crisis. If loans are packaged and sold off to investors against a fee, banks have less incentive to monitor borrowers and the interests of the bank and its borrowers are no longer aligned. Likewise, unless maturity transformation is limited, banks still run significant liquidity risk. Both, opaque securitisation structures and maturity transformation have fuelled the originate-to-distribute model, followed-up from excessive build-up of risks in the financial system. In that sense, separating “complex” securitisation from the public safety net would reduce moral hazard significantly and would stop the artificial and socially undesirable promotion of these activities.

Would separating securitisation reduce conflicts of interest? Conflicts of interests clearly arise when banks are allowed to sell ABS to investors against a fee without retaining sufficient skin in the game. This was the case during the crisis with the originate-to distribute business model.

Would separating securitisation reduce capital and resources misallocation? In retrospect, the securitisation process and transactions has led to a misallocation of capital and has given rise to real estate bubbles (in the US, Ireland, Spain, etc.) and to credit bubble in debt capital markets. They have fuelled the credit default spread (CDS) markets (used to hedge the credit risk of the underlying) and complex and synthetic fixed income products, such as collateralised debt obligation (CDOs) and CDO².

Would separating securitisation impact on competition? Separating “complex” securitisation from the public safety net would reduce the implicit subsidy that these activities benefit from and would lead to enhanced competition on the merits among banks. 

5.2. Social costs of separating securitisation and debt origination

Would separating securitisation lead to a loss of efficiencies? “Simple and transparent” securitisation potentially generates genuine economies of scope related to risk diversification, regulatory capital release, revenue increases, and cost reductions. Such securitisation facilitates banks’ ability to perform their risk pooling and asymmetric information resolving role. In general, securitisation is a funding source that is not tied to the credit of the bank. Hence, in a time of doubt about a bank’s financial strength, it may be relatively resilient, compared to unsecured or short term secured borrowing. It allows capital market investors (such as insurance companies, pension funds, etc.) to fund EU banks. It also breaks the link between the financing needs of the economy and the banks’ capacity to raise capital. However, “complex” securitisation is characterised by potential diseconomies of scope such as excessive complexity, conflicts of interest, excessive risk taking and increased systemic risk. There has been a sudden stop in “complex” securitisation following the eruption of the financial crisis, and it is not clear to what extent it will re-emerge. Hence, the cost of separating this activity is likely to be low, based on current activity levels. Subsidiarising “complex” securitisation in a separate trading entity would still allow for market-constrained, limited and genuine innovation to take place within the regulated banking group, whilst not promoting it artificially by linking it to the deposit entity that enjoys public safety net support.

In addition, securitisation has allowed for a rapid expansion of the financial sector through greater interconnectedness. The enlargement of the financial system, beyond a certain size, is associated with reductions in real productivity growth. This, in part, is due to the financial sector competing with the rest of the economy for scarce resources. Excessively large financial systems may reduce economic growth because of the increased probability of a misallocation of resources, the increased probability of large economic crashes[77], or the endogenous feeding of speculative bubbles.

6. Derivative transactions

Trading in derivatives is mostly performed over-the-counter (OTC) and is dominated by a handful of large US and European banking groups. Derivatives are contracts between two counterparties. Their notional value has increased significantly and has reached extremely high levels. Chart 6.1 compares the notional value of derivatives and the volume of primary securities, consisting of all issued domestic and international securities, bank intermediated credit and equity market capitalisation, as a percent of world GDP. The Chart illustrates that the notional value of derivatives has increased from 3.5 times world GDP in 1998 to 12 times world GDP when the crisis materialised. The volume of primary securities, in contrast, has remained stable at around 3 times world GDP.

Chart 6.1: Derivative notional values versus primary securities (% of world GDP)

Source: BIS, OECD, World Federation of Stock Exchanges, as reported in Blundell-Wignall, Atkinson, and Roulet (2012)

Variation margins reflect changes in prices and volatility and imply transfers from losing or out-of-the-money counterparties to winning counterparties (zero sum game). The margins are mostly collateralised with cash and sometimes government securities.

The notional value may not be informative about the riskiness of the derivative positions. First, the Gross Market Value (GMV) measures what it would cost to replace all trades at current market prices. It is typically significantly smaller than the notional value. While the notional value of global derivatives was 586 trillion USD in December 2007, the GMV at the same time was only 16 trillion USD. Even when valued at GMV, derivatives can still be very important in terms of balance sheet of the biggest systemically important banks. Second, financial firms have offsetting positions that can be netted and banks expressly hedge most of their positions. The GMV minus netting is the Gross Credit Exposure (GCE). It is against the GCE that collateral is held. It amounted to 3.3 trillion USD in December 2007, against which 2.1 trillion USD was held. The final global open exposure amounted to 1.2 trillion USD.

However, it would be a mistake to conclude from relatively small open positions that systemic risks are equally small. Changes in volatility may shift the GMV quickly and netting provides no protection against such shifts in market risks, because netting is about settlement amounts using prices at the point of close out. When the crisis hit in 2008, the GMV more than doubled from 15.8 trillion USD to 35.3 trillion USD, the GCE increased from 3.3 trillion USD to 5 trillion USD and the estimated collateral had to rise from 2,1 trillion USD to 4 trillion USD. Significant margin calls needed to be met in a highly risky environment. 

 Charts 6.2 and 6.3 illustrate the size if derivatives for EU banks.

Chart 6.2 – Total liabilities for the EU average bank (2007 versus 2012)

Source: ECB

Chart 6.3 – Derivatives as a share of total liabilities for EU Member States

Source: ECB

6.1. Social benefits of separating derivatives activity

Would separating derivatives activity facilitate recovery and resolution? Would separating derivatives activity facilitate monitoring, management, and supervision? ISDA surveys suggest that roughly three quarters of trades comingle these margins in non-segregated accounts and over 90 per cent of the cash is used in rehypothecation. The amount of leverage that can be achieved through this process is high and its nature gives rise to interconnectedness between financial firms.

Would separating derivatives activity reduce moral hazard? Derivative activity allows banks to grow their bank balance sheet aggressively, build up risks quickly, concentrate risks within the leveraged sector, grow notably short-term debt reliance between financial intermediaries, and make financial intermediaries significantly more intertwined. Derivatives may imply high leverage. By making a small down-payment (or initial margin), banks can take large speculative positions in the market and can transform the riskiness of their assets and income flows, while booking revenues from fees and OTC derivative spreads. Through re-hypothecation, the leverage can be multiplied throughout the financial system.

Would separating derivatives activity reduce conflicts of interest? Derivatives play a critical role in regulatory arbitrage under the Basel capital requirement framework, essentially permitting banks to have a wide discretion in risk-weighting their assets for regulatory capital purposes. Likewise, derivatives were used as a tool to profit from tax arbitrage.

Would separating derivatives activity reduce capital and resources misallocation? Derivatives play a critical role in regulatory arbitrage under the Basel capital requirement framework. The structuring of products via securitisation, swaps, use of seniority tranches and CDS insurance was an integral part of the growth of derivatives transactions.

Would separating derivatives activity impact on competition? If a bank is unable to post the necessary collateral, the risk of transacting with the bank is perceived to go up and other banks will begin to take defensive actions which exacerbate the bank’s weak cash position. Ultimately, without state support, the bank will need to sell assets and unwind trades at fire-sale prices, which will amplify its distress. The moment a bank does not have a sufficient cash buffer, short term securities of sufficient quality, or the ability to borrow to meet collateral calls, it essentially becomes reliant on direct official support. For systemically important banks such public support is always there, and the support itself becomes part of the problem (distortionary implicit subsidies). Asset managers and hedge funds prefer to deal with systemically important banks precisely because the public support can be relied upon. In the absence of such cross-subsidisation support, the cost of capital would have to be much higher. Counterparties to the bank would demand segregated accounts and no re-hypothecation. Securities used for collateral would require higher haircuts. Risk premia for lending collateral would rise. 

6.2. Social costs of separating securitisation and debt origination

Would separating derivatives activity lead to a loss of efficiencies? Derivatives activity has grown aggressively and is characterised by potential diseconomies of scope such as excessive complexity, conflicts of interest, excessive risk taking and increased systemic risk. Subsidiarising derivative activity in a separate trading entity would still allow for market-constrained, limited and genuine innovation to take place within the regulated banking group, whilst not promoting it artificially by linking it to the deposit entity that enjoys public safety net support. Derivative activity has allowed for a rapid expansion of the financial sector through greater interconnectedness. The enlargement of the financial system, beyond a certain size, is associated with reductions in real productivity growth. This, in part, is due to the financial sector competing with the rest of the economy for scarce resources. Excessively large financial systems may reduce economic growth because of the increased probability of a misallocation of resources, the increased probability of large economic crashes, or the endogenous feeding of speculative bubbles.

7. Private equity/Venture capital

Private equity (PE) is an asset class consisting of equity instruments provided to firms that are not publicly traded on an exchange. Private equity is about buying stakes in businesses, transforming business and then realising the value created by selling or floating the business. Because it is equity, it is risk capital allocated to firms for the purpose of funding early stage ventures, growth and diversification opportunities, restructuring and management buy-outs and buy-ins in established companies. Private equity is inherently an illiquid and long-term oriented investment traded only on acquisition and exit. This differs from trading on public markets, trading liquid asset classes such as currencies, stocks, bonds and other derivatives. It typically involves the acquisition of a major stake in a targeted company and comprises also of active management of business operations therein. It can be distinguished from hedge funds which apply various trading strategies to accomplish supra-competitive returns.

One of the parties to a private equity transaction is the fund manager (private equity firm) who is in charge of managing the pooled money in the fund coming from investors and who makes investment decisions. The fund manager can be either a single person or an investment firm. He is responsible for raising funds, sourcing investments and managing them as well as realising capital gains. Another party is the private equity fund which is the investment vehicle pooling money from investors. The target company with its shareholders and management is a third party. In case of leveraged buy-outs, the bank providing debt instruments is a fourth player.

Private equity can generally be seen as a term encompassing the three subgroups venture capital, growth capital as well as management buy-outs and buy-ins.

Venture capital (VC) is that part of private equity that entails finance provided to early-stage, high-potential and possibly, high-growth start-up companies. This commonly covers the seed to expansion stages of investment. The venture capital funds follow an active investment model and provide funding in exchange for management influence and equity in the company invested in. Most often, expertise and experience of the venture capital funds' personnel is one of the main contributing factors. Venture capital firms earn money by owning equity in companies that they invested in which possess novel technologies and/or business models. In the history, most of the venture capital-backed companies have been active in biotechnology, IT, software and such. These firms are too small and their credit history is too limited to acquire debt financing from banks, but their capital need is usually too large to be satisfied by own means. In exchange for the high risk venture capital funds embrace with their investment in young companies, they are rewarded with considerable managerial control and ownership.

Growth capital is the subset of private equity investments aimed at relatively mature companies that need external financing for expanding or restructuring their operations, business diversification, market expansions or for acquisitions. This often involves a minority investment by a private equity firm to companies which are yet unable to generate sufficient funds through debt offerings. Most commonly, growth capital is common or preferred equity and lies at the intersection of mature private equity and venture capital.

The most mature private equity investments are buy-outs and buy-ins where private equity firms target companies to be acquired with equity instruments from current shareholders and restructured in order to be sold off at a later stage for a profit. These companies are usually considered to be fundamentally undervalued because of their unrealized organisational, product or management capacities. Because the targeted companies are rather large, most often these private equity deals are leveraged buy-outs (LBO) where a bank additionally provides debt financing so that the private equity firm can acquire the majority control over the mature company. This varies from venture capital and growth funds which typically do not acquire a majority stake. A special case of these activities are management buy-outs (MBO) or buy-ins (MBI) where the incumbent or an external management team raises funds to acquire a significant share in the company.

The European Private Equity and Venture Capital Association (EVCA) estimated that European private equity funds raised approximately EUR 265 billion in 2007-2012. In Europe over the period 2007-2012, banks made up 11% of all investors in private equity funds, with pension funds having the largest exposure to private equity (21%).

Banks may be involved in two different ways in private equity: as the equity investors (bank-affiliated deals) or as both the equity investor and the debt financier (parent-financed deals). Fang et al. (2012) estimate that bank-affiliated private equity groups account for 30% of all private equity investments in the US.

Private equity is also dealt with in the "Volcker Rule" provision of the Dodd-Frank Act which limits banks' exposure to private equity and hedge funds to no more than 3% of Tier 1 capital. 

7.1. Social benefits of separating private equity and venture capital

Would separating private equity facilitate monitoring, management, and supervision? Would separating private equity facilitate recovery and resolution? Hoenig and Morris (2013) argue that several non-traditional activities, including private equity, are less transparent than traditional banking activities as the success of the underlying investments depends on the opaqueness of the bank's position and on the speed at which their exposures can be changed. Transparency concerns are relevant for private equity. For example, they have led to the adoption of the non-voluntary code of conduct in the UK (Walker guidelines), which require companies to provide the same kind of information to the public that is required for publicly traded companies. The main difficulty with private equity is that valuing private equity investments is inherently difficult for the market. Therefore market monitoring will be less effective, although the relatively long-term nature of private equity mitigates some of the uncertainty. Supervision cannot address all shortcomings as supervisors only have snapshots of various operations while the underlying activities can become very risky in a short time span changing a bank's risk profile. However, as Hoenig and Morris (2012) point out, there is a risk that such activities following separation would migrate into shadow banking leading to even less transparency and monitoring.

The resolvability of banks may be affected as it is difficult to determine reliably the true value of the private equity activities, which raises obstacles to the resale of the bank during resolution. Also, almost all bank-affiliated and parent-financed private equity deals run through special purpose vehicles (SPV) which may complicate these procedures.

Would separating private equity reduce moral hazard? Gilligan and Wright (2012) argue that private equity tries to address the principal agent problem between managers and shareholders, as private equity backed companies do not pay material cash bonuses to senior managers and get a return if the business is sold or floated. On the other hand, despite the long-term nature of private equity investments (compared to proprietary trading) and the informational advantages enjoyed by private equity firms specializing in investing in certain industry sectors, equity investments made by private equity funds remain inherently risky. Private equity firms managing the funds can assess the true valuation of the targeted company only with significant uncertainty, given the start-up nature and/or restructuring required for the target companies. This may lead to significant discrepancy between the price paid for gaining managerial control over a targeted company and the true fair value of the company once taken over. Fang et al. (2012) argue that banks still run serious risks when investing in private equity funds, and have experienced substantial losses. As the potential downside when investing in private equity is high, the potential to reduce moral hazard when separating risky private equity activities from commercial banking is also significant. There is also evidence that private equity performance is highly pro-cyclical. Fang et al (2012) also explain that bank-affiliated deals, even though have similar characteristics and financing compared to stand-alone deals, perform worse if they are made during peaks of the credit market.

Would separating private equity reduce conflicts of interest? Private equity may lead to conflicts of interest between banks and other stakeholders. A bank may take advantage of its superior information about firms as private equity provider to make decisions that benefit the banks at the expense of the target firm (similarly as in underwriting). Furthermore, banks may have stronger incentives to finance in-house deals compared to outside investors as these deals may lead to more cross-selling opportunities to the banks. It can bring additional revenues to banks as the private equity target can be a potential customer for the bank. On the other hand, Fang et al. (2012) explain that banks become exposed to both the equity and debt of the target through private equity and commercial lending and hence that there is a better alignment of stakeholder versus debt-holder's incentives.

Would separating private equity reduce capital and resources misallocation? The removal of the safety net from private equity can restore the level of these activities to the level dictated by the market. Given, however, the significant asymmetric information and other market failures relevant to the financing of companies, and in particular start-ups, it is not obvious that separation would lead to an improvement in capital and resources.

Would separating private equity impact on competition? The reduction in the implicit subsidy in private equity activities would have a positive effect on competition among banks, but relatively limited given the relative size of these activities for banks. Furthermore, it would lead to a level playing field between banks and non-bank-affiliated private equity firms (depending on the extent that the implicit subsidy is removed).

7.2. Social costs of separating private equity and venture capital

Would separating private equity lead to a loss of efficiencies? Economies of scope would be lost when separating private equity activities from the deposit issuing entity that is also active in lending to SMEs. Firstly, as Gilligan and Wright (2012) point out, banks' involvement in private equity activities provides scope for diversification of risk. The Frontier Economics report (2013), commissioned by EVCA, argues that the advantages of private equity include the possibility to diversify portfolios, while earning returns, and having access to otherwise unavailable investment opportunities. Private equity provides institutions (including pension funds, banks and insurance companies) with investment opportunities that they may not be able to pursue otherwise, improving the diversification of their portfolio. Given the attractiveness of diversifying the portfolio and the possibility of earning greater returns from this diversification, the preservation of private equity activities within the deposit-taking bank may provide the potential for advantageous returns and portfolio characteristics for the bank, as well as the bank's customers. A second efficiency relates to the informational advantages that can be exploited by banks' involvement in private equity which stem from relationship banking. Fang et al (2012) claim that through the screening of loans and monitoring banks obtain private information about their clients which they can reuse (and similarly they could use information gathered during past banking relationships to make private equity investment decisions). Thirdly, the banks' engagement as a private equity investor could carry a positive signal about the quality of the investment. This would be credible if the bank has past relationships with the firm and if the bank has a good reputation. Popov and Roosenboom (2012) estimate that while the ratio of venture capital to R&D has averaged around 6% between 1991 and 2005, venture capital has accounted for 9.7% of industrial innovation during that period.

What is the impact on stakeholder groups of separating private equity? Private equity is often argued to be an important source of funding for SMEs. The channelling of fund to SMEs via private equity may make also bring benefits to SMEs through external expertise provided by private equity firms in regards to investments into specific industry sectors. Hence, a separation of private equity activities and the increased funding costs for private equity may have an adverse impact on SMEs. Furthermore, several studies suggest that private equity has an important impact on growth, though innovation and increased productivity.[78]

8. Lending to SMEs and households

Relationship-oriented banking refers to the traditional originate-and-hold model of banking; Banks build up and maintain long relationships with their clients, have an alignment of interests with their clients (they fare well if their customers fare well), have limited scope for trading, and are encouraged to monitor and serve their clients.[79]

Lending to SMEs (as well as syndicated lending and lending to less advanced economies) is relatively intensive in “soft information” that cannot be easily exchanged and sold.

8.1. Social benefits of separating lending to SMEs and households

Would separating lending to SMEs and households reduce moral hazard? Lending to households and SMEs is risky and hence separating the activity would allow shielding deposits from potentially large losses. On the other hand, in normal times the risk is primarily of a non-systemic nature. Loan portfolios are typically relatively granular and obey the law of large numbers and proper risk management. Also, the ability to build up tail risk is not present, to the extent that it is for trading activities. The scalability of lending to SMEs and households is not as prominent as for certain other banking activities.

Would separating lending to SMEs and households facilitate recovery and resolution? Separating lending to SMEs and households would not facilitate resolvability to a significant extent, nor would it facilitate the management, monitoring and supervision of the banking group, given that retail lending does not give rise to complex interconnectedness and intra-financial sector contagion.

Would separating lending to SMEs and households reduce conflicts of interest? Separating lending does not reduce conflicts of interest, because it is a typical example of relationship-oriented banking in which the interests of the lender and the borrower are aligned. The lender does well if the borrower does well, and vice versa.

Would separating lending to SMEs and households reduce capital and resources misallocation? Would separating lending to SMEs and households impact on competition? Separating lending would not result in reduced resource misallocation, given that monitoring small and medium sized borrowers is a core activity of deposit taking banks. Linking savers and borrowers implies that asymmetric information get resolved. If lending and intermediation is left to the market, this is likely to lead to under-provision of credit for SMEs and households.

8.2. Social costs of separating lending to SMEs and households

Individuals are typically risk averse and this characteristic is reflected in their preferences. Those with an excess of funds typically have a preference to lend short, while those with a shortage of funds have a preference to borrow long. Still, in the presence of perfect financial markets (Arrow and Debreu (1954)), there would be no need for maturity-mismatching intermediating banks, since savers and borrowers would execute their transactions directly with sufficiently rewarded and willing counterparties in the financial markets (see also Modigliani and Miller (1958)). So, the true raisons d'être of banks are market imperfections such as information asymmetries, transaction costs, tax distortions and market incompleteness.

Given the existence of market imperfections, there is a role for banks in bringing risk-averse savers and borrowers together. However, banks create a mismatch between the maturity of their assets and liabilities by issuing demandable and other short-term debt and granting long-term loans. Among many others Diamond (1984) and Gorton and Pennacchi (1990) try to understand the exact circumstances under which each of these two separate activities might require the existence of an intermediary, as opposed to being implemented directly through arm's-length financial markets. Although this literature yields many insights, only a few papers address the more fundamental question of why it would make economic sense for a single institution to carry out both functions under the same roof. Real synergies have to exist between the two activities, since if there exist none, there would be no rationale for the existence of loan making and deposit taking banks.

Kashyap et al. (2002) show that, indeed, as long as markets are imperfect, synergies exist between deposit-taking and loan-making activities. They argue that banks offer credit lines or loan commitments to their borrowers, such that the latter hold the option to draw down the loan on demand over a specified period of time. Once the decision to extend a credit has been made, the borrower can show up at any time and withdraw funds, just as with a demand deposit. In that sense, banks provide their customers with liquidity on both the liability and asset side to accommodate their unpredictable needs, extending the original Diamond and Dybvig (1983) argument[80]. Now, given that financial markets are imperfect, a bank cannot accommodate liquidity shocks instantaneously by raising new external finance, so that a buffer stock of liquid assets needs to be held. Holding this buffer is costly for several reasons: opportunity costs, tax distortions, increased agency costs, etc. So, if demand withdrawals and loan draw downs are not perfectly correlated, a real synergy arises and a bank would be able to hold a smaller total liquid asset stock than two separate institutions would have to hold jointly.

Other arguments have also been raised. Dermine (2003a,b) lists several synergies between loan making and deposit taking that lead to real cost reductions. For example, there could be joint operating expenses in delivering deposits and loans, or the terms of mortgage loans could simply require the opening of deposit accounts. Diamond and Rajan (2001) argue instead that banks commit themselves to bearing withdrawal risk by issuing demandable deposits. Hence, the bank will be committed to do the utmost to collect from borrowers to repay depositors. If not, a run might be precipitated and the bank would fail. Similarly, Calomiris and Kahn (1991) argue that deposits may discipline bankers and hence, by submitting themselves to demandable deposits, bankers may attain a lower cost of capital. Finally, Mester et al. (2001) argue that deposits may help banks in monitoring borrowers, thereby becoming superior lenders.

Would separating lending to SMEs and households lead to a loss of efficiencies? Lending and deposit taking naturally belong to each other and separating one from the other would give rise to important efficiency losses. As argued above, Kashyap et al. (2002) show that, indeed, so long as markets are imperfect, synergies exist between deposit-taking and loan-making activities. Dermine (2003a,b) list other synergies between loan making and deposit taking that lead to real cost reductions. For example, there could be joint operating expenses in delivering deposits and loans, or the terms of mortgage loans could simply require the opening of deposit accounts. Diamond and Rajan (2001) argue that banks commit themselves to bearing withdrawal risk by issuing demandable deposits. Hence the bank will be committed to do the utmost to collect from borrowers to repay depositors. If not, a run might be precipitated and the bank would fail. Also, deposits might help banks in monitoring borrowers, thereby becoming superior lenders (Mester et al., 2007).

Given the existence of market imperfections, there is a role for banks in bringing risk-averse savers and borrowers together.

Given its impact on household consumption and SME investment, increasing the private funding costs would give rise to social costs and reduced GDP and economic growth.

9. Lending to large corporates

The assessment and analysis of lending to large corporates is similar to lending to households and SMEs elaborated in section 8, except for the fact that large corporates depend much less on banks for their funding, as they have easier access to debt issuance and capital markets in general. The information asymmetry (and hence market imperfection) is also less prominent, given that more analysts scrutinise large corporates.

As a result, lending to large companies need not be linked to a deposit-issuing entity, but it can be performed by a deposit entity.

10. References

Anand, A. and K. Vankatamaran (2013), “Should Exchanges impose Market Maker obligations?”, http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2179259.

Arrow, K. and G. Debreu (1954), "Existence of an equilibrium for a competitive economy", Econometrica, 22, 3, pp. 265–290.

Benston, G. (1990), The separation of commercial and investment banking: the Glass-Steagall Act revisited and reconsidered, Oxford University Press.

Blundell-Wignall, A., P. Atkinson, and C. Roulet (2012), “The business models of large interconnected banks and the lessons of the financial crisis”, National Institute Economic Review, 221: R31.

Calomiris, C. and C. Kahn (1991), “The Role of Demandable Debt in Structuring Optimal Banking Arrangements”, American Economic Review, 81, 3, pp. 497-513.

Dermine (2003a), “European Banking, Past, Present, and Future”, pp. 31-95 in The Transformation of the European Financial System (Second ECB Central Banking Conference), eds. V. Gaspar, P. Hartmann and O. Sleijpen, Frankfurt, ECB.

Dermine, J. (2003b), "ALM in Banking", Working Paper, INSEAD.

Diamond (1984), “Financial intermediation and delegated monitoring”, Review of Economic Studies, 51, pp. 393-414.

Diamond (1991), "Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt", Journal of Political Economy, 99, 4, pp. 689-721.

Diamond and Dybvig (1983), “Bank runs, Deposit Insurance, and Liquidity”, Journal of Political Economy, 91, pp. 401-419.

Diamond, D. and R. Rajan (2001), “Liquidity Risk, Liquidity Creation, and Financial Fragility: A Theory of Banking", Journal of Political Economy, 109, 2, pp. 287-327.

Diamond, D. and R. Rajan (2001), “Banks and liquidity", American Economic Review, 91, 2, pp. 422-425.

Drucker, S. and M. Puri (2005), „On the benefits of concurrent lending and underwriting“, Journal of Finance, 60, pp. 2763-2799.

Ellis, K., R. Michaely, M. O’Hara (2000), “When the underwriter is the market maker: An examination of trading in the IPO aftermarket”, Journal of Finance, 55, 3, pp. 1039-1074.

Fang, L., V. Ivashina and J. Lerner (2012), “Combining banking with Private Equity investing”, Harvard Business School Finance Working Paper No. 10-106.

Frontier Economics report (2013), “Exploring the impact of private equity on economic growth in Europe”, report prepared for the EVCA, May.

Gambacorta, L. and A. van Rixtel (2013), “Structural bank regulation initiatives: approaches and implications”, Bank for International Settlements, WP No. 412.

Gilligan, J. and M. Wright (2012), “Private equity demystified: 2012 update”, ICAEW.  

Gorton, G. and G. Pennacchi (1990), “Financial intermediaries and liquidity creation”, Journal of Finance, 45, 1, pp. 49-71.

Hartzell, J., J. Kallberg, and C. Liu (2008), ”The role of corporate governance in initial public offerings: Evidence from real estate investment trusts’, Journal of Law and Economics, 51, 3, pp. 539-562.

Hebb, G. and D. Fraser (2002), “Conflict of interest in commercial bank security underwritings: Canadian evidence”, Journal of Banking and Finance, 26, 10, pp. 1935-1949.

Hebb, G. and D. Fraser (2003), “Conflict of interest in commercial bank security underwritings: United Kingdom evidence”, Quarterly Journal of Business & Economics, 42, pp. 79-95.

Hoenig, T. and C. Morris (2011), “Restructuring the banking system to improve safety and soundness”, Federal Reserve Bank of Kansas City, May.

Holmstrom, B. and J. Tirole (1998), “Private and public supply of liquidity”, Journal of Political Economy, 106, 1, pp. 1-40.

HM Treasury (2012), “Banking reform: Delivering stability and supporting a sustainable economy”, June, London.

HM Treasury (2013),“Banking reform: a new structure for stability and growth”, February, London.

JP Morgan research June 2013

Richardson, M. (2013), “Why the Volcker rule is a useful tool for managing systemic risk”, NYU Stern School of Business working paper.

Kanatas, G. and J. Qi (2002), “Underwriting by commercial banks: Incentive conflicts, scope economies, and project quality”, Journal of Money, Credit and Banking, 30, 1, pp. 119-133.

Kang, J-K. and W-L. Liu (2007), “Is universal banking justified? Evidence from bank underwriting of corporate bonds in Japan”, Journal of Financial Economics, 84, 1, pp. 142-186.

Kashyap, A., R. Rajan, and J. Stein (2002), ”Banks as liquidity providers: an explanation of lending and deposit-taking”, Journal of Finance, 57, 1, 33-73.

Kovner, A. (2010), “Do underwriters matter? The impact of the near loss of an equity underwriter”, FRB New York Staff report No. 459.

Kroszner, R. (1998), “Rethinking banking regulation: a review of historical evidence”, Journal of Applied Corporate Finance, 11, 2.

Kroszner, R. and R. Rajan (1994), “Is the Glass-Steagall Act justified? A study of te U.S. experience with universal banking before 1933”, American Economic Review, 84, 4, pp. 810-832.

Mester, L., L. Nakamura, and M. Renault (2007), “Transactions accounts and loan monitoring”, Review of Financial Studies, 20,pp. 529-556.

Modigliani, F. and M. Miller (1958), “The cost of capital, corporate finance, and the theory of investment”, American Economic Review, 48, pp. 261-297.

Mollica, M. and L. Zingales (2007), "The Impact of Venture Capital on Innovation and the Creation of New Businesses," October.

Popov, A., and P. Roosenboom (2012), “Venture Capital and Patented Innovation: Evidence from Europe”, Economic Policy, 28, 71, pp. 447-482.

Popov, A. and F. Smets (2013), “Financial markets: Productivity, procyclicality, and policy”, in Financial Innovation: Too Much or Too Little?, M. Haliassos (ed.), The MIT Press: Cambridge, MA.

Puri, M. (1994), “The long-term default performance of bank underwritten securities issues”, Journal of Banking and Finance, 18, pp. 397-418.

Rajan, R. (1992), “Insiders and outsiders: The choice between informed and arm’s length debt”, Journal of Finance, 47, 4, pp. 1367-1400.

Richardson, M. (2013), “Why the Volcker rule is a useful tool for managing systemic risk”, NYU Stern School of Business Working Paper.

Saunders, A. and I. Walter (1994), Universal banking in the United States, Oxford University Press, New York.

Stein, J. (2002), “Information production and capital allocation: Decentralized versus hierarchical firms”, Journal of Finance, 57, 5, pp. 1891-1921.

Stiglitz, J. (2010), The financial crisis of 2007-2008 and its macroeconomic consequences, time for a visible hand: Lessons from the 2008 world financial crisis, Oxford University Press, pp. 19-49.

Zingales (2012), “Why I was won over by Glass-Steagall”, Financial Times, June 10.

Annex A7 – Strength of Separation

1. Introduction

Building on chapters 4 and 5 of the impact assessment, this Annex will continue the discussion on what forms of separation could be envisaged as well as the strength of the fence, i.e., the combination of measures of a legal, operational, economic, and governance nature that could be adopted to ensure the independence, robustness and effectiveness of the ring-fenced entity.  To decide the optimal strength of the fence, different proposed solutions will be evaluated to see how they would address the specific objectives associated with structural reform.  This is done by evaluating the various proposals against the social benefits and social costs of implementing them. 

The Annex is structured as follows:  section 2 outlines the main forms of separation that are considered in this Impact Assessment (accounting separation, functional separation and ownership separation) and assesses each of these in light of the specific objectives of structural reform (reduce moral hazard, improve resolvability, facilitate monitoring, management and supervision, reduce conflicts of interests, reduce a misallocation of capital, reduce losses of efficiency and impact on competition) and identifies ownership separation and "subsidiarisation"[81] as the options going forward.  Section 3 proceeds to a more in-depth analysis of what subsidiarisation actually means in terms of what existing legislation requires, what changes to current legislation would be required to maintain the integrity of subsidiarisation, and what additional rules can be used, and in what combinations, to reinforce the independence of the separated entity from the rest of the wider corporate group.  Finally,  section 4 aims at specifically evaluate different “ring-fences” in that respect against the social benefits and social costs of implementing them which will provide an answer as to how suitable they may be for addressing the specific objectives of structural reform.  Given the large permutation of possible combinations of additional restrictions on economic, legal, governance and operational links that may exist, this Annex elaborates and assesses a wider range of ring-fencing approaches than the ones compared in the main body of the Impact Assessment.

2. Different forms of structural separation

This Annex presents three broad forms of separation: (1) accounting separation; (2) functional separation through subsidiarisation; and (3) ownership separation (i.e., a prohibition of certain business lines).  These forms of separation display an increasing level of severity and intrusiveness in a financial institution’s structure.

More specifically:

1. Accounting separation means that a financial institution that provides integrated financial services would have to provide separate accounts for each of its different business segments and make these reports publicly available.  This would constitute the lightest degree of structural separation compared to the current status quo. 

2. Functional separation through subsidiarisation is a requirement for a financial institution to transfer certain activities of different business units into a separate, legal subsidiary.

3. Ownership separation implies a prohibition on a financial institution to engage in certain activities.  This is the most intrusive degree of structural intervention as it would involve winding down or selling off assets connected to certain activities to an independent third party.

These three forms of separation are not necessarily mutually exclusive; functional separation, for example, presupposes a degree of accounting separation.  The following sections will discuss each of these forms of separation and evaluate them against the operational objectives of structural reform. 

2.1. Accounting separation

Accounting separation would require financial institutions to provide separate accounts for the main activities they are engaged in (whether through business divisions or subsidiaries).  As a result, the costs and ultimately revenues of each business division or subsidiary (and transfers between them) would in principle be separately identified.  Accounting separation therefore increases transparency and makes it easier to understand different parts of a financial institution and easier to monitor and control potential financial transfers between divisions and subsidiaries.

Accounting separation has typically been used in the utilities sector (e.g., in gas and water companies) to enable the development of competition in contestable parts of the value chain and to prevent cross-subsidies between the various activities of the operator.  Accounting separation is also obligatory for firms that provide services of general economic interest next to any commercial activity.

Currently there are two accounting frameworks applicable for financial institutions in the EU: (1) the International Financial Reporting Standards (the “IFRS”);[82] and (2) the Directive on the annual financial statements, consolidated financial statements and related reports of certain types of undertakings (the “Accounting Directive”).[83]

The IFRS is mandatory for the consolidated accounts of listed companies but EU Member States have the option to apply IFRS also to the consolidated accounts of non-listed companies.  The Accounting Directive applies by default to individual accounts.  IFRS is optional for individual accounts in general. 

The introduction of the IFRS in EU has led an increase in disclosures as IFRS requires more disclosure than the Accounting Directive.  For example, the IFRS requires detailed information on consolidation and fair value measurement but also on different risks borne by the financial institution (e.g., credit risk, liquidity risk or market risk).[84]  More importantly, IFRS standards have led to improvements in the level of quality of disclosures which means that it is now easier to assess a bank's risks. 

The IFRS asks for specific information per business segment (e.g., investment banking, retail, corporate banking, and assets management).[85]  However, companies have leeway to define their segments which means that the definition and the scope of the segments can differ from one bank to another.  Moreover, the information provided is in aggregate form.  Therefore, there is not always enough detail in the financial statements to identify the financial flows either between business segments (e.g., "do retail deposits fund investment banking activities?") or between business divisions and subsidiaries.  It is furthermore not possible to identify where the profit, the cash or the own funds come from within a wider corporate group. 

Because the Accounting Directive leaves Member States some flexibility in the way they transpose the requirements of the Directives into national law (for example with regard to measurement of financial instruments, re-evaluation of tangible or intangible assets), there are now different national, general accepted accounting principles.  As a consequence, it is difficult to compare the financial statements of EU financial institutions when they are based on the Accounting Directive.  The disclosure requirements in the Accounting Directive are also less stringent than those of the IFRS.  For example, disclosure requirements regarding risk or business segments are different. 

In addition, in accordance with the Directive on the access to the activity of credit institutions and the prudential supervision of credit institutions and investments firms (the "CRDIV")[86] all financial institutions must report to prudential supervisors on own funds/capital requirements – “COREP.”  Prudential supervisors also demand additional quarterly reports (“FINREP”) on a consolidated level from “IFRS financial institutions.”  This is optional for non-IFRS banks.  Finally, financial institutions are also required to publically disclose information regarding their prudential calculation (so-called “Pillar 3 disclosures”).[87]

2.1.1. Social benefits and social costs

As explained below, the net social benefits of using accounting separation appear modest.

Would accounting separation facilitate recovery and resolution?  The simplification of and more transparency in a financial institution’s accounts could be one contributing factor for facilitating the drawing up of a resolution plan.  However, accounting separation in itself is highly unlikely to significantly increase resolvability because it does not address the issues that impact on resolvability: for example, nature of activities (e.g., risky and easily scalable), the level of complexity of the legal structure, size and intra-group connectedness.

Would accounting separation reduce moral hazard?  To the extent accounting separation provides more transparency and to the extent information is publicly available this could contribute to more informed choices and monitoring by investors (depositors, creditors, and shareholders).  However, accounting separation does nothing to reduce or remove the incentives to take risk that arise because of the perception of the implicit public safety net; it would most likely have an insignificant impact on the incentives of investors to monitor and control bank risk taking and therefore on the incentives for managers to take excessive risk from the beginning.  The impact on moral hazard would therefore be limited.

Would accounting separation facilitate monitoring, management and supervision?  Accounting separation leads to more information in the public domain which could make it easier for relevant supervisory authorities to understand a financial institution's balancing sheet.  Similarly, it could make it easier for investors to monitor balance sheet activities.  However, given that this degree of separation would not affect incentives to act on information, it is doubtful that accounting would facilitate monitoring, management and supervision to such an extent that incentives for excessive risk taking would change.

Would accounting separation reduce conflicts of interest?  As accounting separation does nothing to actually put a wall between the types of activities which may give rise to conflict of interests it is highly unlikely that it would have any significant impact on reducing conflicts of interests.

Would accounting separation reduce resource and capital misallocation? No. Accounting separation is more a transparency measure. It will most likely have no impact on incentives to allocate capital and human resources to trading and intra-financial activity and away from lending activity.

Would accounting separation lead to a loss of efficiencies? No. This is because accounting separation has no impact on diversification benefits and synergies arising from universal banking.

Would accounting separation impact on competition? Accounting separation in itself is unlikely to address the competition concerns that may follow from an implicit public safety net; bank creditors' perception that the government will intervene to protect them from the risk of bank failure will remain and any distortions of competition arising from those banks who benefit from the implicit subsidy (i.e., lower funding costs) and therefore have the ability to expand at the expense of those banks that do not benefit from the implicit subsidy will remain.

To conclude, accounting separation appears to be ineffective in addressing the operational objectives of structural reform and this form of separation is therefore discarded from further evaluation.

2.2. Functional separation through subsidiarisation

Under functional separation through subsidiarisation financial institutions would have to transfer certain activities to a new legal entity (the "trading entity") or at least ensure that certain activities are carried out in a separate legal entity to insured deposit taking.[88] This trading entity would have to be authorized as an investment firm[89] or, to the extent it would intend to take uninsured deposits as a credit institution.[90] When part of a wider corporate group it would have to be ensured that the trading entity could be swiftly isolated from the entity taking insured deposits (the "deposit entity") should the financial health of the trading entity be at risk.

Subsidiarisation would require the trading entity and the deposit entity to maintain self-standing reserves of capital and of loss-absorbing debt, as well as to comply with other prudential requirements on an individual, sub-consolidated or consolidated basis. Subsidiarisation provides a degree of independence and to some extent also insulates the deposit entity from shocks and losses.  For more on subsidiarisation as a stand-alone fence, see section 4 below.  Moreover, and importantly, subsidiarisation can be coupled with additional rules of a legal, operational, economic, and governance nature to further regulate the relationship between the ring-fenced entity and the wider corporate group.

This form of separation has been followed in different ways in a number of instances in other countries.  For example, in the United States bank holding companies are allowed to provide non-banking financial services (e.g., securities dealing and insurance) under the condition that these services are located in separate subsidiaries.  In a similar manner, the UK Government has also proposed a ring-fence around certain banking activities based on functional separation through subsidiarisation, as have the French and German Governments.[91]

Finally, the act of separating certain activities could take place through the transfer of relevant assets and related liabilities to an existing legal entity (wholly-owned or not) or through the creation of a new legal entity (wholly-owned or not) to which assets and liabilities are then transferred.  Separation would affect shareholder and creditor rights to the extent that current rules provide for their consent for the increase/decrease of capital, and/or for mergers (including cross-border mergers) and divisions.[92]  It cannot be excluded that relevant legislation would therefore have to be amended to ensure that shareholders and creditors could not oppose separation.

2.2.1. Social benefits and social costs

As will be explained below, functional separation through subsidiarisation has high potential for scoring well on addressing the operational objectives associated with structural reform.  This is because subsidiarisation provides a degree of independence and to some extent also insulates the deposit entity from shocks and losses, but also because it can be further strengthened by the addition of various restrictions of legal, operational, economic, and governance nature.  These restrictions can be added in various combinations to further regulate the relationship between the ring-fenced entity and the wider corporate group and strengthen the wall in between them.  It is important to note that the social benefits and the social costs increase with the strength of the fence: The better insulated and the more independent the separated entity is vis-à-vis the wider corporate group and, in particular, the deposit entity, the better subsidiarisation will address the specific objectives, which as a result will impact positively on the social benefits.  However, the stronger the fence is the higher the social costs may also be.  For a more developed discussion and analysis of subsidiarisation and the strength of the fence, see below section 4.

Would subsidiarisation facilitate recovery and resolution? Functional separation could facilitate resolvability and resolution as balance sheets would become smaller and more structured. This could expand the resolution options available to relevant authorities, e.g., it could be easier to divide groups into different parts and sell off/close them down. If the effects of a failure can be functionally insulated, it could also be easier to concentrate supportive funding to those parts that the society wants to support and therefore reduce public sector support for other socially less desirable functions (e.g., retail versus certain trading services). The stronger the fence the more subsidiarisation would facilitate recovery and resolution.

Would subsidiarisation reduce moral hazard? Complying with capital requirements on a stand-alone basis may have the effect of better aligning risk-taking incentives within a banking group. For example, the cost of increasing trading activities would partly be reflected in the stand-alone capital requirements. Banks would not be in a position to expand their trading activities entirely on the basis of capital from the deposit entity. Depending on the degree of restrictions of legal, economic and governance nature, functional separation could significantly curb moral hazard which may have led boards and managers to encourage excessive risk taking as they knew that big losses would be paid largely by taxpayers rather than stakeholders. The stronger the fence the larger the impact would be on reducing moral hazard.

Would subsidiarisation facilitate monitoring, management and supervision? Functional separation could increase transparency and clarity, and reduce complexity in the structure of a group. How much easier market monitoring, management and supervision of risks in the different subsidiaries become will depend on the legal, economic and governance links between the separated entity and the wider group (i.e., the strength of the fence).

Would subsidiarisation reduce conflicts of interest? Subsidiarisation will provide thicker walls between various activities in a group. Again, the more legal, economic and governance links between various group entities are restricted (the higher the fence) the better conflicts of interest could be resolved.

Would functional separation reduce resource and capital misallocation? By separating certain activities from insured deposit taking, the funding of those activities would become more risk-sensitive. As a result, financial institutions would have less of an incentive to encourage excessive risk taking. However, the extent to which funding becomes fully risk sensitive depends on the degree of separation. If the incentives for excessive risk taking are reduced so would the incentives to allocate capital and human resources to trading and intra-financial activity and away from lending activity.

Would functional separation lead to a loss of efficiencies? Subsidiarisation would not necessarily affect a financial institution's ability to provide a universal set of services within a group. Therefore, effects on potential efficiencies resulting from such diverse services may be limited. However, there is a link between the strength of the fence and a reduction in efficiency gains. For example, the requirement to comply with capital requirements on a stand-alone basis would imply additional capital costs for the two entities due to possible diversification of capital effects. Also, the stronger the fence is on requiring subsidiaries to be self-standing and independently funded, the more functional subsidiarisation could reduce efficiency gains stemming from economies of scope and scale.

Would subsidiarisation impact on competition?

Similar to accounting separation, subsidiarisation in itself and without additional restrictions is unlikely to address the competition concerns that may follow from an implicit public safety net. This is because it does not address bank creditors' perception that the government will intervene to protect the too-big-to-fail banks ("TBTF" banks) from the risk of failure. Therefore creditors have no incentive to demand banks to pay the right compensation for bearing risks, which in turn lowers funding costs for the TBTF banks. This enables the latter to benefit from lower funding costs which in turn enables them to distort competition by expanding at the expense of those banks that do not benefit (or benefit less) from the implicit subsidy.  Moreover, because different Member States have a different ability and willingness to intervene to rescue banks, competition among Member States (and wider, on an international level) also remains distorted.

2.3. Ownership separation

Ownership separation is a prohibition on those financial institutions that are caught by the relevant thresholds, and therefore subject to structural reform (the "affected institutions"), to engage in certain activities, even through legally separated affiliates.  This is the most intrusive form of structural separation as the affected institutions would accordingly have to wind down or divest any assets supporting such activities as well as the liabilities related to the assets that were up for divestiture.  If divested, those assets and related liabilities would have to be sold off/transferred to a completely different legal and structurally separate entity with which the affected institutions could have no links of any nature.  This would impact on the universal banking model to the extent that not all activities could be provided by one and the same financial institution.

Ownership separation could be accomplished by way of division of assets, transfer of all assets and liabilities through a merger or an acquisition or through a spin off that would lead to the creation of an independent company through the sale or distribution of new shares of an existing business/division.

Whichever way, it cannot be excluded that changes to current EU law would be required to ensure that ownership separation could not be opposed; For example, current rules that: (i) require shareholders’ approval of any increase or reduction of capital;[93] and (ii) require approval by the General Meeting of merger/cross-border merger/divisions.[94]  Moreover, CRDIV currently provides that Member States can object to the change of ownership of a bank only on certain prudential grounds.  A ban on the acquisition of a deposit bank by an investment bank – for the reason that it is an investment bank – could well be held to breach the obligation on the competent Member State authority to object to an acquisition only on the limited set of prudential criteria set out in Article 23, which does not include the criterion that an acquirer of a deposit bank is not an investment bank.

Imposing ownership separation on financial institutions could also have an impact on the EU Internal Market.  In particular, ownership separation could limit the freedom to provide services to the extent that the affected institution would no longer be permitted to carry out certain activities in combination with insured deposit taking.  Finally, the right to property is a fundamental right protected in the Charter of Fundamental Rights

2.3.1. Social benefits and social costs

As the below description will illustrate, there are good arguments why ownership separation may lead to high social benefits.  It could, however, also lead to significant losses of economies of scope and trigger a migration of certain activities toward non-bank credit intermediaries (so-called "shadow banks"), which could imply high social costs.

Would ownership separation facilitate recovery and resolution?  Ownership separation would facilitate resolution mainly because certain risks linked to the prohibited activities would no longer feature on the balance sheets of the financial institution.  Depending on the scale of the relevant activities, banks may also become less complex and smaller in size as a result of divesting them or winding them down.

Would ownership separation reduce moral hazard?  The primary advantage of ownership separation is that, while it may be complex to implement, it is an “unclouded” solution which after completion could alleviate the need for further regulation within the context of structural reform.  Ownership separation could effectively remove the implicit state subsidy from those financial institutions that are TBTF.  This would remove the related moral hazard, depending on the extent of the ownership separation, i.e., the activities to be separated.

However, ownership separation could give rise to more homogeneous and less diverse entities and business models and allow for less economies of scope, such as risk diversification.  It can therefore not be excluded that completely separated entities may fail more, and, if so, more as a herd, which would complicate system-wide resolution.

Would ownership separation facilitate monitoring, management and supervision?  Monitoring, managing and supervision could be facilitated, as the activities that are most scalable and complex and consequently difficult to monitor would be located outside the financial institution.

Would ownership reduce conflicts of interest?  The impact on conflicts of interests will depend on the range and type of activities that are prohibited from the group.  See Annex A6 for a discussion of the different activities.  Depending on the scope of the activity prohibition, conflicts of interest may be significantly reduced through ownership separation.

Would ownership separation reduce resource and capital misallocation?  With the most risky activities no longer part of its balance sheet, a financial institution could be able to focus on and allocate more resources to lending to the real economy.

Would ownership separation lead to a loss of efficiencies?  The primary disadvantage is the loss of potential economies of scope from integration within the universal banking model.  Economies of scope consist of risk diversification, cost economies of scope, and revenue economies of scope.  Economies of scale would be present to the extent there are significant fixed costs for some operations, even though diseconomies of scope can arise.  Similarly, economies of scale are found to be exhausted at relatively low levels of bank assets, and thereafter diseconomies of scale become also important.

As a result of the loss in economies of scope (and scale), private funding and capital costs increase with the strength of separation and hence would be the highest under ownership separation.  If some of these activities perform an important role in the economy these additional costs may have further efficiency effects in other areas of the economy.

Would ownership separation impact on competition?

Ownership separation could eliminate implicit cross-subsidies and introduce effective market discipline on the trading entity.  This would have the effect of evening out the level playing field among small, medium-sized and large banks not only within a single Member State but also across Member States (and wider).

Finally, it cannot be excluded that divested activities will migrate from affected institutions to shadow banks where there may be less scope for control by supervisors (whether or not located with the EU).  Work done by the Financial Stability Board (the "FSB") has highlighted that the disorderly failure of shadow bank entities can carry systemic risk, both directly and through their interconnectedness with the regular banking system.  The FSB has also suggested that as long as such entities remain subject to a lower level of regulation and supervision than the rest of the financial sector, reinforced banking regulation could drive a substantial part of banking activities beyond the boundaries of traditional banking and towards shadow banking.  The European Commission has recently adopted a Communication setting-out a roadmap for tackling the risks inherent in shadow banking.  The measures foreseen in this roadmap, including a series of regulatory measures such as a framework for the interaction between banks and the shadow banking sector, are intended to ensure that the potential systemic risks to the financial sector are covered and that the opportunities for regulatory arbitrage are in limited, in order to strengthen market integrity and increase the confidence of savers and consumers.  This includes tightening the prudential rules applied to banks in their operations with unregulated financial entities in order to reduce contagion risks.[95]

2.4. Ownership separation and functional separation through subsidiarisation are the preferred options going forward

Based on the above evaluation of each of the different forms of functional separation the following conclusions are appropriate:

· While accounting separation would leave potentially existing economies of scale and scope intact, it would have no or only limited impact on any of other criteria used to measure social benefits and social costs. It will therefore not be subject to any further analysis.

· Ownership separation scores very high on social benefits but also come at high social costs. There is also a risk which is not insignificant that some of the prohibited activities might migrate to the shadow banking sector.

· Conversely, subsidiarisation would come at a comparably lower social cost than ownership separation. As section 4 illustrates there is also a case that subsidiarisation can deliver similar social benefits as ownership separation at a lower social cost (this lower cost may vary depending on the strength of the fence). The additional costs of ownership separation as opposed to subsidiarisation would therefore need to be justified by additional social benefits on an activity by activity basis. For example, as highlighted in Chapter 5, as regards ownership separation of proprietary trading, the balance between benefits in terms of e.g., facilitating resolvability and reducing moral hazard and conflicts of interest clearly outweigh the costs in terms of foregone efficiency, as these are very limited.

· While individual weightings might differ, the sum of social costs and benefits of ownership separation and subsidiarisation may be more or less equal and they are both candidates for the preferred policy option.  Given, however, that subsidiarisation is a wider concept and can exist in various degrees the remainder of this Annex will focus on subsidiarisation and the determination of the appropriate degree/strength of subsidiarisation.

3. Subsidiarisation

The remainder of this Annex will discuss functional separation through subsidiarisation in more depth to see how this form of separation can best address the specific objectives of structural reform.[96]  In particular, the remaining part will discuss various "degrees" of subsidiarisation that can be achieved by adding rules in various combinations to reinforce the fence between the separated entity and the rest of the wider corporate group.

To that end, section 3.1 aims at describe what the simplest form of separation through subsidiarisation actually means in terms of how current key banking legislation on prudential requirements and resolution apply to the separated entities.  Section 3.2 discusses more normatively what changes to current legislation could be envisaged to secure the effectiveness of subsidiarisation and limit that prudential concerns arise.  Section 3.3 lists additional rules that can be added in various combinations on top of current legislation to build on subsidiarisation and reinforce the independence of the separated entities and the wall in between them (i.e., to build a stronger fence between the separated entity and the rest of the group, in particular the deposit entity).  Finally, section 4 aims at determine the appropriate strength of the fence by specifically evaluating the social benefits and costs of various combinations of rules (different fences) section.

3.1. Subsidiarisation under current regulatory framework

In accordance with existing legislation regarding prudential requirements and recovery and resolution,[97] the following would apply automatically in case of subsidiarisation:

Separate management bodies: Under subsidiarisation, the trading entity and the deposit entity will have their own separate management bodies[98] and follow the requirements regarding composition and selection of members set out in the CRDIV.[99]

Prudential requirements: In accordance with CRDIV and the Regulation on prudential requirements for credit institutions and investment firms (the "CRR")[100] the general rule is that the basic prudential requirements with regard to own funds, capital, leverage (as of 2019), liquidity, large exposures[101] and prudential reporting apply on an individual and consolidated basis.[102] This means that both the trading entity and the deposit entity would have to abide by these rules on an individual or consolidated basis. The CRR also provides competent Member State authorities under certain circumstances with some discretion to impose additional capital requirements in the form of buffers on so-called global and other systemically important institutions. This may include both the trading entity and the deposit entity.[103]  

Bail-in tools: The BRRD requires that each credit institution and investment firm maintains a sufficient amount of liabilities in their balance sheet that could be subject to bail-in powers.[104]

Capital transfers and dividend payments: According to CRDIV, capital transfers and dividend payments are allowed only when it can be established that the institution has sufficient financial resources to do so and when this does not lead to a breach of any prudential requirements regarding, for example, capital. This means that capital transfers and dividend payments from the deposit entity to the parent or the trading entity could only be effectuated under those circumstances.[105]

Corporate governance arrangements: The CRDIV sets out the basic rules regarding the management body and other governance arrangements. In particular, the CRDIV provides that each entity within a wider corporate group has to have its own risk management strategy which must cover measures to manage, monitor and mitigate risks which the institution is and might be exposed to. This risk management strategy needs to be approved and implemented by each separate legal entity's management body.[106] The trading entity shall also formulate its remuneration policies in line with the rules set out in the CRDIV which require that the remuneration policy is consistent with and promotes effective risk management, is in line with business strategies and long-term interest of the institution.[107]

Disclosures: Where the trading entity is part of a wider corporate group and to the extent it constitutes a “significant subsidiary” of an EU parent institution,[108] it shall disclose information regarding its own funds, capital requirements, capital buffers, credit risk and credit risk mitigation techniques, remuneration policy, leverage on an individual or sub-consolidated basis.[109] When the trading entity is not part of a wider corporate group all disclosures and prudential reporting requirements obviously apply on an individual basis.

Recovery plans: Finally, according to the BRRD, group recovery plans as well as individual recovery plans for each institution that is part of the group need to be drawn up which means that each of the trading entity and the deposit entity will have to draw up its own recovery plan.[110] This includes identifying arrangements for cooperation and coordination with relevant authorities in third countries where a group includes entities incorporated abroad.[111]

3.2. What amendments to the current regulatory framework could be envisaged to enhance the effectiveness of subsidiarisation?

As described in the previous section, subsidiarisation itself already provides a certain degree of separation in terms of legal, economic and governance links with the rest of the group. Subsidiarisation therefore insulates the deposit entity from shocks and losses to a certain extent. To enhance the effectiveness of subsidiarisation and reduce chances that prudential concerns arise, a number of amendments to existing legislation could be envisaged. These amendments relate to legislation regarding prudential requirements and recovery and resolution.

Amendments ensuring adherence to the prudential requirements on a sub-consolidated basis: The requirement that prudential obligations must be fulfilled on an individual basis can be waived under current legislation and under certain conditions, for example for a subsidiary or a parent established in the same Member State as the group if certain conditions are met.[112] To ensure the effectiveness of the subsidiarisation it could be envisaged to prescribe that obligations should apply at sub-consolidated level – i.e., among entities belonging to the same sub-group.[113] It would also have to be clarified that it can only be applied among similar entities. Doing so will require changing those provisions in the CRR that provide for derogation to the application of prudential requirements on an individual basis. With regard to intra-group large exposures in particular it should also be envisaged to specifically prescribe that the limits currently set out in Article 395(6) of the CRR will not apply once an EU law is adopted and that large exposures rules shall apply as between entities belonging to different sub-groups of similar entities.

Amendments ensuring that disclosures are made on an individual or sub-consolidated basis among similar entities: In order to ensure the effectiveness of monitoring and supervision of the separated entity, it could be envisaged to define the trading entity as a "significant subsidiary" of an EU parent within the meaning of Article 12 of the CRR because otherwise disclosure requirements regarding own funds, remuneration policy and leverage ratio apply on a consolidated basis for the wider corporate group.

Amendments ensuring that the deposit entity is not made liable for costs related to resolution actions involving the trading entity: In view of recovery and resolution, the BRRD currently allows entities within a wider corporate group that are covered by the supervision of the parent undertaking to enter into agreements with each other to provide financial support to any party of the agreement that experiences financial difficulties.[114] Moreover, group resolution plans shall identify how the group resolution actions could be financed and, where appropriate, set out principles for sharing responsibility for that financing between sources of funding in different Member States.[115]

In principle, the agreement to provide financial support may only be concluded if the supervisory authority considers none of the parties in breach of the CRDIV in relation to the rules on capital or liquidity, or is at risk of insolvency. However, to avoid that the deposit entity becomes liable for costs related to resolution actions involving the trading entity which may put undue stress on the balance sheet of the deposit entity it could be envisaged to specifically set out in legislation that for the deposit entity such agreements can by their nature jeopardize the liquidity or solvency of the deposit entity or create a threat to financial stability and therefore cannot be concluded. Similarly with regard to the shared financial responsibility for resolution actions, it should be specified that the deposit entity should not be liable for any costs related to resolution actions involving the trading entity. Both these clarifications would require amendments to the BRRD.

                                                            *          *          *         

In addition to the above amendments, to safeguard the objectives of subsidiarisation, there are a number of additional rules that could be used in various combinations to further strengthen the separation between the trading entity and the rest of the group, in particular the deposit entity. The following section lists a number of such rules. These rules are not mutually exclusive.

3.3. Potential additional rules to further strengthen separation 3.3.1. Rules to achieve stronger legal and operation separation

To further strengthen the insulation of the trading entity, to avoid an unrestricted mixture of activities, to limit that liabilities or potential liabilities are shifted around between entities within the wider corporate group, and to limit conflicts of interests, one or several of the following additional rules could be added to the subsidiarisation:

Intra-group ownership restrictions:

· The trading entity is not allowed to hold shares or voting power in a deposit entity. The purpose of this rule would be to avoid the creation of potential conflicts of interests and culture shocks that may affect banking standards, and to reduce risks related to moral hazard that may arise from funding and subsidy leakages that can occur between the different entities; and/or

· The deposit entity is not allowed to hold shares or voting power in a trading entity. The purpose of this rule would be to avoid that the deposit entity has to bear any losses related to the trading entity; and/or

· The trading entity is not allowed to hold shares or voting power in critical infrastructure needed for the operation of the deposit entity. The purpose of this rule would be to ensure the continuity of vital banking services provided by the deposit entity in case the trading entity fails.

Operational continuity:

· Another option could be that the wider corporate group, including the parent, would have to ensure the operational continuity of the deposit entity under all circumstances; irrespective of the financial health of the trading entity. This would include but not be limited to: access to critical infrastructure, staff, data and information, and services.

3.3.2. Rules to achieve stronger economic separation

To further ensure that subsidiarisation results in separate and independent entities that can effectively operate as stand-alone entities and to increase transparency for supervisors and market participants, the following requirement could be considered in addition to subsidiarisation:

· Separate debt issuance: Both the deposit and trading entities could issue their own debt (i.e., independently from each other and any other entity, including the parent, in the wider corporate group). The purpose of this rule would be to avoid any effects of increasing contagion risk within a group in the event of distress.

· Disclosures: The trading entity and the deposit entity could make all financial and supervisory disclosures on an individual or sub-consolidated basis. The purpose of this rule would be to facilitate monitoring, management and supervision by investors, creditors, managers and supervisors.

Specifically on reducing intra-group interconnectedness:

To further ensure the insulation and independence of the trading entity when it forms part of a wider corporate group and to limit contagion risks, the following rules, which in various combinations contribute to reducing the interconnectedness among group entities, could be considered with regard to interactions between the trading entity and the rest of the group – including the parent:

Exposures[116] to an individual group entity or group of entities:

· Intra-group exposures could only be allowed between "similar" entities. The purpose of this rule would be to remove any contagion risks between the trading entity and the deposit entity as it would effectively result in a zero per cent large exposure limit to any intra-group exposures between trading entities and entities not belonging to the same sub-group; or

· A large exposure limit could apply to the exposures of all deposit entities to each trading entity or wider to include also other financial sector entities (the limit is applied on a sub-consolidated basis and equal to [x]% of the eligible capital of the sub-group of deposit entities). This limit could apply on a net basis and after taking account of credit risk mitigation techniques. The purpose of this rule would be to reduce contagion risks from trading entities to deposit entities; or

· A large exposure limit could apply to the total amount of exposures of all trading entities to all deposit entities (the limit is applied to aggregated exposures on a sub-consolidated basis and equal to [x]% of the eligible capital of the sub-group of trading entities). This limit could apply on a net basis and after taking account of credit risk mitigation techniques. The purpose of this rule would be to limit contagion risks from deposit entities to trading entities.

Treasury Management:

Treasury management plays a pivotal role in the running of a financial institution as it is responsible for the management of the institution’s capital, liquidity and funding. However, to ensure an appropriate relationship between the trading entity and the deposit entity that is in the spirit of structural reform, the following rule could be considered:

· The trading and deposit entities should have their own separate treasury management. The purpose of this rule would be to limit the deposit entity’s dealing in investments and to avoid any potential conflicts of interest that may arise.

Contracts and other transactions:

To ensure that both entities are acting in their own self-interest and are not subject to any pressure or duress from the other party, and to increase transparency and facilitate monitoring and supervision, one or several of the following rules could be considered:

· All transactions within the group (including, if relevant, the parent) could be made at arm’s length commercial basis. The purpose of this rule is to avoid undue influence by one entity over the other and to increase transparency;[117] or

· Only transactions between the trading entity and the deposit entity could be at arm's length commercial basis. The purpose of this rule would be to avoid undue influence by the trading entity over the deposit entity and to increase transparency.

· Extensions of credit and guarantees to the trading entity from the deposit entity could be secured by high-quality collateral. The purpose of this rule is to act as a mitigant to any prudential concerns.

Intra-group financial support:

In view of recovery and resolution, one option could be that the trading entity is not allowed to enter into agreements to receive or in any other form, direct or indirect, benefit from financial support from the deposit entity (or only if it would be absolutely satisfied that it can do so without putting into question the "integrity" of the separation).[118] The purpose of this rule would be to ensure that the deposit entity cannot be burdened by financial obligations of the trading entity that ultimately may put at risk its own solvency.

Resolution actions:

With regard to the financing of groups' resolution actions, one option could be that the deposit entity should not be liable for any costs related to resolution actions involving the trading entity.[119] The purpose of this rule would be to ensure that the deposit entity cannot be burdened by financial obligations toward the trading entity that ultimately may put at risk its own solvency.

Specifically on reducing interconnectedness between banking groups

To reduce the interconnectedness between banking groups, to mitigate systemic contagion risks and protect the insured deposits, one or several of the following rules could be considered:

Exposures to an individual counterparty or groups of connected counterparties:

· Exposures of all entities within the wider corporate group to financial institutions not part of the same wider corporate group could be limited beyond current rules regarding large exposures. The purpose of this rule would be to limit the effects of a financial institution’s failure on the entire financial system ; or

· All the exposures of all trading entities within a wider corporate could be limited. The purpose of this rule would in particular be to limit the effects of a trading institution’s failure on the entire financial system.

· To the extent that an ownership separation of certain activities was to be combined with subsidiarisation, one could consider limits (including aggregate limits) to the large exposure regime for the deposit entity toward the trading entity or any financial sector entity.  This could be particularly important for the deposit entity to manage risk especially in cases where an ownership separation of a significant amount of trading activities that give rise to major risks could cause a migration of such activities to the shadow banking area where they are more likely to remain outside the scope of supervision.[120]

3.3.3. Rules to achieve stronger corporate governance

As described in the Commission's Staff Working Document on Corporate Governance in Financial Institutions, the latest financial crisis revealed serious flaws and shortcomings in board performance at a number of financial institutions.[121] Corporate governance requirements are intended to encourage a bank to be well managed. Good corporate governance may therefore, in an indirect way, help limiting the probability of failure and also mitigate the impact of a failure.

To strengthen the separation of the trading entity, to limit conflicts of interest and to ensure that the trading entity has a relationship with the rest of the group that is in the spirit of structural reform, to align incentives, while still leaving flexibility for the parent to deliver group strategies, one or several of the following rules could be considered:

The composition of the management body:[122]

· A sufficient number of management body members should be non-executive members. The purpose of this rule would be to reduce conflicts of interests and provide for more independence within a management body;[123]

· Another option could be that there should be no (or limited to a minority) cross-membership between the management body of the trading entity and that of the deposit entity. The purpose of this rule would be to reduce conflicts of interest and culture between the two entities; and/or

· Another option could be that there should be a limit to the number of management body members from the parent sitting on the management bodies of the trading entity and the deposit entity. The purpose of this rule would be to limit undue influence of the parent over the deposit entity and to reduce conflicts of interest; and

· The same parent member is not allowed to sit on both the trading entity and the deposit entity management body. The purpose of this rule would be to reduce, if not avoid, parental influence over the deposit entity and to reduce conflicts of interest.

The duties of the management body:

· One option could be to impose a statutory duty for managers and all management bodies (including the parent's) to uphold the objectives of the separation. The purpose of this rule would be to make the duty to uphold the objectives of the separation a fiduciary duty to ensure that business strategies are aligned with the objectives of the separation;[124]

· Another option could be to state that the management body of the deposit entity and the parent should have a statutory duty to protect depositors. The purpose of this rule would be to make the duty to protect depositors a fiduciary duty to ensure that the deposit entity’s business strategies are aligned with the objectives of the separation.

Governance structures, Remuneration:

· There could be a statutory duty for the trading entity and the deposit entity to ensure that the internal remuneration policy is in line with the objectives of the separation. The purpose of this rule would be to ensure alignment between remuneration policies, risk management and internal control systems.

4. Determining the appropriate strength of separation

The previous section identified a number of rules that could be used in the context of subsidiarisation and explained what the purpose of each rule would be when building a fence. This section evaluates how various groups of rules (legal, economic and governance) would address the specific objectives associated with structural reform. Although a great variety of groups could surely be envisaged the subsequent  section will be limited to assessing three different models of fences displaying various degrees of strength ranging from limited separation (subsidiarisation only) to very strong separation (subsidiarisation ++). This range constitutes the most representative picture of how rules could be played around with to establish an appropriate fence. The assessment will include an evaluation of each of these models measured against the social benefits and social costs[125] of implementing them to see how far they go toward addressing the microeconomic objectives of structural reform.

4.1.  Social benefits and social costs linked to legal, economic and governance rules 4.1.1. Social benefits and social costs of legal rules

The legal rules discussed in 3.3.1 are basically rules prohibiting certain ownership: (i) the trading entity cannot hold shares or voting power in a deposit entity; (ii) the deposit entity cannot hold shares or voting power in a trading entity; and (iii) the trading entity cannot hold shares or voting power in critical infrastructure needed for the operation of the deposit entity.

· Would the legal rules facilitate recovery and resolution? Ownership restrictions can have the effect of further insulating the trading entity from the deposit entity in terms of funding; more insulation enhances resolvability without taxpayer support. It may also reduce complexity in resolvability. Moreover, if there were a sibling structure entirely different resolution tools could be deployed for the trading entity and the deposit entity. The risk of contagion would also be reduced as the parent/holding company would act as a firewall between the two sister companies

· Would the legal rules reduce moral hazard? Ownership restrictions are consistent with the principle of insulating the trading entity from the deposit entity that structural reform seeks to achieve through functional separation. The more independent the trading entity is and the more intra-group relationships are assimilated to relationships with third parties, the more credible it may be to the market place that the trading entity will not benefit from government support in case of financial trouble. This can discipline any incentives for excessive risk taking as well as ensuring better alignment of incentives. Ultimately, however, the effect on moral hazard will also depend on the extent other parts of the intra-group relationship are regulated (i.e., see the discussion below regarding economic and governance links).

· Would the legal rules facilitate monitoring, management and supervision? More insulation of the trading entity from the deposit entity would mean that the objectives and needs of each separate entity would be even clearer and therefore easier for management to implement and supervise. Reducing the intra-group interconnectedness will make each entity's accounts more clear and likely more transparent, which will facilitate supervision.

· Would the legal rules reduce conflicts of interest? Different legal entities with separate management bodies with their own separate objectives can reduce conflicts of interest.

· Would the legal rules reduce resource and capital misallocation? On its own there is no guarantee that an ownership restriction would significantly disincentivize the allocation of capital and human resources to trading and intra-financial activity and away from lending activity. It will depend on the entire group structure and what other measures there would be to curb incentives to take excessive risks.

· Would the legal rules lead to a loss of efficiencies? Not on its own; it would depend on what additional rules may apply among entities within the wider corporate group and that may restrict to a varying degree efficiencies of scale and scope to be shared among group entities. However, the legal rules under consideration would have a significant effect on legal corporate structures and holding patterns within the wider corporate group.

Would the legal rules impact on competition? Generally, legal rules complement economic and governance rules and, taken together, they may credibly remove the implicit subsidy and improve competition. However, an ownership separation applied across an entire sector would be the most effective and credible way to eliminate the implicit public safety net and could therefore improve the level playing field.

As is illustrated above, an ownership restriction on its own can be effective with regard to certain issues. Its effectiveness will, however, increase if it is imposed on both the trading and the deposit entity and if it is coupled with additional rules that can target other parts of intra-group relationships.

4.1.2. Social benefits and social costs of economic rules

The economic rules discussed in  section 3.3.2 concerned: (i) separate debt issuance; (ii) financial and prudential disclosure requirements; (iii) large exposure rules (both within a banking group and among different banking groups); (iv) treasury management; (v) arm's length base approach in contracts and other transactions; and (vi) intra-group financial support for resolution purposes.

· Would the economic rules facilitate recovery and resolution? The effect of making in particular the trading entity more independent of the other entities in the wider corporate group in the sense that it is not reliant on resources provided by the wider corporate group is that it enhances resolvability without taxpayer support.

· Would the economic rules reduce moral hazard? The more economically independent in particular the trading entity is of the other entities in the wider corporate group in the sense that it is not reliant on resources provided by the wider corporate group, the greater the incentives of investors (depositors, creditors, and shareholders) to monitor and control bank risk taking, which in turn decreases incentives for managers to take excessive risk as they would have to bear the consequences of their own risk actions.

· Would the economic rules facilitate monitoring, management and supervision? The more the trading entity's funding has to come from external capital markets the more shareholders and creditors would be incentivized to monitor risk taking. Moreover, reducing the intra-group interconnectedness will make each entity's accounts more clear and self-contained and likely more transparent which will facilitate supervision.

· Would the economic rules reduce conflicts of interest? On their own, economic rules most likely have a limited impact on reducing conflicts of interests.

· Would the economic rules reduce resource and capital misallocation? The more economically independent in particular the trading entity is of the other entities in the wider corporate group in the sense that it is not reliant on resources provided by the wider corporate group, the more its funding will be risk sensitive and also more costly. This may reduce incentives to invest in projects which are not worthwhile on average, but risky enough to have at least some chance of making money, which could lead to a better allocation of capital in the economy.

· Would the economic rules lead to a loss of efficiencies? In principle, the imposition of economic rues could lead to some losses of economies of scale and scope. However, imposing additional economic rules does not prevent the bank from deciding to place many activities on either side of the fence, excess capital could under certain conditions still be transferred among the entities within the wider corporate group and customers could still use a single group for all of their services, which means that a certain level of diversification benefits would still remain (for more on economies of scale and scope see Annex A9). Similarly, groups may continue to benefit from operating cost reductions from pooling certain resources such as, for example, IT and finance systems and from shared marketing and advertising campaigns. Even so, the introduction of separate debt issuance and individual/sub-consolidated application of prudential requirements would be associated with a cost for the bank.

· Would the economic rules impact on competition? To the extent that economic rules could successfully and credibly remove the implicit public subsidy there would be scope to argue for an improved level playing field. However, it is doubtful that economic rules on their own would achieve that.

As is illustrated above, economic rules can be effective in terms of addressing the operation objectives of structural reform. The effectiveness may depend on the combination of economic rules imposed and, as for the legal rules, the effectiveness will increase when coupled with also other types of rules.

4.1.3. Social benefits and social costs of governance rules

The governance rules discussed in section 3.3.3 concerned: (i) the composition of the management body; (ii) the duties of the management body; and (iii) remuneration.

· Would the governance rules facilitate recovery and resolution? Governance rules are unlikely to have any significant impact on the resolvability of entities.

· Would the governance rules reduce moral hazard? Rules governing not only the composition of management bodies but also the fiduciary duties of members of the management body and senior management could have a rather significant impact on disciplining incentives for excessive risk taking.

· Would the governance rules facilitate monitoring, management and supervision? Governance rules regarding duties to uphold the objectives of the separation could significantly impact on the way it operates with regard to excessive risk taking. This could mean that the objectives and needs of each separate entity would be even clearer and therefore easier for management to implement and supervise.

· Would the governance rules reduce conflicts of interest? Governance rules would be of significant importance to reduce conflicts of interest and of culture.

· Would the governance rules reduce resource and capital misallocation? To the extent governance rules have an impact on incentives they are of some relevance.

· Would the governance rules lead to a loss of efficiencies? No. Long-term, dedicated governance to specific business lines may improve managerial attention on mastering associated risks, and running the specific business line in the most effective way possible.

· Would governance rules impact on competition? However, it is doubtful that governance rules on their own would have enough impact on the implicit public subsidy to be able to significantly improve the level playing field within Member States and across Member States (and wider).

As is illustrated above, governance rules can be very effective with regard to in particular impact on moral hazard, monitoring and supervision and managing conflicts of interest. The effectiveness will, however, increase when coupled with additional rules of both legal and economic nature.

4.1.4. Conclusion on social benefits and social costs

On the basis of the above, the following table summarises how effective and efficient rules of legal, economic and governance nature can be in achieving the specific objectives. This clearly highlights that the different rules are largely complementary, as they achieve the specific objectives in different ways. For example, rules providing for a stricter economic separation are particularly effective in reducing moral hazard, whereas rules providing for a stricter degree of legal separation are particularly helpful in facilitating resolution, and stricter governance rules address conflicts of interest particularly effectively. As could be expected, the stricter the legal and economic separation, the higher the private cost in terms of foregone efficiency (to the extent economies of scale and scope result in such efficiency). Governance separation, while complementary to the other restrictions, would in isolation not be associated with the same efficiency loss.

Table 1: Mapping the effectiveness and efficiency of legal, economic and governance separation

Objective || Moral hazard || Recovery and resolution || Monitoring, management, supervision || Conflicts of interest || Capital misallocation || Efficiency loss

Legal rules || + || ++ || + || + || ≈ || +

Economic rules || ++ || ++ || +/≈ || ≈ || ++ || +

Governance rules || +(+) || ≈ || + || ++ || ≈ || ≈

4.2. Evaluation of different models of a fence

As illustrated above, each category of rules has a complementary role to play. In providing a robust degree of separation, a combination of rules in these different categories is accordingly likely to be necessary. However, the rules can be combined in many different ways. To ensure an effective as well as efficient assessment, the rest of this Annex assesses three different packages that combine these rules in different ways. More specifically:

· Separation 1 - subsidiarisation: This implies a limited, low fence and is based on subsidiarisation only in according with current legislation and under the assumption that amendments to current legislation suggested in section 3.2 have been implemented. This provides for a degree of separation, mainly of economic nature. This fence constitutes the base from which all other fences depart and is the least “intrusive” for financial institutions because it leaves them more flexibility to determine their internal structure and business models. This fence is limited to separating certain activities to a separate legal entity which (along with the deposit entity) will be subject to prudential and other rules on an individual, sub-consolidated and consolidated basis as described in sections 3.1 and 3.2.[126]

· Separation 2 – subsidiarisation+: Separation 2 provides for a stricter economic separation (e.g., separate debt issuance and arm’s length relations). Moreover, it also provides for a somewhat stricter degree of legal (deposit entity not to own trading entity) and governance separation (e.g., cross-membership of boards) where the trading entity forms part of a wider corporate group;

· Separation 3 – subsidiarisation++: This would provide a fence of maximum strength to protect the objectives of the separation. Accordingly, this model would provide for stricter independence of not only the separated trading entity but also the deposit entity. This would include a provision that a trading entity could not own a deposit entity (in addition to the prohibition on deposit entities owning trading entities). For corporate groups, this would require a sibling structure where the trading entity and the deposit entity are sister companies operating entirely separately and with no connection other than ultimately sharing the same parent company. This fence would also aim at limiting the interconnectedness among various banking groups to dampen the impact of more systemic shocks. To this end, it would be stipulated that current large exposures limits be made more restrictive. Finally, the high fence under Separation 3 would limit the number of board members from the parent sitting on the boards of the trading entity and the deposit entity.

Graph 1 illustrates the three different models discussed above and highlights the corresponding restrictions.

*TE = trading entity

**DE = deposit entity

4.2.1. Separation 1 - subsidiarisation

An evaluation of subsidiarisation as a stand-alone fence gives rise to questions as to its effectiveness:

Would subsidiarisation facilitate resolvability: Subsidiarisation requires the trading entity and the deposit entity to maintain self-standing reserves of capital and of loss-absorbing debt, as well as to comply with the other prudential requirements on an individual or sub-consolidated basis which facilitate resolution. Subsidiarisation would therefore likely bring more clarity, transparency and structure to a banking group's internal organization which could facilitate the architecture of a recovery and resolution plan and also facilitate resolvability. This is because, in the first instance, the economic, operational and governance separation facilitates the preservation of the viable residual elements and the resolution and wind-down of non-viable elements. In the second instance, if the whole entity is to be resolved, asset valuation (and the application of pre-insolvency bail-in tools) is facilitated by the easy identification of assets and losses.

Would subsidiarisation reduce moral hazard (excessive risk taking): Because subsidiarisation requires the trading entity and the deposit entity to maintain self-standing reserves of capital and of loss-absorbing debt (which may not be diminished below safeguard levels set out in the CRR and the BRRD), as well as to comply with the other prudential requirements on an individual basis, subsidiarisation provides a certain degree of independence and to some extent also insulates the deposit entity from shocks and losses affecting the trading entity. In principle, it also means that risks and costs are shifted to who bears them. This should have the effect of curbing the incentive to take excessive risks.

However, if the trading entity is part of a wider corporate group the parent company remains in control of the trading entity and the deposit entity.[127] This parent/subsidiary relationship may signal that the implicit safety net is still present in which case the trading entity will continue to benefit from "preferential" access to and cost of funding. This effect would be exacerbated if the trading entity is the parent undertaking owning the deposit entity. Moreover, because excess capital can be shifted around within the wider corporate group there may be an additional incentive for the parent/trading entity (and for the deposit entity especially if its board members' remuneration schemes are based also on performance of the entire corporate group) to encourage the allocation of capital and human resources to trading and away from, for example, lending activity and continued excessive risk taking.

Would subsidiarisation facilitate monitoring, management and supervision: Each of the entities will be subject to separate capital requirements and a requirement to maintain a sufficient level of loss-absorbing debt, which could impact on shareholders’ incentives to monitor risk taking. Separation may also improve transparency and therefore improve the ability for shareholders to better monitor the activities.

The responsibility and independence of management could increase through separation and this may make it easier to ensure that objectives and needs are more clearly defined. This could overall benefit the management of these entities. Separation 1 would result in a clearer corporate and business structure which could facilitate supervision and will likely also facilitate the application of recovery and resolution measures.

Would subsidiarisation reduce conflicts of interest: Governance separation could contribute to reducing conflicts of interest between the trading business and the deposit-taking business. Because governance separation would to some extent enhance managers' independence this could further contribute to reducing conflicts of interest and culture shocks. However, because it is not excluded that the trading entity would own the deposit entity and, if so, that they would be engaged in the same or similar line of business,[128] their interests inevitably would collide and, thus, the interest of creditors or employees with stock options could be at stake.

As long as there is no ownership restrictions on the trading entity and as long as governance rules do not reach further than the basics outlined in the CRDIV, and especially as long as there is no duty on the parent to uphold the objectives of separation, it may be questioned whether the objectives of the separation will be sufficiently safeguarded and whether the risk taking management culture will not continue to influence the operations of the deposit entity.

Would subsidiarisation reduce resource and capital misallocation: While separation of certain activities and separate governance could contribute to limiting the incentives for financial institutions to allocate resources and capital away from lending activities to trading activities it is unlikely that this would have a very significant impact under Separation 1. This is because excess capital can be shifted around within the wider corporate group, which provides an incentive for the parent - and for the deposit entity if its board members' remuneration schemes depend also on performance of the entire corporate group - to allocate capital and human resources to trading and intra-financial activity rather than to lending activity. This could particularly be the case as there is no duty on the parent board to maintain the objectives of separation.

Would subsidiarisation lead to loss of efficiencies: Separation 1 would have an impact on a banking group's ability to benefit fully from diversification in the sense that the scope for using income smoothing techniques through diversification of activities and centralisation of liquidity and risk management would be reduced. Therefore the increased costs would only reflect the loss of diversification for meeting prudential needs on a stand-alone basis. However, the option to place many activities on either side of the fence, the fact that excess capital could under certain conditions still be transferred among the entities within the wider corporate group, the ability to pull funding needs, and the fact that customers could still use a single banking group for all of their services means that a certain level of diversification benefits would remain. Similarly, banking groups may continue to benefit from operating cost reductions from pooling certain resources such as IT and finance systems, and from shared marketing and advertising campaigns.

Would subsidiarisation improve competition: Because Separation 1 does not remove the implicit public subsidy any competitive distortions (large banks versus small banks, within a Member State or across Member States or wider) will remain.

Based on the above, this degree of separation is unlikely to be sufficiently strong to effectively achieve the operational goals of structural reform.

4.2.2. Separation 2 – "subsidiarisation +"

Separation 2 puts more emphasis on making the trading entity and the deposit entity more self-standing and, where the trading entity forms part of a wider corporate group, regulating the intra-group relationship.

To start with, it would therefore be required that each sub-consolidated group of homogeneous entities (i.e., the "deposit group," and the "trading group") would have to issue its own debt independently from all other entities within the wider corporate group (including the parent).[129] This means that each sub-consolidated group would, to a large extent, have to fund themselves independently from the group.

To the extent that the trading entity forms part of a wider corporate group, the CRDIV prudential rules regarding disclosure need to apply fully to both the trading and the deposit entities on a sub-consolidated basis.

To reduce the interconnectedness of entities that form part of a wider corporate group and to ensure a transparent and appropriate relationship between them, it would first be required that a deposit entity cannot own shares or voting powers in a trading entity. Second, all intra-group contracts and other transactions would have to be made on an arm's length basis and on normal commercial terms. The latter limits intra-group relationships to the same general level as general third party relationships. It would also be made clear that the trading entity could not benefit from resolution support from the deposit entity.

To further reduce conflicts of interests at governance level it would be required that there be: (i) no cross-membership between the boards of the trading entity and the deposit entity (or only a minority of members can sit on both boards); and (ii) a statutory duty for the parent and the deposit entity boards to uphold the objectives of the separation.

Finally, there would be an obligation imposed on the parent to secure the deposit entity's operations (e.g., access to critical infrastructure such as payments services, staff, data, information and other services) irrespective of the wider group's financial health.

The fence under Separation 2 would be of virtually the same strength as the current UK fence.

4.2.2.1. Social benefits and social costs

Would subsidiarisation+ facilitate resolvability: Separation 2 has the effect of further insulating the trading entity from the deposit entity in terms of funding; more insulation enhances resolvability without taxpayer support. This would also curtail implicit government guarantees, reducing the risk to the public finances and making it less likely that banks will run excessive risks in the first place. In addition separate funding also reduces complexity of funding sources.

Would subsidiarisation+ reduce moral hazard (excessive risk taking): The fence under Separation 2 focuses more on independence and regulating the relationship among entities within the wider corporate group than the previous fence. The rule on separate debt issuance would make it more difficult for the trading entity (and the deposit entity) to get access to "intra-group funding" which means that it has to procure funding from the "external" market place. This would have the effect of making funding of risky activities more expensive as the funding of operations would better reflect their underlying riskiness. Conducting all transactions on an arm's length commercial basis would also most likely lead to an increase in sales/service price. Cross-subsidization of trading activities with "cheap" deposits would be significantly reduced. Compared to Separation I, these higher costs should have the effect of restraining even more the incentives of the trading entity to take excessive risks. However, to the extent that the trading entity is part of a wider corporate group and controls the deposit entity that effect would be countered. This could to some extent be mitigated when the trading entity cannot get any support in resolution from the deposit entity and when the management body of the parent (i.e., possibly the trading entity itself) as well as the management body of the deposit entity would be under a duty to uphold the objectives of the separation. These requirements could provide additional credibility to the fence and signal to the market place that the implicit safety no longer exists. This should change the incentives for institutions to become more moderate in risk-taking and therefore reduce moral hazard.

Would subsidiarisation+ facilitate monitoring, management and supervision: The more the trading entity's funding has to come from external capital markets the more shareholders should be incentivized to monitor risk taking. The fact that the parent (possibly the trading entity itself) management body would be under a duty to uphold the objectives of the separation could significantly impact on the way it operates with regard to excessive risk taking. Again, more insulation of the trading entity from the deposit entity would mean that the objectives and needs of each separate entity would be even clearer and therefore easier for management to implement and supervise. Reducing the intra-group interconnectedness will make each entity's accounts more clear and likely more transparent, which will facilitate supervision.

Would subsidiarisation+ reduce conflicts of interest: Prohibiting cross-management body memberships could be very effective in reducing conflicts of interest and culture between the trading business and the deposit taking business. However, in case the trading entity is the parent of the deposit entity this requirement would likely be far too strict as it would not allow the parent any influence over wider corporate group strategies and would sever the line of accountability through the parent to investors and likely make the banking group significantly less attractive as an investment object.

Importantly, the duty on the parent and the deposit entity boards to uphold the objectives of separation could have a significant impact on conflicts of interest as well as on the extent of risk taking within each of the entities. This would particularly be the case if the trading entity owns the deposit entity.

Would subsidiarisation+ reduce resource and capital misallocation: Notwithstanding the ability to shift around excess capital within the wider corporate group, the fence under Separation 2 restricts intra-group relationships and puts them on the same base and terms as strictly third party commercial relationships. Coupled with the governance duties to uphold the objectives of the separation, Separation 2 could better contribute to limiting the incentives for financial institutions to allocate resources and capital away from lending activities to trading activities.

Would subsidiarisation+ generate loss of efficiencies: The stricter fence would make it more expensive for the group to benefit from diversification and other economies of scope. This would be reflected in higher funding costs of relatively risky activities but possibly in even lower funding costs for safer activities. As under Separation 1 there is still the option to place many activities on either side of the fence, excess capital could under certain conditions still be transferred among the entities within the wider corporate group and customers could still use a single group for all of their services, which means that a certain level of scope economies would remain. Similarly, groups may continue to benefit from operating cost reductions from pooling certain resources such as, for example, IT and finance systems and from shared marketing and advertising campaigns (operational efficiencies).

Would subsidiarisation+ improve competition: Only to the extent that Separation 2 could constitute a credible threat that the implicit public subsidy would be reduced would competitive distortions (large banks versus small banks, within a Member State or across Member States or wider) be lowered correspondingly.

Based on the above, it is concluded that Separation 2 (subsidiarisation +) constitutes a more credible fence than Separation 1, and, on a balance, it appears that the social benefits outweighs the social costs; however, this fence remains permeable.

4.2.3. Separation 3 - "subsidiarisation ++"

Separation 3 aims at provide a fence of maximum strength to protect the objectives of the separation while still leaving room for banking groups to provide a universal set of services and thus allowing for some efficiency and diversification benefits at the group level. It builds on the model for Separation 2 as described above in section 4.2.2.

To this end it would be stipulated that the trading entity could not own a deposit entity (in addition to the prohibition on deposit entities owning trading entities).[130] For corporate groups, this would require a sibling structure where the trading entity and the deposit entity are sister companies operating entirely separately and with no connection other than ultimately sharing the same parent company.

Separation 3 would also aim at limiting the interconnectedness among various banking groups to dampen the impact of more systemic shocks. To this end, it would be stipulated that current large exposures be subject to stricter limits than currently applicable.

Finally, the high fence under Separation 3 would limit the number of board members from the parent sitting on the boards of the trading entity and the deposit entity. This would allow the parent to contribute to strategy and objective setting while leaving each subsidiary freedom to set its own agenda.

4.2.3.1. Social benefits and social costs of Separation 3 (subsidiarisation ++)

Would subsidiarisation++ facilitate resolvability: It would be easier in a sibling structure to deploy entirely different resolution tools for the trading entity and the deposit entity. The risk of contagion would also be reduced as the parent/holding company would act as a firewall between the two sister companies and stricter large exposures would apply. I.e., if the trading entity were to fail, the regulator could shut it down without affecting its deposit entity sister firm in a critical way.

Would subsidiarisation++ reduce moral hazard (excessive risk taking): Separation 3 would require some type of sibling structure for corporate groups. This type of structure would be consistent with the principle of insulating the trading entity from the deposit entity, which is what structural reform seeks to achieve through subsidiarisation. The more independent the trading entity is and the more intra-group relationships are assimilated to relationships with third parties,  the more credible it may be to the market place that the trading entity will not benefit from government support in case of financial trouble. Coupled with the fact that the trading entity would no longer have access to cheaper internal funding, incentives for excessive risk taking should be significantly reduced. The limit on the number of parent members sitting on the boards of the trading entity and the deposit entity (coupled with the duty of the parent to uphold the objectives of the separation) would also further discipline any incentives for excessive risk taking as well as ensuring better alignment of incentives.

Would subsidiarisation++ facilitate monitoring, management, supervision and regulation: Separation 3 would clearly separate banking groups' structures and operations and make them clearer and transparent which makes it easier for regulators, investors, creditors and managers to see potential weaknesses and to discipline excessive risk taking.

Would subsidiarisation++ reduce conflicts of interest: Separate governance through a sibling structure could contribute to reducing conflicts of interest between the trading business and the deposit taking business and bring about thicker walls between the sister companies. Limiting the number of parent members sitting on the boards of the trading entity and the deposit entity could also further reduce conflicts of interest between ultimate shareholders and the trading and deposit entities.

Would subsidiarisation++ reduce resource and capital misallocation: A sibling structure would further cement the separation of the trading entity and the deposit entity and lend more credibility to the integrity of the fence between the two entities. Since cheap access to internal funding would no longer be available, the trading entity would have to rely more on the external market, which would make its funding more risk sensitive and hence more expensive (at least for the trading activities). Because of the additional insulation of the trading entity, Separation 3, compared to the other fences, has a more credible positive impact on reducing any implicit subsidy. Reduced incentives to invest in projects which are not worthwhile on average, but risky enough to have at least some chance of making money could then also lead to a better allocation of capital in the economy.

Would subsidiarisation++ generate loss of efficiencies: Separation 3 would further make it more expensive for the banking group to benefit from diversification and other economies of scope. However, there is still the option to place many activities on either side of the fence. Also, excess capital could under certain conditions still be transferred among the entities within the wider corporate group and customers could still use a single group for all of their services. This means that a certain level of economies of scope would remain. Similarly, groups may continue to benefit from operating cost reductions from pooling certain resources, such as IT and finance systems, and from shared marketing and advertising campaigns.

Would subsidiarisation++ improve competition: Next to ownership separation, Separation III constitutes the most credible threat that the implicit public subsidy would be eliminated.  Removing the implicit subsidy would contribute significantly to removing competitive distortions between large banks versus small banks, within a Member State or across Member States or widerBased on the above it is concluded that Separation 3 (subsidiarisation ++) provides the most non-permeable separation short of ownership separation but also comes at a higher cost for financial institutions compared to Separation I and II.  For this reason it has not been retained for assessment in the main body of the Impact Assessment.

Annex A8 : Trading activities and Bank Structural Separation: Possible Definitions and Calibration of Screening exemption thresholds

A. Pagano, J. Cariboni, M. Marchesi, N. Ndacyayisenga, M. Petracco Giudici, H. Joensson  

Forename(s) Surname(s)

2013

European Commission

Joint Research Centre

Institute for the Protection and Security of the Citizen

Contact information

Jessica Cariboni

Address: Joint Research Centre, Via Enrico Fermi 2749, TP 361, 21027 Ispra (VA), Italy

E-mail: jessica.cariboni@jrc.ec.europa.eu

Tel.: +39 0332 78 9372

Fax: +39 0332 78 5733

http://ipsc.jrc.ec.europa.eu/

http://www.jrc.ec.europa.eu/

Legal Notice

Neither the European Commission nor any person acting on behalf of the Commission

is responsible for the use which might be made of this publication.

Europe Direct is a service to help you find answers to your questions about the European Union

Freephone number (*): 00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.

A great deal of additional information on the European Union is available on the Internet.

It can be accessed through the Europa server http://europa.eu/.

JRC84704

© European Union, 2013

Executive summary

Commission Services have undertaken work with the aim of reviewing the examination thresholds suggested by the High Level Experts Group (HLEG) to select bank to be subject to structural separation. This work is focusing in particular on how to define trading activity, estimating the institutional implications of different examination thresholds, and benchmarking the results against other readily available metrics.

Commission Services have assessed a number of options for defining trading activity in order to determine the scope of the institutions subject to a separation requirement. Due to the absence of publicly available data for banks’ specific business lines, this analysis has been done on the basis of publicly available accounting (balance sheet) data from commercial providers. The analyzed options are:

•      HLEG definition (assets held for trading and available for sale);

•      a more narrow definition that excludes available for sale assets as mostly composed of securities held for liquidity purposes ;

•      a definition focused on the gross volume of trading activity, which is likely to focus on proprietary traders and market-makers; or

•      a definition focused on net volumes, which is likely to only capture those institutions that have a higher share of unbalanced risk trading (proprietary traders).

For each of the last three options, absolute and relative thresholds have been assessed. Depending on the option chosen, about 30-35 banks are selected. Even though the selected banks represent less than 20% of the sample, their assets account between 50% and 75% of the assets in the sample, and by and large between 40% and 60% of EU banking assets. It is important to remark that 20 banks are selected under all definitions and they represent 50% of the sample in terms of total assets. By considering only banks with total assets over 30 bn EUR, 16 banks are selected by all definitions to be proposed for structural reform.

Table of Contents

1. Introduction and scope

Following the suggestions of the report by the High Level Experts Group (HLEG) on reforming the structure of the EU banking sector, the so called Liikanen report,[131] Commission Services have been investigating the impact that a functional structural reform could imply.

The Liikanen report suggests that “proprietary trading and all assets or derivative positions incurred in the process of market-making must be assigned to a separate legal entity within the banking group”. Separation would only be mandatory if trading activities amount to a significant share of balance sheet, or if their volume can be considered significant from the viewpoint of financial stability. In particular the HLEG suggests that separation should be mandatory if the sum of “bank’s assets held for trading and available for sale“ is (i) either a significant volume (above 100 bn EUR), or (ii) a significant share of the bank’s balance sheet (between 15-25%). However, the Supervisors would determine the activities that would actually need to be separated, so that above mentioned thresholds would serve for “examination” purposes by Supervisors (examination thresholds).

During the consultation[132], HLEG definition of trading activities has been criticized, especially as Available-For-Sale assets can be composed to a large extent by high quality liquid assets: government or corporate bonds held by banks for liquidity rather than for trading purposes. It was also pointed that current reforms, including European Market Infrastructure Regulation[133], Capital Requirement Directive IV[134], Tax on Financial Transactions[135] and the bank recovery and resolution directive (BRR), should also be considered when analyzing the impact of functional structural separation.

Within this context, this report analyses how to possibly modify the definition of trading activities relying on balance sheet data and how to calibrate examination thresholds for functional structural separation. The alternative definitions proposed by Commission Services are based on specific bank activities such as proprietary trading and market making.

The remaining of the report is structured as follows. Section 2 describes the data public available in SNL for analysing trading activities and introduces different definitions of trading activities which will be analysed in the remaining of the report, starting from the HLEG definition. Section 3 presents the results of the cluster analysis for each definition and set potential thresholds. Section 4 compares results of the various definitions and Section 5 concludes.

Some appendices will provide additional analyses and technical details. Appendix A provides details on the sample used for the analyses. Appendix B presents some additional graphs on the banks selected under the HLEG definition but not under alternative definitions. Appendix C investigates on the stability of thresholds across the years. Appendix D discusses definitions taking into account EMIR/FTT initiatives. Appendix E present an analysis based on balance sheet structural indicators. Appendix F gives few technical details on the clustering techniques used for setting the thresholds.

2. Data and definitions of trading activities

2.1 Description of the sample and of the data used

The banks sample used is extracted from the SNL database[136].[137] It covers 245[138] EU banks and spans the years 2006-2011, for which consolidated balance sheet data have been considered. Only banks with total assets higher than 1% of their home-country GDP are included in this calibration exercise.[139] This set of banks (see Table 1) represents around 75% of the EU banking sector in terms of total assets with respect to 2011[140].

Due to the presence of missing items the number of banks may differ according to definitions and to the reference year. In general there are around 190 banks for which all needed variables are available. In all cases, most of the banks for which data are not available are small[141]. In terms of total assets not including the banks with missing data reduces the coverage of EU27 banking sector to 74%. Hence all the statistics dealing with shares over total assets will refer to the entire sample set of 245 banks.

Table 1: Breakdown of the sample in bucket of sizes based on total assets averaged between 2006-2011. 500 bn EUR roughly corresponds to the 75th percentiles of the total assets of the banks in the sample of banks considered by the European Banking Authority in its capital exercise[142]. 30 bn EUR is the size above which banks will be supervised directly by the SSM.

Banks’ Buckets || Banks’ size (Total Assets) || Number of banks in the sample || Share of total assets in the sample

Small banks || Up to 30 bn EUR || 141 || 4%

Medium banks || between 30 and 500 bn EUR || 85 || 32%

Large banks || larger than 500 bn EUR || 19 || 64%

Total || 245 || 100%

The balance sheet items reclassified in SNL used to estimate trading activities are reported in Table 2.

Table 2: Classes of assets and liabilities used for estimate the size of the trading activity available in SNL.

Item || Label || Data definition

Total Assets || TA || All assets held by the banks at the indicated date, as carried on the balance sheet and defined under accounting principles in use.

Trading Securities Assets      || TSA || Assets part of a portfolio managed as a whole and for which there is evidence of a recent actual pattern of short-term profit-taking, excluding derivative assets.

Derivatives Assets || DA || Derivatives with positive replacement values not identified as hedging or embedded derivatives.

Available For Sale Securities     || AFS || Total securities designated as available for sale

Derivatives Liabilities || DL || Derivatives with negative replacement values not identified as hedging instruments.

Trading Liabilities || TSL || Liabilities taken with the intent on repurchasing in the near term, part of a portfolio managed as a whole, and for which where there is evidence of a recent actual pattern of short-term profit-taking, excluding derivative liabilities.

Figure 1 plots the average breakdown of trading activities under the HLEG definition (DA + TSA + AFS) for each bank size bucket, using average 2006-2011 data. It can be observed that:

- the share of total trading activities on total assets is larger for large banks;

- larger banks have more derivatives and trading securities;

- smaller banks focus relatively more on AFS.

Figure 1: Average share of trading activities in the asset side of the balance sheet on total assets by bucket of size. Data are the average 2006-2011. Source: SNL and JRC calculations.

Figure 2 shows how the share of trading activities under the HLEG definition and the relative importance of the three classes change over time for the large banks, highlighting how both the total share and the relative share of derivatives or securities held for trading are volatile.

This implies that, in order to have thresholds that remain stable over time, it is necessary to consider the average behavior over a certain time horizon. In the present work a 6-year average is considered.[143]

Figure 2: Evolution of trading activities over time for the large banks in the sample. Source: SNL and JRC calculations.

Among the proposed alternative definitions, trading liabilities are also considered. Figure 3 shows the shares of the two classes of trading activities in the liability side over total assets. Large banks tend to have large share of trading liabilities.

Comparing Figure 3 with Figure 1, trading securities assets tend to be higher than trading securities liabilities while derivative assets and liabilities tend to be more balanced for all bank size bucket.

Figure 3: Average share of trading activities on liability side on total assets by bucket. Data are the average 2006-2011. Source: SNL and JRC calculations[144].

2.2 Use of alternatives definitions of trading activities: discussion and proposition

In this section, different definitions of trading activities using balance sheet data are introduced and discussed and eventually used to calibrate the de minimis exemption rule.

Starting from the HLEG definitions and alternative definitions are considered with the aim to take into account the comments from stakeholders collected through the HLEG report consultation[145]. In particular: some consultation replies have pointed out that Available-For-Sale assets are normally composed of high quality liquid assets, such as government bonds or corporate bonds of high-quality, held by banks for liquidity purposes rather than for trading purposes. [146],[147] Some criticisms have also been raised on the fact that the HLEG definition does not seem to provide a good proxy of the risks undertaken through trading activities.

These comments have led the Commission Services to consider the removal of AFS from the definition of trading activities and to consider some additional definitions which may better focus on the two main risks due to trading: market risk and counterparty risk. While market risk can be considered to be more related to one directional bets on market movements, counterparty risk depends more on the volume of trading undertaken by the bank[148].

On the basis of these considerations, the Commission Services have decided to analyse the definitions presented in Table 3, also considering limitations due to the need to rely on publicly available accounting data.

Table 3: Definitions of trading activities based on the labels of balance sheet items introduced in Table 1.

|| Definition || Description

1 || TSA + DA + AFS || HLEG definition

2 || TSA + DA || Exclusion of AFS under the assumption that they are mostly held for liquidity purposes

3 || (TSA + TSL + DA + DL)/2 || Exclude AFS + Gross volumes of securities and derivatives held for trading (to focus on market and counterparty risk)

4 || |TSA – TSL| +|DA - DL| || Exclude AFS + Net volumes of securities and derivatives held for trading (to focus on market risk)

A second issue linked with the choice of the definition is the scope of the separation. The HLEG recommends that assets or derivatives positions incurred in the process of market-making must be assigned to a separate legal entity. The Commission Services also consider the possibility that only proprietary trading is separated.[149]

It is still worth to notice that while precise legal and economic definitions of proprietary trading and market making is subject to a certain level of uncertainty[150], one can - in very general terms - consider that proprietary trading is more exposed to market risk while market making is more related to counterparty risk.

To capture both market makers and proprietary traders, Commission Services propose to use definition 3, which focuses on gross volumes of trading activities by summing assets and liabilities. In fact, this measure could be considered a proxy for banks’ exposure to counterparty and market risk.

To capture proprietary traders only, Commission Services propose to use definition 4, which focuses on unbalanced positions between assets and liabilities (net volumes). This idea is based on the assumption that market makers normally try to keep a balanced net position while still showing important gross positions; while proprietary trading is based on directional bets that the market will move in a certain direction, so that it generates asymmetric positions on the two sides of the balance sheet.

Commission Services have also investigated additional definitions (see Appendix D) aimed at taking into account some other ongoing Commission legislative initiatives that might impact on the volumes of trading activities, notably EMIR (European Market Infrastructure Regulation) and FTT (Financial Transaction Tax).

3. Thresholds calibration

The HLEG suggests examination thresholds for functional structural separation based on the amount of trading activities and share of trading activities on total assets, equal to 100 bn EUR or 15-25% respectively. The results presented in this report are based on the assumption that the ‘or’ logical operator best suits for the purposes of a functional structural separation.

The scope of this section is to calibrate thresholds for each definition of trading activities introduced in Table 3, using banks’ balance sheet data spanning the period 2006-2011.

We develop a clustering exercise[151] in the two dimensional space (amount, share) to build banks clusters and set the thresholds according to them. The analysis is based on averages of balance sheet data over the considered period.[152]

Given the results of the clustering exercise, thresholds are chosen in such a way that

- respect the shape of clusters and avoid splitting them;

- are multiple of 10 bn EUR for the amount and of 2.5% for the share of trading activities;[153]

- show consistency across years;

- show as much consistency as possible with the thresholds set by the HLEG.

3.1. HLEG definition

Figure 4 shows the results of the clustering exercise for the HLEG definition. The top plot presents all banks in the sample, while the bottom left one zooms on total trading activities below 250 bn EUR, excluding the well separated group of very large players (SocGen, HSBC, CrAgr, Barclays, RBS, BNPP, Deutsche Bank). The bottom right plot is used to identify those banks which are close to the thresholds.

On the amount dimension (x-axis), the presence of a stretched cluster of banks (green  diamonds) between 30 bn EUR and 150 bn EUR does not allow to fix a threshold on the amount dimension in the neighbourhood of 100 bn EUR that does not intersect this cluster. However, as shown in the bottom left plot, there is an important gap between 50 bn EUR and 100 bn EUR, for shares lower than 15%. On this basis, a threshold is suggested equal to 80 bn EUR. This smaller amount compared to the 100 bn EUR is suggested by the following:

the presence of several banks belonging to the green diamonds cluster in the area close to 100 bn EUR; the presence of many fewer banks in the area close to 80 bn EUR, which could help the threshold to be more stable over time.

On the share dimension (y-axis), it is more difficult to identify a unique threshold due to the very dense presence of banks of smaller size. This difficulty is in line with the decision of the HLEG to provide a range (15%-25%) instead of a single threshold on the share dimension. The clustering analysis suggests that a threshold, within the suggested range, could be fixed at 20%, so as to separate light blue stars from the red circles.

Fixing the examination thresholds at 80 bn EUR for the amount or 20% for the share allows preserving the stretched cluster of green diamonds. It can also be noticed that a rather populated cluster of small banks with a high share of trading activities (light blue stars) would be captured with this choice. Not considering AFS will reduce the number of very small banks selected (see next sections).

The bottom right plot of Figure 4 highlights those banks which are close to the thresholds (see also Appendix C for a detailed discussion). The use of a measure of distance from the thresholds would allow softening the definition of hard thresholds resulting from cluster analysis and could be used as additional information for supervisors to discretionally enlarge the list of banks subject to separation.[154]

Figure 4: Clusters under Definition 1 (Liikanen) .

3.2. Alternative definitions

In this section we use the cluster analysis to set potential thresholds for the alternative definitions. To the list of criteria we already mention to properly set the thresholds based on clusters we add the need of having values differing as little as possible from thresholds “equivalent” to the HLEG ones.

The “equivalent” thresholds are obtained shifting and/or rescaling the HLEG thresholds (80 bn EUR or 20%) using the sample weighted average of the items of balance sheets cut, added or rescaled in the various definitions.[155]

Table 4 shows the equivalent thresholds. Table 5 presents a choice of potential thresholds based on cluster analysis. Graphs are presented in Figure 5, 6 and 7. Comparing the equivalent thresholds in with those obtained via cluster analysis we observe similar results.

Table 4: HLEG “equivalent” thresholds for the alternative definitions.

HLEG “EQUIVALENT” THRESHOLDS

Definition || THRESHOLDS || SELECTED BANKS || SELECTED BANKS BY SIZE

TradAct bn EUR || Shar TradAct || Number || % of the sample || % of the sample in total assets || Large || Medium || Small

1 || TSA + DA + AFS || 80 || 20% || 52 || 21% || 75% || 19 || 16 || 17

2 || TSA +DA || 66 || 12% || 40 || 16% || 65% || 14 || 13 || 13

3 || (TSA+TSL + DA+DL)/2 || 62 || 10% || 36 || 15% || 65% || 13 || 16 || 7

4 || |TSA-TSL| + |DA-DL| || 37 || 8% || 28 || 12% || 45% || 9 || 6 || 13

Table 5: Proposed thresholds based on cluster analysis.

THRESHOLDS FROM CLUSTERS

Definition || THRESHOLDS || SELECTED BANKS || SELECTED BANKS BY SIZE

TradAct bn EUR || Shar TradAct || Number || % of the sample || % of the sample in total assets || Large || Medium || Small

1 || TSA + DA + AFS || 80 || 20% || 52 || 21% || 75% || 19 || 16 || 17

2 || TSA +DA || 70 || 15% || 32 || 13% || 60% || 13 || 10 || 9

3 || (TSA+TSL + DA+DL)/2 || 70 || 10% || 36 || 15% || 65% || 13 || 16 || 7

4 || |TSA-TSL| + |DA-DL| || 30 || 8% || 33 || 14% || 51% || 11 || 7 || 15

Figure 5: Cluster for Definition 2 (exclude AFS).

Figure 6: Clusters for Definition 3 (volumes without AFS).

Figure 7: Clusters for Definition 4 (net volumes on derivatives and securities without AFS)

From these scatter plots and the associated cluster analyses, one can notice that the information about proximity to the thresholds (bottom right plots) becomes more and more relevant in order to better preserve similarity among selected banks which may be partly lost when using hard thresholds.

To have an insight in the characteristics of the set of selected and non-selected banks for each definition, Figure 8 shows the two dimensional box-plot of the share of trading activities (x-axis) versus share of covered deposits on total assets (y-axis).[156] This figure shows that the selected banks have much larger share of trading activities (by definition), but smaller share of covered deposits[157]. This together with the relative dimensions of the rectangles including the central values (measures of dispersions) suggests that banks with large share of trading activities tend to finance it with sources of funding other than covered deposits.

Figure 8: Boxplot in two dimensions for the HLEG definition 1: the stars identify the median values, the light cyan and orange rectangles include 80% central values while the dark cyan and orange rectangles include the 50% central values

Figure 9: Boxplot in two dimensions for definition 2: the stars identify the median values, the light cyan and orange rectangles include 80% central values while the dark cyan and orange rectangles include the 50% central values

Figure 10: Boxplot in two dimensions for definition 3: the stars identify the median values, the light cyan and orange rectangles include 80% central values while the dark cyan and orange rectangles include the 50% central values.

Figure 11: Boxplot in two dimensions for definition 4: the stars identify the median values, the light cyan and orange rectangles include 80% central values while the dark cyan and orange rectangles include the 50% central values.

4. Comparison across definitions

This section presents a comparison among banks selected according to each definition (and relative thresholds). Twenty banks are selected by all definitions which correspond to less than 10% of the sample in terms of individuals. Considering the size, they actually account for about 50% of the sample total assets. Other 44 banks are selected by at least one definition, and, among them 17 are selected only by HLEG. In term of total assets all 44 account for 30% of the sample.

Further, it is worth noticing that:

· among the 20 banks selected by all definitions (see Table 6) 10 are large banks, 6 medium-sized and only 4 are small-sized. Ten of these banks are in the list of the G-SIBs.

· The 17 banks selected only by the HLEG (see Table 7) include 5 large banks, 4 medium sized banks and 8 small banks. These small banks are selected by HLEG due the size of their AFS assets (see Appendix B). None is a G-SIB, although only recently Lloyds have been excluded from this list.

· the 27 banks are selected according to more than one definition but not all (see Table 8) include 4 large banks, 10 medium sized banks and 13 small banks. These small banks are selected by the definition 4 and are mainly located in Denmark (4 banks) and Italy (5 banks).

· the 14 EU G-SIBs provided by the Financial Stability Board[158] are all selected by at least one definition. While the HLEG definitions select all of them, definitions 2 3 and 4 select 12, 13 and 11 G-SIBs respectively. BBVA and Standard Chartered are selected by only one alternative definition (4 and 3 respectively).

                        

In Tables 6 to 9, a double-line frame circles banks with total assets above 30 bn EUR.

Table 6: List of the banks selected by all definitions (ordered by size bucket). Size: L=Large, M=Medium and S=Small

|| Bank || Size || MS || 1 || 2 || 3 || 4 || G-SIBs

TSA+DA +AFS || TSA+DA || (TSA+TSL+DA + DL)/2 || |TSA - TSL| + |DA - DL|

1 || BNP Paribas || L || FR || Y || Y || Y || Y || Y

2 || Crédit Agricole Group || L || FR || Y || Y || Y || Y || Y

3 || SocGen || L || FR || Y || Y || Y || Y || Y

4 || Deutsche Bank || L || DE || Y || Y || Y || Y || Y

5 || Unicredit || L || IT || Y || Y || Y || Y || Y

6 || Santander || L || ES || Y || Y || Y || Y || Y

7 || Barclays || L || UK || Y || Y || Y || Y || Y

8 || HSBC || L || UK || Y || Y || Y || Y || Y

9 || RBS || L || UK || Y || Y || Y || Y || Y

10 || Nordea || L || SE || Y || Y || Y || Y || Y

11 || KBC || M || BE || Y || Y || Y || Y ||

12 || Danske Bank || M || DK || Y || Y || Y || Y ||

13 || DekaBank Deutsche GZ || M || DE || Y || Y || Y || Y ||

14 || Portigon || M || DE || Y || Y || Y || Y ||

15 || Mediobanca || M || IT || Y || Y || Y || Y ||

16 || SEB || M || SE || Y || Y || Y || Y ||

17 || Arbejdernes Landsbank || S || DK || Y || Y || Y || Y ||

18 || DiBa Bank || S || DK || Y || Y || Y || Y ||

19 || Banca Generali || S || IT || Y || Y || Y || Y ||

20 || FIMBank || S || MT || Y || Y || Y || Y ||

Table 7: List of the banks selected only in HLEG definitions. Size: L=Large, M = Medium and S= Small.

|| Bank || Size || MS || 1 || 2 || 3 || 4 || G-SIBs

TSA+DA+ AFS || TSA+DA || (TSA+TSL+DA + DL)/2 || |TSA - TSL| + |DA - DL|

1 || Lloyds Banking Group || L || UK || Y || || || ||

2 || Rabobank Group || L || NL || Y || || || ||

3 || Crédit Mutuel Group || L || FR || Y || || || ||

4 || Dexia || L || BE || Y || || || ||

5 || Intesa || L || IT || Y || || || ||

6 || Landesbank Berlin Holding AG || M || DE || Y || || || ||

7 || NORD/LB || M || DE || Y || || || ||

8 || Banque Intl. Luxembourg || M || LU || Y || || || ||

9 || Banque et Caisse d'Epargne || M || LU || Y || || || ||

10 || Credito Emiliano SpA || S || IT || Y || || || ||

11 || Aktia Plc || S || FI || Y || || || ||

12 || AXA Bank Europe || S || BE || Y || || || ||

13 || Banca Comerciala Carpatica SA || S || RO || Y || || || ||

14 || Raiff. Landsbk Steiermark || S || AT || Y || || || ||

15 || Getin Noble Bank || S || PL || Y || || || ||

16 || ICICI Bank UK Plc || S || UK || Y || || || ||

17 || SaarLB || S || DE || Y || || || ||

Table 8: Banks selected according to some but not all definitions (Y if selected, empty otherwise). Size: L=Large, M=Medium and S=Small.

|| Bank || Size || MS || 1 || 2 || 3 || 4 || G-SIBs

TSA+DA+ AFS || TSA+DA || (TSA+TSL+DA + DL)/2 || |TSA - TSL| + |DA - DL|

1 || Groupe BPCE || L || FR || Y || Y || Y || || Y

2 || ING || L || NL || Y || Y || Y || || Y

3 || Commerzbank || L || DE || Y || Y || Y || ||

  4 || BBVA || L || ES || Y || || || Y || Y

5 || Belfius Banque || M || BE || Y || || Y || ||

6 || DZ Bank AG || M || DE || Y || Y || Y || ||

7 || Landsbk Baden-Württ. || M || DE || Y || Y || Y || ||

8 || Bayerische Landesbk || M || DE || Y || || Y || ||

9 || Monte Paschi Siena || M || IT || || || Y || ||

10 || DNB ASA || M || NO || || || Y || Y ||

11 || Helaba || M || DE || Y || Y || Y || ||

12 || Swedbank || M || SE || || Y || Y || ||

13 || Handelsbanken || M || SE || || || Y || ||

14 || Standard Chartered || M || UK || Y || || Y || || Y

15 || Spar Nord Bank A/S || S || DK || || Y || Y || Y ||

16 || Lån & Spar Bank || S || DK || || Y || || Y ||

17 || Vordingborg Bank A/S || S || DK || || Y || || Y ||

18 || Nordfyns Bank A/S || S || DK || || || || Y ||

19 || Naspa Dublin || S || IE || Y || Y || || Y ||

20 || Banca Finnat Euramerica || S || IT || Y || || || Y ||

21 || Banca Pop di Sondrio || S || IT || || || || Y ||

22 || Banca Pop di Spoleto || S || IT || || || || Y ||

23 || Banca Sella Holding || S || IT || || || || Y ||

24 || Istituto Centrale delle BP || S || IT || || || || Y ||

25 || NIBC Bank NV || S || NL || Y || || Y || ||

26 || Handlowy w Warszawie || S || PL || Y || || Y || ||

27 || Banco Coop Español || S || ES || Y || Y || || Y ||

Table 9: G-SIBs.

|| Bank || Size || MS || 1 || 2 || 3 || 4 || G-SIBs

TSA+DA+AFS || TSA+DA || (TSA+TSL+DA + DL)/2 || |TSA - TSL| + |DA - DL|

1 || BNP Paribas || L || FR || Y || Y || Y || Y || Y

2 || Crédit Agricole Group || L || FR || Y || Y || Y || Y || Y

3 || SocGen || L || FR || Y || Y || Y || Y || Y

4 || Deutsche Bank || L || DE || Y || Y || Y || Y || Y

5 || Unicredit || L || IT || Y || Y || Y || Y || Y

6 || Santander || L || ES || Y || Y || Y || Y || Y

7 || Barclays || L || UK || Y || Y || Y || Y || Y

8 || HSBC || L || UK || Y || Y || Y || Y || Y

9 || RBS || L || UK || Y || Y || Y || Y || Y

10 || Nordea || L || SE || Y || Y || Y || Y || Y

11 || ING || L || NL || Y || Y || Y || || Y

12 || Groupe BPCE || L || FR || Y || Y || Y || || Y

13 || BBVA || L || ES || Y || || || Y || Y

14 || Standard Chartered || M || UK || Y || || Y || || Y

5. Conclusions

This report stems from the decision of the Commission Services to investigate alternative definitions of trading activities in order to address some of the questions raised during the stakeholder consultation following the HLEG report. Various definitions of trading activities have been analysed and corresponding examination thresholds for functional structural separation have been identified on the basis of cluster analysis.

In total three alternative definitions of trading activities have been proposed in addition to the HLEG one. In all of them the AFS assets are not considered, since this class is mainly composed of assets held for liquidity purposes, rather than for trading purposes. In particular, one definition also focuses on total volumes of trading activities, averaging the asset and liability sides, in order to capture both market and counterparty risks. Finally, another definition focuses also on net volumes of trading activities, netting assets and liability sides, to capture directional bets on market movements (market risk).

All definitions are based on publicly available balance sheet data and they have been computed considering averages over a 6-year period (2006-2011). Examination thresholds have been calibrated via a cluster analysis, aiming at identifying banks (those above the thresholds) that would be subject to functional structural separation, after a supervisory scrutiny that would determine the part of their assets to be separated.

On the basis of the results presented in the report and in the enclosed appendices the following conclusions can be drawn:

· independent of the definition of trading activities the percentage of banks selected in our sample (composed of 245 banks[159]) is rather small in terms of individuals, varying in the range 13%-21% in terms of number and between 50% and 75% in terms of total assets, i.e. by and large between 40% and 60% of EU banking assets. Most G-SIBs are identified by all definitions

· with respect to HLEG definition, the cluster analysis suggests thresholds of 80 bn EUR for the amount of trading activities, 20% for their share. These are to a large extent in line with the thresholds proposed by the HLEG (100 bn EUR, 15%-25%).[160] The HLEG definition selects more banks than the others, partly due to the presence of the AFS assets (not considered in the other definitions).

· 20 banks are selected under all definitions (around 10% of the sample). These banks represent a large share of the total number of selected banks under all definition (varying roughly from 40% to 65%). They also account for almost 50% of the sample total assets.

· the definition based on gross volumes is used to have a proxy able to identify market making and proprietary trading. It shows a quite stable pattern along the years; the total number of selected banks is lower than for HLEG.

· the definition based on net volumes is used to identify proprietary trading only. In general the results based on this definition are slightly less stable over time with respect to the others; the total number of selected banks is lower than for HLEG.

· the use of a measure of distance from the thresholds would allow softening the definition of hard thresholds resulting from cluster.

Appendix A: Description of the SNL sample

The analysis is based on the SNL database[161] since it allows for a detailed disaggregation of balance sheet items related to assets held for trading and it allows distinguishing derivatives for trading from derivative for hedging. This breakdown is not available in other commercial databases such as Bankscope.

However the coverage of SNL for EU countries is still not complete, since SNL was historically focused on US. This appendix presents quality checks made on the sample of banks from SNL and quantifies how much this database is relevant for European banks.

In SNL we found 244 banks that are headquartered in EU27 and have a consolidated total assets available in 2011. These banks have also total assets to GDP ratio larger than 1% in 2011. The sample accounts for 75% of the total EU banking sector[162]. It contains most of the 65 banks listed by European Banking Authority (EBA) in its capital exercise except Banco Espirito Santo (Portugal), Raiffeisen Bank International (Austria), and KBC Bank (Belgium).These entities are “replaced” in the sample by their holdings or by some other relevant entity of the same financial group.

Comparing the available sample with data from another data provider (Bankscope) one can observe 8 large financial institutions having total assets greater than 100 bn EUR missing in the SNL sample: NatWest, Caisse de Depots et Consignations, Sparkasse Hessen-Thueringen, BNG, Depfa, Exor, UKAR, Deutsche Pfandbriefbank.

Table I.1 reports the repartition of the SNL sample by countries. No bank is available in Czech Republic or Estonia. Even if out of EU27, DNB ASA is in the sample as part of the capital exercise of EBA.

Table I.10: Geographical repartition of banks in SNL (total assets are given in billion EUR)

|| Number of Banks || Total Assets in 2011 (bn EUR) || || Number of Banks || Total Assets in 2011 (bn EUR)

BE || 7 || 1,044 || LU || 3 ||    148

BG || 4 ||      7 || HU || 2 ||     35

CZ || 0 ||      0 || MT || 2 ||       7

DK || 28 ||  798 || NL || 8 || 2,296

DE || 56 || 5,937 || AT || 17 ||   669

EE || 0 ||       0 || PL || 6 ||     78

IE || 6 ||  442 || PT || 5 ||    293

GR || 5 ||  296 || RO || 2 ||       7

ES || 19 || 3,244 || SI || 3 ||     27

FR || 7 || 6,983 || SK || 1 ||       2

IT || 27 || 2,879 || FI || 3 ||   106

CY || 4 ||      80 || SE || 5 || 1,484

LV || 1 ||       2 || UK || 21 || 7,747

LT || 2 ||       2 || NO || 1 ||   274

Depending on the definition of trading activities considered, the number of banks included in the calibration exercise might differ, since there are differences in the availability of the various items considered in each definition. For example AFS and trading liabilities are less populated than other items.

To get an insight in the banks which are not considered due to lack of data, Table I.2 provides the list of banks which are excluded from definition 1 (HLEG) due to lack of AFS data. The sample reduces from 245 to 185 banks.

Table I.11: 60 banks (average total assets (2006-2011) have no data for the computation of HLEG definition. These banks are located in DK, DE but also in AT, BE, and PL.

|| Company Name || Assets in bn EUR || || Company Name || Assets in bn EUR

BE || Argenta Bank || 32 || DE || Kreissparkasse Heilbronn || 7

BE || Landbouwkrediet NV || 11 || DE || Kreissparkasse Köln || 24

DK || Djurslands Bank || 1 || DE || Kreissparkasse Ludwigsburg || 9

DK || DLR Kredit A/S || 18 || DE || Kreissparkasse Waiblingen || 7

DK || FIH Erhvervsbank A/S || 15 || DE || L-Bank Baden-Württemberg || 60

DK || Hvidbjerg Bank A/S || 0 || DE || LBS Baden-Württemberg || 10

DK || Jyske Bank || 30 || DE || LBS Norddeutsche || 8

DK || Kreditbanken A/S || 0 || DE || LBS Westdeutsche || 10

DK || Lollands Bank A/S || 0 || DE || MBS in Potsdam || 9

DK || Nordjyske Bank A/S || 1 || DE || MERKUR BANK KGaA || 1

DK || Nykredit Realkredit || 168 || DE || Münchener Hypothekenbank || 35

DK || Østjydsk Bank A/S || 1 || DE || Nassauische Sparkasse || 13

DK || Ringkjøbing Landbobank A/S || 2 || DE || National-Bank AG || 4

DK || Salling Bank A/S || 0 || DE || Sparda-Bank Baden-Württemberg || 11

DK || Skjern Bank A/S || 1 || DE || Sparda-Bank Südwest eG || 8

DK || Sparekassen Faaborg A/S || 1 || DE || Sparkasse Essen || 8

DK || Sparekassen Himmerland || 1 || DE || Sparkasse Hannover || 13

DK || Svendborg Sparekasse || 0 || DE || Sparkasse Herford || 5

DK || Sydbank || 19 || DE || Sparkasse Krefeld || 8

DK || Totalbanken A/S || 0 || DE || Sparkasse Leipzig || 9

DK || Vestfyns Bank || 0 || DE || Sparkasse Münsterland Ost || 8

DK || Vestjysk Bank A/S || 4 || DE || Sparkasse Nürnberg || 9

DE || Bausparkasse Mainz || 2 || DE || Sparkasse Saarbrücken || 6

DE || Bayerische Landesbausparkasse || 10 || DE || Stadtsparkasse Düsseldorf || 12

DE || Berliner Volksbank eG || 10 || DE || Stadtsparkasse München || 15

DE || Deutsche Apotheker || 38 || DE || UmweltBank AG || 1

DE || Die Sparkasse Bremen AG || 11 || DE || Wüstenrot Bank AG || 15

DE || Hamburger Sparkasse || 36 || AT || Raiffeisen Bausparkasse GmbH || 9

DE || Kreissparkasse Biberach || 6 || AT || Raiffeisen-Landesbank Tirol AG || 6

DE || Kreissparkasse Göppingen || 5 || PL || Getin Holding SA || 8

Appendix B: Additional graphs on selected banks

Figure II.1 shows the share of trading activities component for banks selected only under HLEG definition. It shows that their selection is mainly due to the size of their AFS (blue bars). Figure II.2 plots the total amount of trading activities for these banks. The order is given by their total assets.

Figure II.12: Breakdown of HLEG definition 1 trading activities, ordered by descending bank’s total assets.

Figure II.2: Size of trading activities (HLEG definition 1) ordered by descending bank’s total assets.

Appendix C: Stability over the time with respect to the proposed thresholds

Setting hard thresholds could raise problems with respect to the stability of the set of selected banks over the years. Being able to identify those banks which are close to the thresholds (above or below) may help the supervisor to monitor borderline situations. For instance banks closest to the thresholds could discretionally be added by the supervisor to the list of selected banks.

This appendix presents an analysis investigating the stability of selected sample of banks across the years focusing only on those which are “close” to the thresholds. We basically compute the geometric distance of each bank from the two thresholds, and we focus on the 10% banks with lowest distances (above or below the thresholds).[163]

It is worth mentioning that banks which are close to the thresholds considering the six year average, tend to be close to thresholds also for the three year moving averages (see Table III.1-2-3-4). We also observe that many of the banks close to the thresholds are medium sized banks (notably BBVA, which has been recently added to the list of the G-SIBs) that the supervisory authority might want to monitor.

We start from HLEG and we present in Figure III.1 the scatter plots relative to the six year average and also to the three year moving periods. The straight lines in the graphs represent the thresholds chosen under this definition (80 bn EUR in absolute term and 20% for the share).[164]

Figure III.13: 10% banks (red circles) with minimal distance from the thresholds according to HLEG definition 1.

Table III.1  Banks with minimal distance from the thresholds according to Definition 1 appearing in the 2006-2011 average and in at least three other average periods. 1=banks included in the 10% set closest to the thresholds; 0=banks not included in the 10% set closest.

|| Average 2006-2011 || Average 2006-2008 || Average 2007-2009 || Average 2008-2010 || Average 2009-2011

Allied Irish Banks || 1 || 1 || 1 || 1 || 0

Bayerische Landesbank || 1 || 0 || 1 || 1 || 1

BBVA || 1 || 1 || 1 || 1 || 1

BPI || 1 || 1 || 1 || 1 || 1

CGD || 1 || 1 || 1 || 1 || 1

Danske Bank || 1 || 1 || 1 || 1 || 1

HSH Nordbank || 1 || 1 || 1 || 1 || 0

Intesa || 1 || 1 || 1 || 1 || 1

Monte dei Paschi Siena || 1 || 1 || 0 || 1 || 1

Rabobank Group || 1 || 1 || 1 || 1 || 1

SEB || 1 || 0 || 1 || 1 || 1

Table III.2: Banks with minimal distance from the thresholds according to Definition 2 appearing in the 2006-2011 average and in at least three other average periods. 1=banks included in the 10% set closest to the thresholds; 0=banks not included in the 10% set closest.

|| Average 2006-2011 || Average 2006-2008 || Average 2007-2009 || Average 2008-2010 || Average 2009-2011

Bayerische Landesbk || 1 || 1 || 1 || 1 || 1

BBVA || 1 || 1 || 1 || 1 || 1

Belfius Banque || 1 || 0 || 1 || 1 || 1

Groupe BPCE || 1 || 0 || 1 || 1 || 1

Handlowy Warszawie || 1 || 1 || 1 || 1 || 1

Intesa || 1 || 1 || 1 || 1 || 1

KBC || 1 || 1 || 1 || 1 || 1

Landesbank BW || 1 || 1 || 1 || 1 || 1

Monte Paschi Siena || 1 || 1 || 1 || 1 || 1

Rabobank || 1 || 1 || 1 || 1 || 1

SEB || 1 || 1 || 1 || 1 || 1

Standard Chartered || 1 || 1 || 1 || 1 || 1

Swedbank || 1 || 1 || 1 || 1 || 1

Table III.3: Banks with minimal distance from the thresholds according to Definition 3 appearing in the 2006-2011 average and in at least three other average periods. 1=banks included in the 10% set closest to the thresholds; 0=banks not included in the 10% set closest.

|| Average 2006-2011 || Average 2006-2008 || Average 2007-2009 || Average 2008-2010 || Average 2009-2011

Bayerische Landesbk || 1 || 1 || 1 || 1 || 1

BBVA || 1 || 1 || 1 || 1 || 1

Belfius Banque || 1 || 1 || 1 || 1 || 1

HSH Nordbank || 1 || 1 || 1 || 1 || 1

Intesa || 1 || 1 || 1 || 1 || 1

Landesbank Berlin || 1 || 0 || 1 || 1 || 1

Landesbank BW || 1 || 1 || 1 || 1 || 0

Monte Paschi Siena || 1 || 1 || 1 || 1 || 1

NIBC Bank NV || 1 || 0 || 1 || 1 || 1

Rabobank Group || 1 || 1 || 1 || 1 || 1

SEB || 1 || 0 || 1 || 1 || 1

Spar Nord Bank A/S || 1 || 1 || 1 || 1 || 0

Standard Chartered || 1 || 1 || 0 || 1 || 1

Table III.4 Banks with minimal distance from the thresholds according to Definition 4 appearing in the 2006-2011 average and in at least three other average periods. 1=banks included in the 10% set closest to the thresholds; 0=banks not included in the 10% set closest.

|| Average 2006-2011 || Average 2006-2008 || Average 2007-2009 || Average 2008-2010 || Average 2009-2011

Banco Popolare || 1 || 1 || 1 || 1 || 1

BBVA || 1 || 1 || 1 || 1 || 1

Commerzbank || 1 || 1 || 1 || 1 || 1

Danske Bank || 1 || 1 || 1 || 1 || 1

DZ Bank AG || 1 || 1 || 1 || 1 || 1

Intesa || 1 || 1 || 1 || 1 || 1

KBC || 1 || 0 || 1 || 1 || 1

Landesbank BW || 1 || 1 || 1 || 1 || 1

Nordea || 1 || 1 || 1 || 1 || 1

Portigon AG || 1 || 1 || 1 || 1 || 1

SEB || 1 || 1 || 1 || 1 || 1

Appendix D: Definitions of trading activities with EMIR/FTT corrections

IV.1 Additional definitions and corresponding thresholds

This appendix describes additional definitions developed by Commission Services to take into account some other ongoing Commission legislative initiatives that might impact on the volumes of trading activities, notably EMIR (European Market Infrastructure Regulation), which will promote netting of derivative through CCPs, and FTT (Financial Transaction Tax), which will potentially reduce the size of traded derivatives.

To account for the combined effects of these two initiatives, we assume that only a fraction of the derivatives exposure should be added to the trading estimate.[165][166] The additional definitions that we examine are shown in Table IV.1. We repeat for these additional definitions the same analysis presented in the main text. Results are presented in the following tables and graphs.

Table IV.1: Additional definitions of trading activities to account for EMIR/FTT.

|| Definition || Description

2b || TSA + 35% DA || Exclude AFS + EMIR/FTT effect

3b || (TSA + TSL + 35% (DA + DL))/2 || Exclude AFS + Gross volumes of securities + EMIR/FTT effect

4b || |TSA – TSL| + 35% |DA - DL| || Exclude AFS + Net volumes + EMIR/FTT effect

Table IV.2: Proposed thresholds for definitions accounting for EMIR/FTT based on cluster analysis.

|| || || Cluster-based thresholds || Selected banks

|| Definition || Trading Activities (bn EUR) || Share of Trading Activities (% Total Assets) || Number || Share of the sample

2b || TSA + 35%DA || 50 || 10% || 36 || 15%

3b || (TSA + TSL + 35%(DA+DL))/2 || 50 || 8% || 34 || 14%

4b || |TSA – TSL| + 35% |DA - DL| || 30 || 8% || 31 || 13%

Table IV.3: HLEG equivalent thresholds for definitions of trading activities to account for EMIR/FTT.

|| || || HLEG equivalent thresholds || Selected banks

|| Definition || Trading Activities (bn EUR) || Share of Trading Activities (% Total Assets) || Number || Share of the sample

2b || TSA + 35%DA || 53 || 10% || 36 || 15%

3b || (TSA + TSL + 35%(DA+DL))/2 || 49 || 8% || 34 || 14%

4b || |TSA – TSL| + 35% |DA - DL| || 36 || 8% || 25 || 10%

Figure IV.1: Clusters for definition 2b (HLEG without AFS and including EMIR/FTT effect)

Figure IV.2: Clusters for definition 3b (volumes without AFS and including EMIR/FTT effect). Note: We note that set of 4 banks are circled in Figure 13 (Deka bank, Portigon, DNB ASA, Swedbank). These banks are characterized by a relatively high share of trading activities

 

Figure IV.3: Clusters for definition 4b (net volumes without AFS and including EMIR/FTT effect).

Comparing the banks selected through the all 7 definitions we found that:

· The 20 banks systematically selected in the 4 definitions without EMIR/FTT correction (main text) would be also proposed for structural separation using the definitions corrected for EMIR/FTT reforms, with the exception of KBC and Santander which are not selected by definition 4b.

· None of the banks that were never selected by any definition 1, 2, 3 or 4 is selected by any of these additional new definitions.

· G-SIBs are always selected by the definitions with EMIR/FTT correction, except Santander under definition 4b.

Figures IV-4-IV-6 details on the differences between the selected banks with and without EMIR/FTT correction.

Figure IV.4: Comparing definition 2 and 2b.

Figure IV.5: Comparing definition 3 and 3b.

Figure IV.6: Comparing definition 4 and 4b.

IV.2 Sensitivity to the parameter capturing EMIR/FTT effects for definitions 2b and 3b

This section describes the sensitivty of the thresholds to the parameter chosen to introduce the effect of EMIR/FTT. We present the results obtained, for definition 2b (HLEG, no AFS and EMIR/FTT) and definition 3b (volumes, no AFS and EMIR/FTT), using two different correction coefficients: 45% and 25% instead of 35%.

The following graphs show cluster analysis results varying correction coefficients, while Table IV.4 shows shows the variability of the thresholds.

Table IV.4: Sensitivity of the thresholds to the EMIR/FTT correction parameter for definitions 2b and 3b.

Definition 2b || 35% (baseline) || 25% || 45%

Threshold Trading Activities (bn EUR) || 50 || 50 || 50

Threshold Share Trading Activities || 10% || 8% || 10%

Definition 3b || 35% || 25% || 45%

Threshold Trading Activities (bn EUR) || 80 || 80 || 80

Threshold Share Trading Activities || 10% || 10% || 12%

With respect to the list of banks selected, using the new correction coefficients, the differences with respect to the baseline scenario (35% correction) are:

- For definition 2b

o for 25% correction coefficient Raiffeisen-Landesbk Steiermark, Bayerische Landesbank  and BBVA (G-SIB bank) are added

o for 45% correction coefficient only  Bayerische Landesbank  is added

- For definition 3b

o for 25% correction coefficient Credito Emiliano, NIBC Bank NV and Bank Handlowy w Warszawie SA are not selected and there is no new entry

o for 45% correction coefficient only Credito Emiliano is not selected while  Intesa, Rabobank Group and Lloyds Banking Group are added

We conclude by pointing out the the possible effects of EMIR/FTT need to be further invesigated whenever more accurate data on their impact would be available.

Appendix E: Structural indicator analysis

This appendix presents some statistics on a set of structural indicators that can be computed using balance sheet data. The aim is to show banks in the sample, apportioned into groups depending on their distance from the thresholds behave in terms of such indicators. We focus on the 4 definitions presented in the main text.

For each definition banks in the sample are assigned to four different classes, based on their distance from the thresholds[167]. In order to differentiate between selected and non-selected banks the latter are assigned a "negative distance”. The four classes are the following:

‘IN Top’ banks, i.e. 10% of banks in the sample with largest positive distance  from the thresholds (i.e. far away from the thresholds and selected). 2 groups are identified around the thresholds: 1 slightly above the thresholds’ values is called ‘IN Borderline’ and contains the remaining banks proposed for structural reforms but not in the top 10%. 1 slightly below the thresholds’ values is called ‘OUT Borderline’ and contains the banks not proposed for this reform[168].

Both groups are small in size (less than ten banking entities) and accounts for roughly 5-10% of the sample. These groups are key elements to assess if the calibrated thresholds for the choice of the listing: they need to make sense and be stable over time.

 ‘OUT Bottom’ banks, the rest of the sample, i.e. banks not selected and most distant from the thresholds.

For each definition, the following structural indicators are computed for the four classes above

1) Share of RWA for market risk on total RWA

2) Number of branches (complexity)

3) Non loans assets on total assets

4) Adjusted IFRS Tier1 leverage ratio

Share of RWA for market risk on total RWA

The following graphs (VI.1-VI.4) represent the evolution over 2008-2011 of the share of RWA for market risk on total RWA for the various groups of banks defined above under the various definitions.

Figure V.14: RWA shares over total assets median values for the four groups with respect to definition 1-4 over the years

One can observe that the share of RWA for market risk is higher for selected banks. In particular banks that are “IN Top” are systematically more exposed to market risk with respect to definition 1 and, in this case, there is also a clear distinction between “IN Borderline” and “OUT Borderline”. The situation is less clear for the other three definitions where only ‘OUT Bottom’ banks result to be systematically distinguishable.

Complexity

The number of branches is considered in order to proxy the complexity of banking groups. Complexity is here understood as a factor of risk (in case of failure for example). The following graph represent the number of branches grouped by distance from the threshold for each definition.

Figure V.5:Complexity calculated as the number of branches

We observe that banks that are above the threshold tend to have more branches except for definition 4 (netting) which excludes some banks with significant number of branches.

Share of non-loans on total assets

The share of non-loans assets measures non-retail banking activities. The following graphs represent the evolution the share of non-loans assets (in %) grouped by distance from the thresholds.

Figure V.9: RWA shares of non-loans over total assets median values for the four groups with respect to definition 1-4 over the years

We expect higher values of this ratio for banks doing a lot of trading activities, thus for banks of groups ‘IN Top’ and ‘IN Borderline’. This is the case for definition 1. For definition 2 we have a switch between IN Top and IN Borderline while for definitions 3 and 4 Borderline groups are not clearly separated.

Adjusted IFRS tier 1 leverage ratio

Adjusted IFRS tier 1 leverage ratio corresponds to a simple indicator which is comparable between banks and transparent in its computation for measuring the risk of the banking entities. This ratio is computed by SNL and corresponds to tier 1 capital divided by tangible assets less derivatives liabilities in the attempt to replicate U.S. GAAP standards by roughly netting the derivatives assets against liabilities. The ratio is expected to be at least 4-5% for the banks that are well capitalized.

The following graphs represent the evolution IFRS tier 1 leverage ratio grouped by distance from the thresholds.

FigureV.13: Adjusted IFRS tier 1 leverage ratio median values for the four groups with respect to definition 1-4 over the years

We can observe that, based on this definition,

‘OUT Bottom’ group is consistenly showing ratio above 5% which indicates low risk. Definition 3 show lower values of IFRS tier 1 leverage Ratio  for the “IN Top” group while with respect we have a complete different picture.

We conclude this annex highlighting the following:

“structural” indicators distinguish the group of banks further above the threshold from the others. The indicator share of market risk on RWA shows a greater capacity in separating the four groups (IN Top, IN borderline, OUT Borderline and OUT Bottom). We tackle complexity of the financial groups considering the number of branches. There is also an increase of complexity within our four groups except for definition 3 (volume).

Appendix F: Cluster methodology for thresholds’ calibration

This appendix briefly presents the main features of the clustering technique used for thresholds identification.

            K-means

Statistical clustering is used to assign banks into groups (called clusters) so that banks in the same cluster are “more similar” respectively to the measures TradAct and ShareTradAct than to those in other clusters. The algorithm requires choosing the number of clusters to be built and then it searches for the centroids minimizing the dispersion of points within each cluster.

Clusters are built starting from random centroids in the two dimensional space identified by the indicators and moving the centroid positions so to minimize dispersion among the clusters. In order to assures that the final positions of the clusters do not correspond to a local minimum of the measure of dispersion we proceed as it follows:

            T-clustering

For our data such models are insufficient, because they do not account for the presence of outliers, which may occur as noise-like structures or as a small tight group of observations in specific areas of the space. In both cases, the presence of outliers can considerably bias the estimation of the centroids and shape (covariance structure) of the groups and seriously affect the final clustering.

For this reason, we opted for a robust counterpart of the Normal Mixture Modeling known in the literature as Robust Trimmed Clustering TCLUST,

The robustness capacity of TCLUST comes from the trimming approach, i.e. the possibility to leave a proportion of observations, hopefully the most outlying ones, unassigned.

The TCLUST approach is defined through the search of k centers and k shape matrices solving a double minimization problem.

The method has been implemented in Matlab in the framework of the FSDA project

Annex A9: Summary of the main findings in literature on Economies of scale and scope in the banking sector

1. Introduction

Banks have been rapidly increasing in size and scope over the two decades leading up to the financial crisis. For example, Wheelock and Wilson (2011) estimate a five-fold increase in the average size of US banks in terms of inflation-adjusted total assets in the period 1984-2008. Also, many banks diversified and expanded their activities, contrary to non-financial firms, in the years leading up to the financial crisis (Elsas et al. (2009)). There is a significant body of literature on economies of scale and scope trying to explain the reasons why banks have chosen to expand their size and range of activities.

As structural reform sets limits on the activities of banks, this literature contributes to the understanding of efficiency benefits that would be lost as a result of the new requirements. Understanding whether there are efficiency gains from scale and scope, and if so, to what extent, until what level, and for which activities, serves as an important element for this impact assessment for structural reform. Structural reform may interfere with the extent to which banks' activities experience economies or diseconomies of scale and scope, as some options may impose full separation of certain activities. Limits on the organizational complexity and diversification of large financial institutions may have important implications, not only for risks and market valuation of large financial firms, but also for corporations, households and other financial institutions through the supply of financial services, the sources of credit available to borrowers, and the allocative efficiency of capital markets.

There are three critical questions on the structural reform proposal: which banks are subject to structural reform, which activities will need to be separated, and what will be the strength of this separation requirement. These questions are linked with the question of economies of scale and scope. The first question, which banks are subject to structural reform, relates to the literature on scale. Since only relatively large banks are likely to be affected by structural reforms proposals, one needs to assess whether, in particular, economies of scale exist for these levels of assets, or if they are exhausted past a certain size. The second question is directly linked to the diversification benefits and synergies between different bank activities. If there are no real (net) synergies, then the fact that banks engage in both types of activity may be interpreted as resulting from regulatory distortions, such as implicit government subsidies. The third question, on the strength of separation, relates to whether any potential efficiencies of scale and scope will be affected given the strength of separation requirements, and to what extent. Except for the case of full separation, other options impose specific legal, economic, and operational restrictions on deposit-taking entities, but allow banks to continue to offer a broad spectrum of services and obtain any related benefits.

The evidence on economies of scale and scope are ambiguous. Studies by the industry (IIF (2010), the Clearing House, (2011)) find significant economies of scale and scope, arguing that larger and more diversified banks are in a position to realise synergies, and therefore promote safer and more stable banks. As a result, structural reform would be detrimental not only for the bank's shareholders, but also for society. Academic studies, however, are more cautious in their conclusions concerning the existence of economies of scale and, in particular, scope. There is no consensus on the optimal size of banks, and there is no evidence that they exist for high levels of bank assets. Similarly, on economies of scope, while there is some evidence that there are benefits from loan and geographic diversification, the evidence for other activities is mixed, and there is a large body of literature suggesting that product diversification has detrimental effects (diseconomies of scope) that outweigh any benefits.

This Annex provides a summary of the main findings of the literature on economies of scale and scope in the banking sector. Section 2 discusses economies of scale and section 3 economies of scope. Section 4 discusses the impact of structural reform requirements on these (potential) efficiencies, and Section 5 concludes.

2. Economies of Scale

A firm is said to be operating with increasing returns to scale if it can lower average cost of production by increasing its size. There are several sources of economies of scale. Firstly, economies of scale may arise as larger banks would benefit from spreading overhead costs (reducing unit operating costs). Secondly, economies of scale may be the result of better diversification, as large banks are more likely to achieve wider scope in multiple activities, while at the same time maintaining scale in an individual activity. Similarly, banks might enjoy economies of scale from the diversification of risk obtained from a larger portfolio of loans and a larger base of deposits (i.e. increased ability to better match assets and liabilities due to scale).

The most direct source of economies of scale arises from spreading overhead costs, in particular those associated with information technology. Given that fixed costs are not very significant for some banking activities, economies/dis-economies of scale are unlikely to affect all types of activities in the same way. Activities with high fixed costs include payment systems, market infrastructure, and technology. The Liikanen report (2012) considers scale economies to be more prominent in payment and clearing services (due to the importance of fixed costs) compared to securities underwriting (which requires a more individual assessment of the relevant, individual deal). Also, Humphrey (2009) finds evidence of strong economies of scale for certain traditional banking services, such as the provision and processing of payment transactions. By observing the market structure in certain activities, Gambacorta and van Rixtel (2013) claim that scale (however not necessarily scale economies) is a defining characteristic in banks' capital market activities, as evidenced by the large share of the top 3 players in total trading volumes for cash equities, fixed income, FX, structured products, and listed derivatives.

On the other hand, there are several adverse effects of larger scale. Firstly, there are concerns that as banks get larger their market power increases, and thereby the likelihood that they would abuse their market position becomes greater. Secondly, larger banks may imply larger risks for a country's public finances as they are more likely to benefit from TBTF subsidies. Thirdly, consolidation may lead to credit availability composition effects as smaller banks tend to have stronger relationships with smaller firms. Also, consolidation may be driven by managerial benefits, which increase with a bank's size.

a. Economies of scale in empirical studies

The estimates on the significance of economies of scale vary significantly. At one extreme, there are some industry studies that claim significant economies of scale. For example, a study by the Clearing House (2011), puts the estimate of economies of scale to around USD 25 to 45 billion annually in the US, by comparing the actual costs of the banks to the costs in a (hypothetical) system under which no bank would be larger than USD 50 billion in assets. They find that the largest benefits are in payment (USD 10-20 billion) and capital markets (trade processing USD 5-15 billion). This study however, is based on a static comparison between institutions without controlling for any underlying differences of the banks.

The results of academic studies are ambiguous on what is the optimal size of banks and in particular on whether there are economies of scale at high levels of bank assets. Early studies have difficulties in establishing significant and substantial economies of scale, at relatively moderate levels of banks size (USD 100 billion). Some recent studies even though provide evidence of economies of scale at high levels of assets they do not show that these dominate the diseconomies of scale (such as higher risk-taking) and do not control for the impact of implicit subsidies. Once implicit subsidies are accounted for, diseconomies of scale arise at relatively moderate levels of bank size. As structural reform would only apply to very large banks, the pertinent question for assessing the impact of the structural reform proposal is whether economies of scale get exhausted, and if so, at what level of bank size as for example it is unlikely that banks in excess of this threshold will be subject to structural reform requirements (see Annex A8 for the discussion on thresholds).

Empirical literature typically explores the cross-sectional efficiency of banks of different sizes at a given point in time. This is achieved by deriving a minimum cost function after first calculating input prices. Then, by applying this to bank data, scale economies are computed from the fitted cost function. An alternative method is to look into the time series efficiency of banks on either side of a bank merger.

b. Cross sectional data

Early empirical academic analysis based on cross-sectional data has found limited evidence of scale economies which only peak at relatively low levels of assets of USD 10-100 million (see Saunders (1996) and Berger and Mester (1997)). However, more recent work on cross-sectional efficiency based on more recent data and improved methodology find stronger evidence of economies of scale.  Wheelock and Wilson (2011), for example, consider that there are both methodological and structural reasons that early work could not detect economies of scale at high levels of assets. Firstly, they consider that the empirical methods used for estimating the cost function use models that do not capture key features of bank production. [169] Secondly, more recent work could find evidence of scale due to regulatory changes that has made it less costly to become larger (such as relaxation of restrictions on geographic branching and product expansion). Thirdly, technological advances may have had a greater impact in recent years (e.g. information processing equipment and software entails relatively high fixed costs and at the same time the costs of acquiring information about potential borrowers, which has been the competitive advantage of smaller banks, has fallen).

Also, Hughes et al. (2001) consider that there is an additional reason for the lack of strong evidence of economies of scale in the early studies. They claim that there is an issue of identification in such studies, as managerial inefficiencies related to scale may mask positive economies of scale. This phenomenon would occur if there are two opposing effects: on one hand, a positive effect from scale and, on the other hand, if banks respond to the lower cost by taking on more risk, there would be a second, adverse effect,  which is linked to the risk-taking effect if banks spend more resources to manage increased risk.  Their work (as well as work by Hughes and Mester (2011)) has taken into account output measures corrected for banks' risk and find evidence of economies of scale at much higher levels, that is, above USD 100 million. Hughes and Mester (2011) under such a setting not only find economies of scale but also that these scale economies may continue to increase with bank size. They also claim that their estimates are not driven by TBTF considerations, as economies of scale apply also to smaller banks too.  They find that when a bank changes size there are two beneficial effects: one "pure" size effect and one due to a "change in output mix" (an activity mix that is more consistent with the size of the bank) and claim that the latter effect is slightly greater than the first. However, while their approach allows identifying a scale effect, they do not examine whether this effect outweighs the negative effects of managerial diseconomies (i.e. the second adverse effect) related to size.[170] Furthermore, they acknowledge that TBTF considerations cannot be dismissed as an additional factor in explaining the banks' increasing size.

In contrast with the more recent work on economies of scale that find increasing evidence at higher levels of assets, Haldane (2012), referring to work by Davies and Tracey (2012),[171] considers that when taking into account TBTF implicit subsidies, these strong results on the existence of economies of scale disappear.  The authors employ credit rating data to adjust the cost of debt by considering only the standalone rating of banks, rather than the rating including government support (see also Annex A4.1 and A4.2 for details on the different credit agency ratings). Through this adjustment, they control for the implicit government subsidy and they find that the banks' funding costs increase with size, lowering the estimates of bank value-added and the measured economies of scale. In particular, they claim that there are no scale economies in a sample of large international banks with assets above USD 100 billion, as shown in the second chart below. Their findings support the claim that efficiency benefits gained from scale are potentially offset by diseconomies arising from the fact that some banks become too large to maintain effective management. Haldane (2012) therefore claims that implicit subsidies may have artificially increased the privately optimal banking scale compared to the optimal level from the social perspective.

Chart  1: Scale economies, from a standard model of bank production(a),(b),(c) || Chart  2: Scale economies, adjusting for the implicit subsidies(a),(b),(c)

||

(a) The results are for scale economies estimates over the period 2001 to 2010. A value equal to one, less than one, or greater than one implies constant returns to scale, scale diseconomies, and scale economies, respectively. || (b)Total assets have been adjusted to constant year-2010 prices using country level inflation rate data.  (c) Presented results are estimated at the median and interquartile range for each bank in each time period. The scale economies mean is evaluated at the mean of the data. Source:  Davies and Tracey (2012)

Therefore overall, academic studies that find evidence of economies of scale at relatively high level of bank assets identify this effect separately from any diseconomies of scale such as increased risk taking related to higher bank size and from the implicit subsidies.

c. Time series data

There are also a number of studies that look into the efficiency of banks on either side of a bank merger. Such studies do not provide strong evidence of ever-increasing economies of scale (see for example Berger and Humphrey (1997), Berger at al (1999) and De Long (2001)). Berger at al (1999) argues that even if there are some economies of scale, these may have been offset by managerial difficulties in monitoring the larger organizations, conflicts in corporate culture, or problems in integrating systems. Also, Amel et al. (2004) find that mergers may be beneficial up to a relatively small size, and they find little evidence that mergers yield significant gains in efficiency.

Overall, while there appears to be relatively strong evidence for economies of scale at relatively low bank sizes, there is no consensus on the optimal size of banks, and there is no strong evidence that the economies of scale apply at very large bank sizes (for example above $500 million in assets).  This is particularly relevant for the current proposal of structural reform as its provisions only applies to banks above the thresholds as discussed in Annex A8, that is, to very large banks in the EU.

d. Dis-economies of scale

Furthermore, even if economies of scale do arise, the question that arises is whether or not these benefits are likely to be passed on to customers (through changes in prices or quality of service because of competitive pressures) and why smaller firms would not be able to form consortia or outsource some activities to realise similar levels of economies of scale as larger banks. Moreover, as explained above, the literature has identified a number of disadvantages related to increased size that relate to the following:

1. By expanding in scale and scope, banks may be able to raise prices above marginal costs, exploiting their stronger market position (market power). However, the literature on banking sector concentration and prices is not clear-cut, largely because studies on mergers do not always control for the efficiency effects of a merger (see for example Herring and Carmassi (2008) and Berger et al. (1999)).

2. Larger banks are more likely to take higher risks. As explained above, Hughes et al. (2001) claim that there is an adverse effect associated with increased size, the risk-taking effect, if banks spend more resources to manage the increased risk and monitoring costs associated with larger size. Haldane (2009) considers that "at least during this crisis, big banks have if anything been found to be less stable than their smaller counterparts, requiring on average larger-scale support".[172] This may be related to the TBTF status of larger banks. The expectation that some banks benefit from TBTF subsidies provides an incentive to banks to artificially increase their size to benefit from such subsidies. This positive relation between size and implicit subsidy is also documented in the literature (see Annex A4.1 and A4.2). As the implicit subsidies increase for a given bank this also leads to less effective market monitoring of the banks effectively subsidising risk-taking by systemically important financial institutions.

3. Larger banks pose greater systemic risks, and this is likely to imply large risks to a country's public finances. There are a number of studies that find that larger banks pose greater systemic risks. Boyd et al. (2006) show that in countries with more concentrated markets, banks have taken a disproportionate amount of risk, relative to their capital buffer. Boyd and Heitz (2012) estimate that the social cost of too-big-to-fail banks due to increased systemic risk is significantly higher than the benefits due to economies of scale. Also Baele et al. (2007) for a European sample, find that larger and more diversified banks have higher systemic risk. On the other hand, Beck et al. (2006) find evidence that concentrated systems have a lower probability of financial crisis, potentially due to better diversification of risks within large banks, leading to a correspondingly lower probability of failure. Demirgüc-Kunt and Huizinga (2011) distinguish between a banks' absolute size and its systemic size measured with respect to the size of the economy. They conclude that while there may be some benefits from absolute size (returns on assets increase with absolute size, even though bank risk increases), systemic size is unambiguously bad (returns on assets falls with banks liabilities-to-GDP ratio and bank risk is unaffected) meaning that the optimal bank size may be larger for banks in larger economies.

4. Greater consolidation may also have consequences in the composition of companies to which funding is channelled in the real economy. Small banks tend to be better at relationship-lending based on "soft information", such as reliability of the firm's owner versus lending by big banks that is based on "hard information" such as financial statements and credit scoring (Berger et al. (2005)). Therefore, greater consolidation may affect the ability of small firms to secure credit availability.

5. Principal-agency problems, and in particular managerial benefits, may have also led to higher than optimal bank size. Anderson and Joeveer (2012) show that there is stronger evidence of returns to scale to bankers as compared to returns to investors, and that these returns to bankers are particularly strong in banks that have a large share of non-interest income. Also, Hughes et al. (2003) find empirical evidence that bank managers may sacrifice value to build empires (through mergers) and not all consolidation that has taken place is value-enhancing.

3. Economies of Scope

There is a substantial and diverse literature on the subject and yet lack of unanimous evidence supporting the existence of economies of scope. While there are several source of economies of scope such as revenue and cost economies of scope and risk diversification benefits, there are also important diseconomies of scope. Conflicts of interest between different activity business lines, increased risk-taking and complexity as well as systemic risk and cultural contamination are reported as factors that lead to adverse effects of extending banks' scope of activities. The empirical evidence is a relatively stronger for geographic diversification. 

As previously described with regards to economies of scale, if the net benefits of economies of scopes are passed on to customers, the client may benefit from a greater range of products on offer by a single institution, and potentially lower prices.  In answering whether or not economies of scope exist, for what activities, and at what levels, one must analyse existing literature. The sections that follow will analyse in further detail examples of evidence for or against the enjoyment of economies of scope when a bank diversifies its activities, domestically through loan and portfolio diversification, and geographically.

a. What are the economies of scope that a financial institution may enjoy?

Economies of scope relate to efficiency gains of diversifying business activities across products and services provided (product diversification), as well as geographical diversification, or a combination of the two. The possibility of a financial institution benefiting from economies of scope is somewhat related to that of scale: as a financial institution grows in size, it may enjoy improved efficiency associated with diversification, and thus, reduced costs of funding. 

The sources of economies of scope and therefore underlying reasons for which a bank may choose to diversify their business practices, functions, or products offered can generally be attributed to:

Revenue economies of scope: clients may place additional value in being able to seek multiple products at a single bank offering diversified services.  Additionally, in providing a service, the bank gains valuable information about their client that may provide process and pecuniary advantages in the provision of other services; Cost economies of scope: banks may reduce their operating costs in engaging in a wide range of activities from pooling their resources.  Sources of these reductions in costs could be from operating centralised IT and finance systems, for example; and Risk diversification: an extension of the cost economies of scope, risk diversification implies that undertaking a wide range of activities (or operating in more geographic areas)  with less than perfectly correlated income stream may benefit the overall banking group by diversifying assets and earnings thus rendering them more resilient to shocks, and reducing costs. Also, diversification across activities may lead to a more efficient use of capital for the bank which may be affected by structural reform as under several options the two entities would be required to meet these requirements on a solo basis. The ICB (2011) acknowledges that scope benefits arising from moving excess capital between the different parts of a corporate group may be lost. If there is less than perfect correlation in the capital of these two entities then such separation requirements will increase the capital needs of the banking entities.

Contrastingly, most literature refers to the following drivers of inefficiencies associated with diversified institutions:

Conflicts of interest: conflicts of interest between banks' employees and banks' customers when banks engage in several activities may arise in several forms. For example, banks can use the informational advantage they gain from conducting several activities to their own advantage. Also the combination of different banking activities, and therefore of multiple clients and interests provides the opportunity to serve some client categories better to the detriment of others. Potential conflicts between traditional banking and securities underwriting business, for instance, led to the 1930 Glass Steagall Act; Increased complexity: diversification of banks tends to increase their organisational and operational complexity, especially if they are large to begin with, which can increase their risk management costs. This complexity can also lead to reduced transparency, making effective supervision harder and complicating resolution; Increased risk-taking: lower costs of funding due to diversification may contribute to the diversified bank taking on additional and excessive risk. While many studies recognise a certain degree of risk diversification benefits, many note that the expansion of types of activities usually enters the realm of much riskier activities, and in parallel with this expansion, these banks often hold less capital than undiversified banks; Increased systemic risk: while individual activity, and therefore risk diversification, can benefit the single bank, it has been found to contribute to overall systemic risk as banks typically diversify into each other's' traditional areas. As a result, the financial system as a whole becomes less diversified;    Cultural contamination: an increasingly-discussed disadvantage of practicing varied banking activities since the beginning of the financial crisis is the changing of attitudes towards business practices. The transferral of behaviours typical of the trading floor in banking activities into the commercial side causes a lack of confidence in the sector, and is seen as detrimental to the proper, useful functioning of the European economy. 

As a result of the above mentioned drivers, benefits of functional and organisational diversification may manifest themselves in having better access to internal capital markets and lower risk, increased supply of financial services, and operational synergies.  On the other hand, potential diseconomies of scope could arise from intensification of agency problems between the divisions of the conglomerate and between the conglomerate and its outsiders, bargaining problems and higher regulatory costs, inefficient rent-seeking, increased systemic risk due to negative externalities and more interdependencies and cultural contamination. 

b. Economies of scope in the empirical literature

Empirical literature provides mixed evidence for significant efficiency gains generated by large and diversified banks, very often concluding that there are diseconomies of scope, making it difficult to ascertain the value added of bigger and more varied banks. Much of the existing literature in the field refers to the Diamond (1984) model of banking activity, which provides the theoretical rationale for a bank acting as an intermediary in the financial market.  This model, which focuses on deposits and loans, finds that diversification benefits are inherent in the role of the bank as an intermediary (delegated monitor) in this market. Furthermore, banks' activities of providing commitment-based loans and accepting deposits are very similar services. That is, in both cases they provide liquidity on demand to accommodate unpredictable needs. If the two activities' demands for liquidity are imperfectly correlated then the two activities can share the costs of the liquid asset stockpile, or buffer, through a single bank (Kashyap, Rajan, and Stein (2002) and see also Gatev and Strahan (2006)). Overall, while this literature suggests economies of scope between deposit taking and loan provision there is limited evidence of such benefits for other kinds of activity diversification.

Economic literature on the subject of economies of scope is rich. One strand of literature looks directly at the effect of diversification (either activity or geographic) on the market valuation of the specific bank.  These studies find the net effect of economies and diseconomies of scope directly from stock market valuations. Other empirical analyses focus specifically on some categories of economies or diseconomies of scope. Therefore they are analysed under the prism of the categories identified in the section above.

c. Market Valuation studies

Without identifying the underlying drivers, the following studies draw links between the degree of operational diversification (either activity or geographic) of the bank and the value the market places on it. Overall, estimates of such valuation studies, in line with other studies, fail to provide robust evidence on the existence of economies of scope.

At one extreme, Laeven and Levine (2007) investigate whether the market valuation of a diversified bank is more or less than the value it would have if the conglomerate were broken into a portfolio of banks each specialising in the interest and non-interest earning activities of the conglomerate (chop-shop approach that compares these entities with similar entities specialising in interest and non-interest income activities).  The paper finds for a sample of banks across 43 countries a diversification discount in the conglomerates' valuation, as those banks that engage in multiple activities are valued lower by the market than the individual specialised units. These results suggest that any economies of scope in diversification are not sufficiently large to compensate for the detrimental effects of diversification. Schmid and Walter (2009) find similar results for US financial conglomerates. Other studies find that diversification has no material impact on the valuation of a financial conglomerate. Lelyveld and Knot (2009) find for 45 large financial conglomerates (firms that are active in both banking and insurance) in the EU that there is no uniform diversification discount or premium. When the conglomerate is split into a banking unit and an insurance unit, 52% of the sample shows a premium, and 48% show a discount. These effects are significantly variable in magnitude, depending on size, complexity, and risk. They also show that while small conglomerates witness a premium on average, larger conglomerates tend to face a discount. Similarly, DeLong (2001) studies US mergers over the period 1988-1995 and finds that activity-diversifying mergers do not have a positive announcement effect on share prices, contrary to focused mergers. At the other extreme, in a study of large international banks from 1996-2008, Elsas et al. (2009) find evidence for a diversification premium. They test how revenue diversification and increasing bank size affects bank value. They show that revenue diversification enhances bank profitability, and in turn higher profitability translates into higher market valuations.  They find the diversification effects materialise through an indirect effect on current operating performance and not through direct effects (of examining the interest vs non interest income of the two business units as in Laeven and Levine (2007)). Similarly, Baele et al. (2007), in a European sample, also find a positive relationship between franchise value and diversification. There are also a few studies on the effects of geographic diversification on market valuation. Gulamhussen et al. (2010), in an international sample across 56 countries, find that internationally diversified banks trade at a premium. However, they also show that geographic diversification has an inverse U-shape effect on bank value in that increased diversification increases the bank's value, only until a certain threshold, after which it starts to decline. Furthermore, Gulamhussen et al. (2012) show that this overall value enhancement comes at a cost: multinational banks have higher expected probability of default. They show that any positive effect of loan and asset diversification in the reduction of bank risk is outweighed by the negative effects of skewed incentives and complexity originating in international diversification and the creation of large multinational corporations. Economies of scope

The empirical literature on economies of scope has focused mostly on the benefits of risk diversification and informational advantages.

d. Revenue and cost economies of scope 

The works of Diamond (1991), Rajan (1992), Saunders and Walter (1994), and Kashyap et al. (2002), for example, theoretically suggest that banks acquiring information about clients during the process of making loans facilitates the efficient provision of other financial services, including securities underwriting. This positive information-sharing process can also work in the opposite direction, whereby underwriting, brokerage and mutual fund services, and other activities may improve loan-making procedures.  

Drucker and Puri (2005) empirically show for the US that there may be economies of scope to be enjoyed in concurrent lending and equity underwriting from spreading fixed costs of acquiring information.  They argue that the concurrent deals could provide benefits for the issuers in lower costs, and the degree of economies of scope enjoyed could be greater for commercial banks than investment banks, given their well-established lending business. These efficiency gains and resultant savings could be particularly pronounced for the issuers who are noninvestment grade-rated.  They indicate that beyond an optimal volume of underwriting deals, which is however not specified, the bank would experience diseconomies of scale. Other papers suggest that there is a positive reputation effect of mixing different activities. For example Ang and Richardson (1994), Kroszner and Rajan (1994) and Puri (1994) find that commercial banks did not try to mislead the public in investing in poor securities in the years leading up to the Glass-Steagall Act, as the performance of commercial and investment bank issues does not differ.

e. Risk diversification

If the returns of two or more sources of income are less than perfectly correlated, then it is possible to reduce risk through diversification.  However, it may be that multinational banks may take on excessive risks (see below).[173] Overall, there is lack of robust evidence on the effectiveness and the benefits related to risk from activity and geographic diversification. Literature considers two potential sources of risk diversification: activity and geographic.

Concerning activity diversification, certain papers analyse banks' income sources into interest-based and non-interest based (fee- or commission-based, for example) categories.

The IMF Staff Discussion Note (2013) finds evidence that a bank benefits from diversifying its business. They find that the returns from the retail section of  a sample of seven G-SIBs complement those from wholesale and trading banking in the particularly prosperous times (2003-2007), and vice-versa in economically tougher times (2008-2011). 

They claim that the trading banking segment complements retail and wholesale banking via product structuring, hedging, and income enhancements through proprietary risk-taking. Banks may diversify their risk portfolio by maintaining all three business lines. However this might not have a direct impact on the probability of default of different types of banks. For example, an earlier IMF staff discussion note (Ötker-Robe et al., 2011) shows from data throughout the crisis that the frequency of distress was notably higher for banks that practiced investment and universal banking activities than for commercial banks (likely reflecting more reliance on more volatile sources of funding and balance sheets more sensitive to mark-to-market accounting).  

Some papers underline the value of derivative activities within banks to allow diversification of risk. Brewer et al. (2000) focus on derivatives contracting and find in the US 1985-1992 that those commercial banks that engaged in the interest-rate derivative products markets have greater growth in lending in commercial and industrial loans than those banks that did not. They claim that their results are consistent with the role of banks as delegated monitors as in Diamond (1984): derivatives trading enables banks to increase their reliance on their comparative advantage as delegated monitors, and allows them to hedge their risks, and this leads to a reduction in delegation costs. This reduction in delegation costs, in turn, provides incentives for banks to increase their lending activities.[174] A similar point is made by Purnanandam's (2007). He shows that banks that make use of derivatives remained more insulated from monetary policy shocks during the period 1986-2003 compared to banks not using derivatives which instead decreased their lending following money supply contractions. Therefore, derivatives may contribute to maintain smooth operating policies in the event of external shock. Baele et al. (2007), on a European sample (over the period 1989–2004), find that a higher share of non-interest income in total income affects banks’ franchise values positively, reduces idiosyncratic risk, and makes banks safer. Nevertheless, they also find that the systematic risk of banks increases. Similarly, Rossi et al. (2009) examine data on Austrian commercial banks (1997 to 2003) and find that diversification reduces risk, in line with the classical diversification hypothesis.  Their results point also to a negative effect of diversification on cost efficiency (as it is associated with higher monitoring costs) but overall diversification increases profit efficiency and reduces banks’ risk. On the other hand, there is a series of articles (Stiroh (2004a), (2004b) and DeYoung and Roland (2001), for example), that claims that increased diversification leads to increased complexity, and excessive risk-taking, leading to overall higher bank risk (see section 2.2.5 below) In terms of geographic diversification, Amihud et al. (2002) study the effect of geographic diversification on risk. They show that cross-border mergers and acquisitions have no net effect on the risk (and returns) of the acquiring banks. They also find this to hold for both total risk and systemic risk as compared to their home market.  On the other hand, Deng and Elyasiani (2008) find that geographic diversification (using distance metrics within the US only, for BHCs) is associated with risk reduction (and lower stock price variability), along with value enhancement.  However, geographic diversification across more remote areas (measured as the distance between the holding company and its branches) is associated with smaller risk reduction. Furthermore, Deng et al. (2007) show that geographic diversification can lead to a funding advantage. They study a sample of over 60 US BHCs from 1994 to 1998 and find that domestic diversification of deposits reduces the bond-yield spread. They also show that medium-sized BHCs experience a greater reduction in bond yield-spread than small-sized and large-sized BHCs, which, they argue, is consistent with TBTF effects in the banking industry.

f. Diseconomies of scope

A number of diseconomies of scope have been identified above. Below we discuss the academic literature on each of these categories. There is an extensive literature discussing, in particular, conflict of interest as well as increased complexity and risk (both in terms of excessive risk-taking and increased systemic risk).  Before going in detail in the empirical finding, one should note that, as discussed in the section of economies of scale, studies measuring economies of scope should also take into account the benefits from safety nets. Indeed, economies of scope may arise due to the presence of the implicit subsidy and banks may be tempted to extend their activities in order to benefit from the safety net over all their activities. For example, when publishing their final report, the ICB received very little quantitative evidence of the magnitude of these claimed diversification benefits or customer synergies (ICB 2011). The lack of evidence stems from a difficulty in separating the two effects that may lower the funding costs of universal banks: an implicit government guarantee, and diversification benefits that result in a pecuniary advantage. However, there are no such studies.

g. Conflicts of interest

A conflict of interest in general can arise when an agent is serving two or more interests and has the ability to put one party in a better position at the expense of the other. Conflicts of interest become particularly important with increased scope, as service-providers have clients from distinct or opposing business lines. Conflict of interest may arise among different areas of activities and large literature has focused on universal bank's underwriting activities.

A conflict of interest may arise in the universal banking model if authorities allowed banks to underwrite their borrowers' capital market issues (Bhattacharya, Boot, and Thakor, 1998). A theoretical paper on combining lending and underwriting is provided by Kanatas and Qi (1998), who assess the conditions needed for the separation of lending and underwriting to be optimal.  They suggest that there is a social cost related to the bank's reduced incentive to monitor its borrowers, as credit risk can be shifted to uninformed public investors when the borrower's project is not performing well. The authors show that legal separation of lending and underwriting may improve social welfare if firms recognise the intermediary's subsequent incentive when choosing projects to be funded and there are large social costs associated with the bank's funding of poor-quality projects (for example through an increased likelihood of runs on banks). The Glass-Steagall Act prohibited commercial banks from underwriting or dealing in corporate securities to prevent lenders with adverse private information from selling securities of weak firms to an unsuspecting public, to offload credit risk. Studies from underwritings in the period pre- Great Depression (pre GSA) such as Kroszner and Rajan (1994) and Ang and Richardson (1994), however, find no evidence that commercial banks misled the public (by for example comparing the relative performance of securities offered by commercial and independent investment banks). Kroszner and Rajan (1994), however, find evidence that the conflict of interest exist and the market and the affiliates adapt their behaviour (for example the market would request a "lemons" discount and affiliates have avoided information-intensive securities and focused on better-known firms than investment banks). In addition to these research on the pre GSA period in the US, there are also some recent studies on the UK and Canadian experience (Hebb and Fraser 2002, 2003) which find evidence that commercial and investment bank issues do not differ in performance. However, other recent studies such as Johnson and Marietta-Westberg (2009) find evidence of conflicts of interest in relation to underwriting, when it is combined with asset management divisions. They show that universal banks with asset management divisions tend to use asset management funds as vehicles to help them earn more equity underwriting business. Similar results are obtained by Ber et al. (2001), who claim that banks must choose between selling the IPO stocks at a high price, generating a substantial amount of cash, and selling these stocks at a low price, generating good returns for investors. Bessler and Stanzel (2009) study conflicts of interest when the asset management analyst is affiliated with the underwriter of an IPO in a sample of German universal banks. They find that the analysts belonging to the lead underwriter tend to produce inaccurate and positively biased stock recommendations to public. Xie (2007) also studies the issue of conflict of interest in relation to securities underwriting and trading.  She finds that the negative effect of conflicts of interest dominates the positive benefits of economies of scale and scope in universal banking.  This negative effect is weaker in the countries with stronger protection of creditor's rights (higher institutional development, stronger security laws, higher accounting standards) or higher information efficiency of the stock market, as the conflict of interest are less likely.  Relating to the question on the strength of the fence, she also finds evidence to support that a subsidiary structure of separation of banking activities would reduce the likelihood of conflicts of interest.   Fecht et al. (2010) report empirical evidence for the German banking sector that proprietary trading can negatively affect retail customers. Stocks sold to retail customers of the bank underperform compared to other stocks in the bank’s proprietary portfolio and other stocks in the households’ portfolios. Customer portfolio performance is also significantly worse in banks that do proprietary trading. The authors attribute banks pushing stocks to their customers to the banks avoiding direct and indirect transaction costs, as well as not wanting to disclose their possible informational advantage to the market. They argue that conflicts of interest are at the source of these findings.

h. Increased complexity

            An adverse effect of engaging into more activities is that the institutional complexity increases, increasing the risk management costs.

· Klein & Saidenberg (2010) suggest that the diversification discount found in the literature (see market valuation studies, discussed above) reflects not only industry diversification, but also organisational structure. Potential costs of organisational structure include bureaucratic rigidity and bargaining problems, and these should be distinguished from the pure activity and geographic diversification benefits. The authors show that BHCs with many subsidiaries have lower profits and market valuations than similar BHCs with fewer subsidiaries, and accordingly argue that the cost of managing complex organisations increases with the degree of heterogeneity of the institution’s subsidiaries.

· This increased complexity may create problems for market participants and regulators to appropriately value and monitor the banks. Flannery et al. (2010), from pre- and mid-crisis data, argue that complex, diversified banks are seen by many as less transparent than other companies, and so the monitoring of their activities becomes more difficult. Iannotta (2006) analyses the opinions of credit ratings agencies to suggest that the greater complexity of large, diversified banks results in greater opacity. In particular, bank opaqueness increases with size and with the volume of financial assets.

· Increased complexity can also arise from geographic diversification. As explained above, Gulamhussen et al. (2012) show that any positive risk-reducing effect of portfolio and international diversification is outweighed by the negative effects of skewed incentives and complexity originating in the creation of large multinational corporations.

i. Excessive risk-taking

Banks may benefit from activity and geographic diversification, which could reduce their overall riskiness, but this positive effect may be more than or partially offset by opposing incentives, leading the banks to take on excessive risks. Some studies have underlined the potential of excessive risk-taking in banks as a result of consolidation and expanding activities in new markets.

While internationalisation, consolidation, and conglomeration offer potential benefits to financial institutions, diversification may also lead to shifts in risk-taking behaviour and the development of new and more sophisticated mechanisms to transfer risk (FSOC 2011). The Financial Stability Oversight Council (FSOC 2011) explain that pre-crisis, supervisors knew that much financial activity had moved from the banking sector to capital markets, but they did not fully appreciate the risks that certain activities posed to the institutions they supervised and to the financial system as a whole. De Nicoló et al. (2004) study the risk profile of financial conglomerates[175] compared to other financial institutions in an international sample of banks.  In a cross sectional analysis of financial institutions, the authors find that the financial conglomerates exhibited higher levels of risk-taking than smaller, specialised firms. Since they control for any positive effects on risk from diversification, their results suggest that the incentives for firms to take on more risk, including moral hazard-induced incentives, appear to have outweighed the risk reductions that would be achieved through scale or scope economies, or through geographic or product diversification.  They consider that this increased risk-taking may be the result of an extension of the safety net to non-bank financial firms if banking and non-banking activities are not effectively ring-fenced. Other papers study the effects of combining interest and non-interest activities on risk. Stiroh (2004a), for US community banks, and Stiroh (2004b), for US commercial banks, find that non-interest share of revenue enhances risk and return volatility, and is negatively related to risk-adjusted profits. This result suggests a robust, negative relationship between non-interest income and performance. Some studies have shown that in financial institutions, marginal increases in revenue diversification are not associated with a significant change in performance (DeYoung and Roland, 2001, for example), which may reflect either a change in managerial focus or may represent the endogenous nature of the diversification decision. It is argued that this problem arises as managers may enter businesses where they have little experience or comparative advantage. In their study of US financial holding companies, Stiroh and Rumble (2006) find that even though diversification benefits exist, they are more than offset by the increased exposure to more volatile non-interest activities. One proposed reason for this relationship may be that FHC managers have overestimated the benefits of diversification. If managers privately reap the gains of higher profits, but do not bear all the social costs from increased risk, there is indeed scope for risk mismanagement and managerial hubris.  Lepetit et al. (2007) find similar results for European banks. Their findings show that banks expanding into trading and fee and commission-earning activities present higher levels of risk and insolvency measures than banks mainly performing deposit-based banking activities.  Upon further division of the sample, the study demonstrates that smaller banks may benefit from a degree of decreased risk when they engage in trading activities. Contrastingly, both the overall asset growth and the increased share of non-interest income are positively related to risk for large banks. Other explanations may include empire-building, over-diversification to protect firm-specific human capital, or corporate control problems. One of the proposed justifications for the shift in activity in these banks is the mismanagement and estimation of risk, and managers seeking higher expected.  There are also some studies focusing on the effects of securitisation on risk. For example, Nadauld and Sherlund (2009) and Keys et al. (2010) provide evidence to suggest that securitisation encouraged underwriters to relax credit quality standards during the peak of the housing bubble.  In particular, the former construct a measure of geographic diversification and concentration to show that the securitisation process, including the assignment of credit ratings, provided incentives for securitising banks to purchase loans of poor credit quality in areas with high rates of house price appreciation. Therefore, they argue that securitisation allowed banks to transfer risk and discouraged banks from devoting as many resources as it normally would to screening and monitoring loans. Fang, Ivashina, and Lerner (2013) focus on private equity investments.  to carry a high degree of risk, and find US bank-affiliated private equity investments in the period 1978-2009 to be highly procyclical, and worse-performing than those of non-bank affiliated private equity deals over the same period.  These investments thus have potential to aggravate any prevailing risk in the system, if bank managers' incentives to grow and maximize utility entail diversifying into private equity investments. j. Increased systemic risk

 In addition to excessive risk for the individual banks several papers show that increased diversification may have an adverse effect on systemic risk.

Wagner (2010) provides a theoretical setting in which diversification of individual institutions deteriorates systemic stability. Diversification, as a "stand-alone" effect, lowers a bank’s probability of failure. However, widespread diversification makes banks more similar to each other by exposing them to the same risks, and thus they are more likely to fail simultaneously following a negative shock to the system. In addition to the theoretical framework, he finds evidence indicating that very large banks became increasingly similar in the years leading up to the crisis, and argues that the current level of diversification likely exceeds the optimal level for social welfare. A similar paper is by De Nicoló and Kwast (2002) who claim that while individual banks have become more diversified the systemic risk potential in the financial sector may have increased. They argue that firm interdependencies provide an indicator of systemic risk potential and they measure interdependencies with correlations of stock returns. The authors find that stock return correlations among large and complex banking organisations in the US over the period 1988-99 have increased, which is consistent with greater potential for economic shocks to become agents of systemic risk in the financial sector  Baele et al. (2007), find similar results for European banks. De Nicolo et al. (2004), in an international sample of banks, find that complexity resulting from conglomeration and consolidation increases systemic risk. Nijksens and Wagner (2010) focus on the effect of trading credit risk transfer instruments (CDSs and CLOs) on banks' risk. They argue that while securitisation may allow banks to purchase protection in the CDS market to shed idiosyncratic exposure, banks also simultaneously buy other credit risk by selling protection in the CDS market. As a result banks may end up being more correlated with each other and this may amplify the risk of systemic crisis. Their analysis of an international sample of banks between 1996 and 2007 shows that the adoption of risk transfer methods by banks increased their riskiness due to higher systemic risk. The authors conclude that credit risk transfer reduce banks’ idiosyncratic risk, but increase systemic risk, by increasing banks’ exposure to risk in the system overall. Brunnermeier et al. (2012) find evidence for US financial institutions over the period 1986-2008 that banks with higher non-interest income make a greater contribution to systemic risk than those practicing "traditional" banking activities. This suggests that activities that are not linked with deposit taking and lending are associated with a larger contribution to systemic risk. Furthermore, after splitting total non-interest income into (i) trading income and (ii) investment banking and venture capital, they find that both components are roughly equally related to systemic risk. FSOC (2011) explains that securitisation has also increased systemic risk even if the creators of ABS did retain a considerable amount of risk on their portfolios. Systemic risk would increase because the regulatory environment allowed creators of ABS to hold less capital than if they had simply held the original assets on their balance sheets. Therefore some activities might contribute disproportionately to systemic risk. Focusing on the use of derivatives, Stulz (2010) argues that while derivatives may provide credit protection however those that provide such protection need to have the ability to repay their obligation. When derivatives shift credit risk from banks to less regulated parts of the financial system then credit derivatives could increase systemic risk. k. Cultural contamination

An additional concern of combining deposit taking and trading activities is the risk that the trading culture contaminates and dominates over commercial bank culture. The transferral of behaviours typical of the trading floor in banking activities into the commercial side causes a lack of confidence in the sector, and is seen as detrimental to the proper, useful functioning of the European economy. For example, Kay (2012) considers that there has been a systematic and deliberate replacement of a culture based on relationships by one based on trading increasingly characterised by anonymity, and the behaviours which arise from that substitution, which has led to a wide erosion of trust in financial intermediaries and in the financial system as a whole. The Commission’s December 2012 8th edition of Consumer Markets Scoreboard shows that consumer trust in the EU banking sector is at an all-time low.

Many proponents of stricter or more banking regulation have noted the negative aspect of cultural impetus and have stressed the importance of re-establishing trust in banks. For example, the European Parliament report on reforming the structure of the EU banking sector stress the important of reforms to change the banking culture.[176] The rapporteur of the report, European Parliamentarian McCarthy, has said that the customer should trust his banker in the same way he trusts his family doctor and that “piecemeal” legislation will not serve the purpose of changing the culture of banking. She considers that structural reform is a necessary addition to the current reform package. "This separation is necessary if we want to change the culture of the retail side of banks, because the investment banking mentality currently pervades retail banking."[177] Also, in the UK, the House of Parliament has established a Parliamentary Commission on Banking Standards (2013) to make recommendations on how to promote a better culture in the banking system.

l. Industry and other studies

In this section we review some studies from the industry examining why banks expand their scope of activity. Some industry studies find significant benefits related to both size and diversification (IIF (2010); The Clearing House (2011, Annex C and D)). Their argument is that larger banks and their scope for achieving greater diversification across business lines and geographies may realise significant synergies, promoting safer, more stable, and ultimately more valuable banks. Continuing this logic, they argue that structural bank regulation initiatives and the separation of business lines would imply costs, not only for banks and their shareholders, but for the economy as a whole. 

The Clearing House (2011), an American banking association, estimates for the US that the 26 largest US banks provide an estimated USD 15-35 billion in direct value to customers every year by providing a wide range of activities. They reach this number by analysing the products and services in which banks provide a unique benefit and quantifying the benefit that each subscribing customer receives.  The authors acknowledge that this process is an imperfect method, but, in particular, estimate that banks with assets of over USD 500 billion are responsible for USD 10 - 20 billion of the total, not counting "indirect benefits to the economy at large." They suggest that the benefits are found in the four product areas of banking (retail, payments and clearing, commercial, and capital markets), though least of all in retail banking.  Within the payments and clearing category of banking activity, securities servicing is seen to be the main source of benefits in product scope that large banks provide, contributing an estimated USD 4-8 billion in related annual benefits.

The Institute of International Finance's (2010) report on systemic risk and systemically important firms finds "considerable real-world evidence" of economies of scale, scope, and diversity among large, international firms with assets in excess of USD 100 billion.  They claim that imposing limits on the activities a bank may perform could severely limit the enjoyment of economies of scope and how they pass these on to the customer, however, without quantifying such a loss, and that there is likely to be "no real gain" for financial stability from such an "arbitrary measure."

Industry estimates of economies of scope and subsequent effects of structural reform also highlight diversification's impact on the risk involved in banks taking part in varied activities. While during the recent financial crisis banks of all types have failed, Barclays Capital (2011) claims that during the period 1988-2009, universal banks defaulted less frequently than retail banks, which in turn defaulted less frequently than investment banks.  Santander's response (2011) to the ICB's consultation, however, notes that a retail bank's assets are less volatile in value than for banks dealing with non-interest-earning activities. Other studies (see for example Van Ewijk and Arnold (2012)) argue that traditional relationship banks were better positioned to deal with the financial crisis than diversified transaction-oriented banks. CEPS (2011) provides a review of banks and an overview of their performance over the crisis for a sample of major European banks.  The authors conclude that retail banks were the best-performing during the period 2006-2009, with less need for state support, and by continuing to fund the real economy, unlike investment and wholesale banks. Since these banks provide a net benefit to the wider economy and in light of these findings, the authors suggest that the authorities' crisis responses should have increased pressure on banks to operate with less complex business structures, therefore justifying the movement back towards the traditional retail banking model. 

Frontier Economics (2013) study the economies of scope for banks engaging into private equity. They argue that that private equity attracts a large range of investors, including banks, and has potential for many advantages for the investors' portfolios. Given the attractiveness of diversifying the portfolio and the possibility of earning greater returns from this diversification, the preservation of private equity activities within the deposit-taking bank may prove advantageous for the bank, as well as the bank's customers. Nevertheless, banks still run serious risks when investing in private equity funds, which have been known to experience substantial losses. For instance, the Finnish Venture Capital Association (2009) states that default rates of private equity-backed companies increases during downturns (which is similar to empirical evidence explained above from Fang et al. (2013)). A systemically important bank bearing the burden of a failing private-equity backed company is extremely dangerous for the company and all the relevant stakeholders, as well as the bank's other customers and the overall financial system.

4. Economies of Scale and Scope and Structural Reform

If economies of scale and scope are present, will banks be in a position to enjoy them post-structural reform? The answer to this question depends on several parameters. Firstly, as discussed in various sections of this Annex, there is evidence that economies of scale are likely exhausted at levels of assets below the thresholds considered by the Commission. Secondly, concerning economies of scope, there is greater evidence for loan portfolio and geographic diversification effects, which are not affected by the structural reform proposal, while for other types of activity diversification results suggest that diseconomies of scope are likely to outweigh any efficiency. Thirdly, the strength of separation requirements has a direct effect on this relationship. At one extreme, accounting separation allows banks to essentially enjoy the same level of economies of scale and scope. At the other extreme, ownership separation would deprive banks of any economies of scale and scope. Functional separation through subsidiarisation would allow some benefits to be maintained depending on the precise requirements.

An advantage of separation through subsidiarisation is that retail banking operations would be protected from investment banking, while some economies of scale and scope advantages that exist within a group could be maintained. These advantages can be preserved, as the overall banking group can stay the same size, and practice the same activities, provided they meet regulatory requirements on a solo basis. Separation through subsidiarisation can be seen as a driver of "net" economies of scope in a banking group. The ability to allow one line of banking business to fail without disturbing overall business functioning and customer relationships could help address the moral hazard created by bank rescues and therefore the subsidiarisation avoids such diseconomies of scope (see ICB (2011) and CEPS (2012)). Transfers of informational knowledge and cost advantages could be enjoyed depending on the precise requirements associated with subsidiarisation requirements, however some duplication of infrastructure would be necessary and a less efficient use of capital would arise (due to regulatory restrictions).

5. Conclusion: Economies of Scale and Scope

Evidence of economies of scale and scope are, at best, mixed. A recurrent problem in the literature is that the presence of implicit subsidies is not typically controlled for, and therefore it is more likely that evidence is found of economies of scale (at relatively larger levels of bank assets) as well as of scope. Furthermore, some studies argue that there are economies of scale or scope but do not comment on the net effect of the economies and diseconomies of scale and scope.

On economies of scale, initial studies have found evidence of such benefits at moderately low levels of assets (USD 100 billion). Given the relatively high thresholds for banks to fall under the requirements of the structural reform proposals, it is likely that the vast majority of these banks will exceed this level of assets. More recent studies that find evidence of higher optimal banking levels either only focus on the positive effects of scale, or do not take into account implicit subsidies (or both).

Concerning economies of scope, the evidence is again mixed. On activity diversification there is some evidence of economies of scope in combining deposit-taking and loan provision, but there is weak evidence that such benefits are significant for other kinds of activity diversification. On the contrary, some activities are likely to lead to conflicts of interest between business lines, increased business complexity, and have an adverse effect on risk-taking and systemic risk. There is a large empirical body of literature suggesting that these diseconomies dominate economies of scope for product diversification. In addition, cultural contamination from investment banking into retail banking has an adverse effect as for example for proprietary trading the bank's aim is to make a profit without providing services to its customers. There is some more positive evidence of economies of scope from geographic diversification. Therefore, while diversification of activity and product lines is mostly associated with significant increases in individual and systemic risk, overall positive effects from diversification are mostly restricted to geographic and loan portfolio diversification.

Given the relatively weak evidence on economies of scale and scope, the business model of large and complex global banking organisations may have been, at least partly, induced by regulatory considerations, rather than inspired by efficiency gains. In this context, obtaining the status of “too big to fail” may have played a role. Therefore, there is lack of evidence on the existence of sufficiently large efficiency benefits that would make such activity restrictions economically suboptimal or even counterproductive.

Bibliography

Amel, D., C. Barnes, F. Panetta, and C. Salleo (2004), “Consolidation and efficiency in the financial sector: a review of the international evidence”, Journal of Banking and Finance, 28, 10, pp. 2493-2519.

Amihud, Y., G. DeLong, and A. Saunders (2002), "The effects of cross-border bank mergers on bank risk and value," Journal of International Money and Finance, 21, 6, pp. 857–877.

Anderson, R.W. and K. Joeveer (2012), "Bankers and bank investors: Reconsidering the economies of scale in banking," FMG Discussion Papers No. 712, Financial Markets Group.

Ang, J.S. and T. Richardson (1994), "The Underwriting Experience of Commercial Bank Affiliates Prior to the Glass-Steagall Act: A Re-examination of Evidence for Passage of the Act," Journal of Banking and Finance, 18, January, pp. 351-95.                  

Baele, L., O. De Jonghe, and R. Vander Vennet (2007), "Does the stock market value bank diversification?" Journal of Banking & Finance, 31, 7, pp. 1999–2023.

Beck, T., A. Demirgüç-Kunt, and R. Levine (2006), "Bank concentration, competition, and crises: First results," Journal of Banking & Finance, 30, 5, pp. 1581-1603.           

Ber, H. Y. Yafeh, and O. Yosha (2001), "Conflict of interest in universal banking: bank lending stock underwriting, and fraud management," Journal of Monetary Economics, 47, 1, pp.189-218.

Berger, A.N. and D.B. Humphrey (1997), "Efficiency of financial institutions: International survey and directions for future research", European Journal of Operational Research, 98, 2, pp. 175-212.    

Berger, A.N. and L. Mester (1997), “Inside the black box: What explains differences in the efficiencies of financial institutions”, Journal of Banking and Finance, 21, 7, pp. 895-947.

Berger, A.N., R.S. Demsetz, and P.E Strahan (1999), "The consolidation of the financial services industry: Causes, consequences, and implications for the future", Journal of Banking and Finance, 23, pp. 135-194.

Berger, A.N., M.A. Espinosa-Vega, W.S. Frame, and N.H. Miller (2005), "Debt Maturity, Risk, and Asymmetric Information," Journal of Finance, 60, 6, pp. 2895-2923.

Bessler, W. and M. Stanzel (2009), "Conflicts of Interest and Research Quality of Affiliated Analysts in the German Universal Banking System: Evidence from IPO Underwriting", European Financial Management, 15, 4, pp. 757-786.

Bhattacharya, S., A. Boot, and A.V. Thakor (1998), "The Economics of Bank Regulation," Journal of Money, Credit and Banking, 30, 4, pp. 745-770.

Boyd, J.H., G.D. Nicolò, and A.M. Jalal (2006). "Bank risk taking and competition revisited: New theory and new evidence." IMF Working Paper No. 297.

Boyd, J. and A. Heitz (2012), “The social costs and benefits of too-big-to-fail banks: a bounding exercise”, working paper, University of Minnesota.

Breedon, T. (2012), “Boosting finance options for finance“, report of an industry-led working group on alternative debt markets.

Brewer, E., W. Jackson, and J. Moser, (2000), "The value of using interest rate derivatives to manage risk at US banking organisations." Economic Perspectives, Federal Reserve Bank of Chicago, pp. 49-66.

Brewer, E. and J. Jagtiani (2011), "How much did banks pay to become too-big-to-fail and to become systematically important?”, Working Papers 11-37, Federal Reserve Bank of Philadelphia.

Brunnermeier M. K., T.M. Eisenbach, and Y. Sannikov (2012), "Macroeconomics with Financial Frictions: A Survey," NBER Working Paper No. 18102.

CEPS (2011), "Business models in European banking: a pre- and post-crisis screening," report prepared by Ayadi, R., D. Llewelyn, W.P. De Groen, Centre for European Policy Studies.

CEPS (2012), "Regulation of European banks and business models: towards a new paradigm," report prepared by Ayadi, R., Arbak, E., De Groen, W.P. and Llewellyn, D., Centre for European Policy studies.

Chairman of the Financial Stability Oversight Council (2011), "Study of the effects of size and complexity of financial institutions on capital market efficiency and economic growth," January.

Davies, R. and B. Tracey (2012), “Too big to be efficient? The impact of too big to fail factors on scale economies for banks”, mimeo.

DeLong, G.  (2001), "Stockholder gains from focusing versus diversifying bank mergers," Journal of Financial Economics, 59, pp. 221-252.

Demirgüc-Kunt, A. and H. Huizinga (2011), “Are banks too big to fail or too big to save? International evidence from equity prices and CDS spreads”, Journal of Banking and       Finance, 37, pp. 875-894.

Deng, S. and E. Elyasiani (2008), "Geographic diversification, bank holding company value, and risk," Journal of Money, Credit and Banking, 40, 6, pp. 1217-1238.

Deng, S., E. Elyasiani, and C. Mao (2007), "Diversification and the cost of debt of bank holding companies", Journal of Banking and Finance, 31,12, pp. 2453-2473.

De Nicolo, G. and M.L. Kwast, (2002), "Systemic Risk and Financial Consolidation: Are They Related?" IMF Working Paper 02/55.

De Nicoló G, P Bartholomew, J Zaman and M Zephirin (2004), “Bank consolidation, Internationalisation and conglomeration: Trends and implications for financial risk,"   Financial Markets, Institutions and Instruments, 13, pp. 173–217.

DeYoung, R and K Roland (2001), “Product mix and earnings volatility at commercial banks: evidence from a degree of leverage model”, Journal of Financial Intermediation, 10, pp. 54–84.

Diamond, D.W. (1984), "Financial Intermediation and Delegated Monitoring", Review of Economic Studies, 51, 3, pp. 393-414.

Diamond, D.W. (1991), “Monitoring and Reputation: The Choice between Bank Loans and Directly Placed Debt”, Journal of Political Economy, 99, 4, pp. 689-721.

Drucker, S. and P. Puri (2005), “On the Benefits of Concurrent Lending and Underwriting”, The Journal of Finance, Vol LX, no.6, pp.2763-2799.

Elsas, R., A. Hackethal, and M. Holzhäuser (2009), "The anatomy of bank diversification", Journal of Banking & Finance, 34, 6, pp. 1274-1287.

Fang, L. H., V. Ivashina, J. Lerner, (2013), "Combining Banking with Private Equity Investing”, NBER Working Paper No. 19300.

Feng, G. and A. Serletis (2009), "Efficiency, Technical Change, and Returns to Scale in Large U.S. Banks: Panel Data Evidence from an Output Distance Function Satisfying Theoretical Regularity", Monash Econometrics and Business Statistics Working Papers 5/09, Monash University, Department of Econometrics and Business Statistics.

Fecht, F., A. Hackethal, and Y. Karabulut (2010), "Is proprietary trading detrimental to retail            investors?”, working paper.

Flannery, M.J., S.H. Kwan, and M. Nimalendran (2010), "The 2007-09 financial crisis and bank opaqueness", Federal Reserve Bank San Francisco Working paper.

Frontier Economics (2013), “Exploring the impact of private equity on economic growth in Europe”, a report prepared for the EVCA, available at:

http://www.evca.eu/publications/frontier_economics_report.pdf

Gambacorta, L. and A. van Rixtel (2013), "Structural bank regulation initiatives: approaches and implications," BIS Working Paper No. 412.

Gatev, E. and P. E. Strahan (2006), "Banks' Advantage In Hedging Liquidity Risk: Theory and Evidence From The Commercial Paper Market," Journal of Finance, 61, 2, pp. 867-892.

Gulamhussen, M. A., C. M. Pinheiro and A. F. Pozzolo (2010), “Do Multinational Banks Create or Destroy Economic Value?”, Available at SSRN: http://ssrn.com/abstract=1572822 or http://dx.doi.org/10.2139/ssrn.1572822.

Gulamhussen, M. A., C.M. Pinheiro, and A.F. Pozzolo (2012), "Internationalization, Diversification and Risk in Multinational Banks", Bancaria No. 11/2012. Available at      SSRN: http://ssrn.com/abstract=2196351.

Haldane, A. (2009), “Credit is Trust”, Speech at the Association of Corporate Treasurers, Leeds, 14 September.                                                                                         

Haldane, A. (2012), “On being the right size”, speech, Bank of England.                                           

Hebb, G.M. and D.R. Fraser (2002), “Conflict of Interest in Commercial Bank Security Underwritings: Canadian Evidence”, Journal of Banking and Finance, 26, pp. 1935-1949.

Herring, R. and J. Carmassi (2010), “The corporate structure of international financial conglomerates: Complexity and its implications for safety and soundness”, chapter 8 of the Oxford Handbook of Banking, Oxford University Press.                                             

Hughes, J., L. J. Mester, C-G Moon (2001), "Are Scale Economies in Banking Elusive or Illusive?”, Departmental Working Papers 200004, Rutgers University, Department of   Economics.

Hughes, J. P., W.W. Lang, L.J. Mester, C-G Moon, and M.S. Pagano (2003), "Do Bankers Sacrifice Value to Build Empires? Managerial Incentives, Industry Consolidation and Financial Performance", Center for Financial Institutions Working Papers 02-18, Wharton School Center for Financial Institutions, University of Pennsylvania.       

Hughes, J. and L. Mester (2011), “Who said large banks don’t experience scale economies? Evidence from a risk-return-driven cost function”, Working Papers 11-27, Federal Reserve Bank of Philadelphia.                                                                                        

Humphrey, D. (2009), “Payment scale economies, competition, and pricing”, ECB Working Paper, No. 1136.                                                                                                               

Iannotta, G. (2006), "Testing for opaqueness in the European banking industry: evidence from bond credit ratings," Journal of Financial Services Research, 30, pp. 287–309.                

ICB (2011), “Final report: Recommendations”, Independent Commission on Banking.              

IIF (2012), "Making resolution robust – completing the legal and operational frameworks for effective cross-border resolution of financial institutions", June, Institute of International Finance.

IMF (2012), “Global Financial Stability Report”, October.                                                     

Johnson, W. C. and J. Marietta-Westberg (2009), “Universal Banking, Asset  Management, and Stock Underwriting”, European Financial Management, 15, 4, pp. 703-732.                                  

Kanatas, G. and J. Qi (1998), “Underwriting by Commercial Banks: incentive Conflicts, Scope Economies, and Project Quality”, Journal of Money, Credit, and Banking, 30, pp. 119-133.

Kashyap, A. K., R. Rajan and J. C. Stein (2002), "Banks as Liquidity Providers: An Explanation for the Coexistence of Lending and Deposit-Taking," Journal of Finance, 57, 1, pp. 33-73.

Kay, J. (2012), “The Kay Review of UK Equity Markets and Long-Term Decision Making", Final Report.

Keys, B., T. Mukherjee, A. Seru, and V. Vig (2010), “Did Securitization Lead to Lax Screening? Evidence from Subprime Loans”, Quarterly Journal of Economics, 125, 1.

Klein, P. and M. Saidenberg (2010), “Organisational structure and the diversification discount: Evidence from commercial banking”, Journal of Industrial Economics, 58, pp. 127–155.

Kroszner, R. S. and R. Rajan (1994), "The Role of Firewalls in Universal Banks: Evidence from Commercial Bank Securities Activities before the Glass-Steagall Act", University of Chicago - George G. Stigler Center for Study of Economy and State 103, Chicago - Center for Study of Economy and State.

Laeven, L. and R. Levine (2007), "Is there a diversification discount in financial conglomerates?”, Journal of Financial Economics, 85, 2, pp. 331-367.                                    

Lelyveld, I. van & K. Knot (2009), “Do financial conglomerates create or destroy value? Evidence for the EU", Journal of Banking and Finance, 33, 12, pp. 2312-2321.

Lepetit, L., E. Nys, P. Rous, and A. Tarazi (2007), “Bank income structure and risk: An empirical analysis of European banks”, Journal of Banking and Finance, doi:10.1016/j.jbankfin.2007.12.002.                                                                           

Liikanen, E. (2012), “Final Report of the high-level expert group on reforming the structure of the EU banking sector”, 2 October, http://ec.europa.eu/internal_market/bank/docs/highlevel_expert_group/report_en.pdf.            

Lumpkin, S. (2011), “Risks in financial group structures”, OECD Journal: Financial Market            Trends, Issue 2.                                                                                                      

Nadauld, T. and S. M. Sherlund (2009), “The Role of the Securitization Process in the Expansion of Subprime Credit”, Available at SSRN:      http://ssrn.com/abstract=1410264 or http://dx.doi.org/10.2139/ssrn.1410264.

Nijskens, R. and W. Wagner (2011), “Credit risk transfer activities and systemic risk: How banks became less risky individually but posed greater risks to the financial system at the same time   ”, Journal of Banking and Finance, 35, pp.1391-1398.

Ötker-Robe, I., A. Narain, A. Ilyina, and J. Surti (2011), “The Too-Important-to-Fail Conundrum: Impossible to Ignore and Difficult to Resolve”, Staff Discussion Paper.                             

Purnanandam, A. K. (2006), “Interest Rate Derivatives at Commercial Banks: An Empirical Investigation”, Journal of Monetary Economics, 54, 6, pp. 1769-1808.       

Rajan, R. G (1992), “Insiders and Outsiders: The Choice between Informed and Arm's-Length Debt", Journal of Finance, 47, 4, pp. 1367-1400.

Rossi, S., M. Schwaiger and G. Winkler (2009), “How loan portfolio diversification affects risk, efficiency and capitalisation: A managerial behaviour model for Austrian banks”, Journal of Banking and Finance, 33, pp. 2218–26.

Saunders, A. (1996), Financial Institutions Management, Second Edition (Burr Ridge, Ill.: Irwin).

Saunders, A. and I. Walter (1994), Universal banking in the United States: What could we gain? What could we lose?, New York: Oxford University Press, 276 pages.

Schmid, M., I. Walter (2009), “Do Financial Conglomerates Create or Destroy Economic Value?”, Journal of Financial Intermediation, 23.

Smith, C. W. and R. M. Stulz (1985), “The Determinants of Firms' Hedging Policies”, Journal of Financial and Quantitative Analysis, 20, 4, pp. 391-405.      

Stiroh, K. (2004a), “Do Community Banks Benefit from Diversification?”, Journal of Financial Services Research, 25, 2, pp. 135-160.

Stiroh, K. (2004b), “Diversification in Banking: Is Noninterest Income the Answer?”, Journal of Money, Credit and Banking, 36, 5, pp. 853-82.

Stiroh, K. And A. Rumble (2006), “The dark side of diversification: The case of US financial holding companies," Journal of Banking & Finance, 30, 8, pp. 2131-2161.

Stulz, R. M. (2010), “Credit Default Swaps and the Credit Crisis”, Journal of Economic Perspectives, 24, 1, pp. 73-92.

The Clearing House (2011), “Understanding the Economics of Large Banks”, http://www.theclearinghouse.org/index.html?f=073048

UK Parliament (2012), “First report of the Parliamentary Commission on Banking Standards”, 21 December.

van Ewijk, S. E., &. I. J. M. Arnold (2012), “The quest for growth: The impact of bank strategy on interest margins," International Review of Financial Analysis, 25, C, pp. 18-27.

Wagner W. (2010), “Diversification at financial institutions and systemic crises”, Journal of Financial Intermediation, 19, 3, pp. 373-386.                                                                             

Wheelock, D. and P. Wilson (2012), “Do large banks have lower costs? New estimates of returns to scale for US banks”, Journal of Money, Credit, and Banking, 44, 1, pp. 171-199.

Xie, L. (2007), “Universal Banking, Conflicts of Interest and Firm Growth”, Journal of Financial Services Research, 32, 3, pp. 177-202.

Annex A10 Quantitative Estimation of a part of the Costs and Benefits of Bank Structural Separation

G. Cannas, J. Cariboni, M. Foryś, H. Joensson, S. Langedijk, M. Marchesi, N. Ndacyayisenga, A. Pagano, M. Petracco-Giudici

Forename(s) Surname(s)

2013

European Commission

Joint Research Centre

Institute for the Protection and Security of the Citizen

Contact information

Sven Langedijk

Address: Joint Research Centre, Via Enrico Fermi 2749, TP 361, 21027 Ispra (VA), Italy

E-mail: sven.langedijk@jrc.ec.europa.eu

Tel.: +39 0332 78 9103

Fax: +39 0332 78 5733

http://ipsc.jrc.ec.europa.eu/

http://www.jrc.ec.europa.eu/

Legal Notice

Neither the European Commission nor any person acting on behalf of the Commission

is responsible for the use which might be made of this publication.

Europe Direct is a service to help you find answers to your questions about the European Union

Freephone number (*): 00 800 6 7 8 9 10 11

(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.

A great deal of additional information on the European Union is available on the Internet.

It can be accessed through the Europa server http://europa.eu/.

JRC88531

© European Union, 2013

Executive summary

This report provides a quantitative contribution on the assessment of costs and benefits in support of the impact assessment of the Banking Structural Reform proposal. The analysis looks at the effects of structural separation on 29 large-sized EU banks which could be affected by the reform according to criteria being proposed for adoption by the Commission.

This report is based on the Systemic Model for Bank Originate Losses (SYMBOL) for the simulation of banking losses and uses balance sheet data for the analysis of allocation of losses and of changes in cost of capital. Policy assumptions and scenarios are provided by DG MARKT.

To estimate the costs and benefits of the structural reform, the balance sheets of the selected sample of banks are separated into a Trading Entity and a Deposit Taking Bank on the basis of publicly available data. Then SYMBOL simulations are run for these separated entities to estimate losses in a financial crisis. Losses are allocated to different private (equity owners and bondholders) and public stakeholders (public finance and Deposit Guarantee Schemes (DGS)/Resolution Funds (RF)) based on available balance sheet data and regulatory scenario assumptions. All simulations assume that Capital Requirements Directive IV (CRD IV) will be implemented and a fully effective bail-in regime will be in place, in line with the Bank Recovery and Resolution Directive Proposal (BRRD). These simulated losses after separation are compared to losses in a baseline without separation for each category of stakeholders.

Separation of banks into a Trading Entity and a Deposit Taking Bank shifts risks and losses towards stakeholders of the Trading Entity as it can no longer rely on capital support from the deposit taking bank to finance its more risky activities. In response, behavioural responses are introduced to the Trading Entities’ balance sheets to reduce their riskiness back to that of the original undivided universal banks. The possible balance sheet adjustments include increased capital, reduced activity and reduced riskiness of assets portfolios. A new set of simulations is performed to assess the overall impact of the reform after behavioural responses.

Aggregate benefits are measured as the reduction in gross and excess losses across all banks due to the behavioural response following structural separation. Aggregate costs are measured as reduced revenues of the TE due to the reduced (risk weighted) assets. Distributional effects are estimates, by measuring benefits (costs) for different stakeholders as the reduction (increase) in expected losses in case of a financial crisis before and after structural separation. These changes in expected losses for the different bank creditors and shareholders are used to estimate changes in risk premia, which are added to the costs of holding additional capital and ‘loss absorbing capacity’ (LAC) to obtain changes in the weighted average cost of capital (WACC) for TEs and DTBs. Possible reduced revenues of the Trading Entities due to behavioural responses are also estimated using an average return on risk weighted assets.

For a financial crisis with losses comparable to the 2008-09 banking crisis, and subject to the caveat that the division of assets and risk-weighted assets between trading and deposit-taking entities and the simulation methodology are uncertain, the simulations show:

· A reduction in gross losses plus recapitalisation needs across all banks in the sample from € 629 bn to a range of €504-583 bn, depending on the behavioural response. The largest loss reductions are the result of reduced activity or risk taking of the banks.

· The amount of losses potentially allocated to bail-in under the BRRD is reduced from € 250 bn to a range of €192-219 bn, mitigating potential financial stability risks and contagion effects.

The impact on banks’ funding costs depends on the extent to which changes in expected losses affect risk premia on equity and bonds:

· If changes to expected losses on capital and bonds are not reflected in lower funding costs, the average WACC across all banks in the sample increases by up to 3 bps as the Trading Entities need to hold more capital and Loss Absorbing Capacity (LAC) which increase their WACC by up to 9 bps. For the Deposit Taking Banks: their probability of default and expected losses decrease; however as they cannot decrease their capital and LAC below the minimum requirements, their average WACC will not change.

· If changes in expected losses are fully reflected in risk premia, the average WACC across all banks in the sample decreases (by 1 to 9 bps) as gross losses reduce. The WACC for the Deposit Taking bank is reduced by 4 to 9 bps as it no longer supports the more risky Trading Entity activities. The increase in WACC for the Trading Entities is limited due to the reduced riskiness for bail-inable creditors (following behavioural responses). The behavioural responses however also reduce the return on equity.

The analysis does not find a direct beneficial effect of structural separation on losses to public finances in case of a financial crisis. This is the result of the assumption that both CRDIV and BRRD are considered as fully effective before the introduction of separation, thus leaving limited scope for further reductions in pass-through to the safety net. This assumption implies a conservative estimate of net public finance benefits of the structural reform.

Broader and indirect costs and benefits - through e.g. market liquidity, lending and investment, asset prices and macroeconomic competition effects - may be substantial as the banks concerned play an important role in providing credit to the economy and market liquidity. These effects are not quantified within this report.

Table of Contents

Executive summary.. 205

1..... Introduction and objectives. 208

2..... The dataset. 210

3..... Methodology.. 211

3.1. Allocation of assets, RWA and capital 211

3.2. Simulation methodology. 211

3.3. Calibration of market behavioural responses. 213

3.4. Costs and benefits. 214

4..... Balance sheet allocation upon separation.. 216

4.1.  Allocation of assets. 216

4.2.  Allocation of Risk Weighted Assets and Capital 218

5..... Results: costs and benefits. 221

5.1.  Simulations of separation before behavioural responses. 221

5.2.  Behavioural responses. 222

5.3.  Simulations of separation after behavioural response. 224

5.4.  Impact on Funding costs. 226

6..... Conclusions. 234

Annexes

APPENDIX A - List of banks. 236

APPENDIX B - Sample statistics before and after structural separation   237

APPENDIX C - The SYMBOL model.. 239

APPEDIX D - The Implied Obligor PD for the Trading Entities. 240

APPENDIX E - SYMBOL simulated losses and historical losses during the recent crisis. 242

APPENDIX F - Assets of large banking groups within the EU-27. 244

1. Introduction and objectives

This report provides a contribution on quantification of costs and benefits to the impact assessment of bank structural separation. The analysis is based on the Systemic Model for Bank Originate Losses (SYMBOL)[178] for the simulation of banking losses and on balance sheet data for the analysis of allocation of losses and of changes in cost of capital.

Benefits of the reform are calculated based on overall differences in simulated losses in a financial crisis under scenarios with and without structural separation. Scenarios with separation include the effect of behavioural responses to the reform. Changes in losses are estimated for different public (public finances, safety nets) and private (equity owners, bail-inable bondholders) stakeholders. Costs are obtained by calculating the impact on the Weighted Average Cost of Capital (WACC) due to variations in the balance sheets following behavioural responses to structural separation and the ensuing reduced revenues of the Trading Entities. Variations in the funding costs of equity and bail-inable liabilities following variations in expected losses after the introduction of the reform can also be considered.

Results are subject to the caveat that the division of assets and risk weighted assets between trading and deposit-taking entities, the simulation methodology and the modelled behavioural responses are susceptible to a degree of uncertainty.[179]

The analysis is based on the sample of 29 banks which could be affected by the reform according to criteria being proposed by the Commission.[180]

Policy assumptions and scenarios to be analyzed are provided by DG MARKT, in particular, this costs-benefit analysis is performed under the assumption that bail-in of liabilities in resolution is fully used and effective before the structural reform, as foreseen in the BRRD.[181] This implies a conservative estimate of net social benefits of the structural reform.[182]

The analysis does not take into account other social and private costs and benefits that might be incurred due to separation such as loss of economies of scope and scale, legal costs, relocation costs, effects on asset pricing and knock-on effects due to reduced market liquidity, reduced conflicts of interest within banks, reduced misallocation of resources, facilitated supervision.[183]

The remainder of the report is structured as follows: in Section 2 the dataset. Section 3 outlines the methodology. Section 4 details how the balance sheet is split between the TE and the DTB upon separation and what adjustments are applied to the data to reflect the impact of the introduction of CRD IV (Basel 3). Section 5 presents the results of the simulations and their translation into costs and benefits. The last section concludes.

Several appendices are enclosed to present detailed figures and technicalities. In Appendices A and B the sample is described. Appendix C describes the SYMBOL model and Appendix D the methodology and adjustments applied for applying SYMBOL to the trading entity. A comparison of historical losses with simulated losses is provided in Appendix E. The estimated share of assets of large banking groups within the EU-27 are summarized in Appendix F.

2. The dataset

The sample of 29 EU banks used in the simulation exercise (see appendix A and B for a full list and descriptive statistics) is identified based on the methodology presented in the JRC report “Trading activities and functional structural separation: possible definitions and calibration of de minimis exemption rule”. In particular the focus is on definition 3 of trading activities introduced in that report.[184] Small-sized bank (assets below 30 bn€) are excluded[185].

It should be noted that possibly not all candidate banks could be subject to separation, and that actual banks selected for separation will depend on the criteria eventually adopted in the legislation. In the absence of further details this seems however a reasonable selection on which to base the exercise.

Total assets of the sample amounts to 22.653 bn€ by end 2011, i.e. roughly 56% of European banks assets (Appendix B). All the banks in the sample are included in the European Banking Authority (EBA) capital exercise sample, excluding one medium-sized bank (Mediobanca).

The source of data for SYMBOL simulations is Bankscope, a proprietary database of banks' financial statements produced by Bureau van Dijk. The inputs needed for SYMBOL simulations are Risk Weighted Assets (RWA), regulatory capital and total assets. SNL Financial data (see Appendix B for descriptive statistics) are used to calculate the split of the balance sheet for each bank, as detailed in Section 3.[186] All data are consolidated as of 2011.

3. Methodology

The SYMBOL model is used to simulate the losses of the sample of these 29 banks in case of a financial crisis under regulatory scenarios with and without bank structural separation and balance sheet data are used for the analysis of allocation of losses and of changes in cost of capital and revenues. All scenarios assume that CRD IV will be implemented and that an effective bail-in regime as contained in the Bank Recovery and Resolution Directive Proposal (BRRD) will be in place.

3.1. Allocation of assets, RWA and capital

In order to simulate the effects of structural reform, each bank in the sample is separated into a Trading Entity (TE) and a Deposit Taking Bank (DTB) using publicly available balance sheet data. The allocation of assets is based on definition 3 of trading activities presented in the JRC report “Trading activities and functional structural separation: possible definitions and calibration of de minimis exemption rule”, which includes securities (excluding loans) and derivatives held for trading.[187]

RWA and capital are allocated between the DTB and the TE following the methodology described in the Commission Services report “Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements”, which is based on obtaining average risk weights for each category of assets using a panel regression, predicting risk weighted assets for each separated entity based on this estimated weights and the allocation from the previous step and obtaining minimum capital requirements based on this predicted risk weighted assets and capital adequacy ratios.[188]

Adjustments are introduced to take into account the impact of the introduction of Basel 3 (CRDIV) on RWA, regulatory capital and minimum capital requirements. These are implemented using the average EU results of the 2011 Quantitative Impact Study (QIS) by the EBA. These adjustments imply increased RWA, a more strict definition of regulatory capital, and the introduction of the Capital Conservation Buffer. Regulatory capital is then topped up to 10.5% of RWA to meet the minimum capital requirements of Basel 3.

In this step, the implicit assumption that separation along these lines will effectively be possible is implicitly maintained.[189]

3.2. Simulation methodology

SYMBOL (SYstemic Model of Banking Originated Losses) is a micro-simulation model[190] which makes use of individual banks’ balance sheet and regulatory capital data to simulate banks’ losses due to the failure of its obligors and to derive the (aggregated) distribution of losses originated in the banking system. The main idea behind this model is that it is possible to estimate and average Probability of Default (PD) of the portfolio of obligors of a bank (the so called implied obligors’ PD) by inverting the Basel FIRB (Foundation Internal Ratings Based) formula for capital requirements.[191] The base SYMBOL methodology is used both for the DTB and for the TE, after taking into consideration the difference in confidence for calibration purposes (i.e. inverting the FIRB at 99% and not at 99.9%).[192],[193]

The distributions of systemic losses (both gross and in excess of banks’ regulatory capital[194]) are then simulated using the SYMBOL model. All simulations are run without contagion effects between the banks in the sample.[195] Adjustments to the final distribution of losses are also made to account for banks assets outside the EU based on detailed unconsolidated balance sheet data, such that losses are reduced proportionally for each banks’ estimated share of non-EU subsidiaries.[196] Losses are allocated to equity, bail-inable liabilities, safety-net tools (DGS, RF) and public finances based on available balance sheet data. 

To determine financial crisis losses in the baseline no-reform scenario, Gross Losses are separately simulated for the TE and the DTB and are then summed to obtain total gross losses of the undivided bank.[197] Total regulatory capital of the undivided bank is used to absorb combined losses of the DTB and TE activities in the baseline scenario; the total bail-in capacity[198] is applied to the undivided bank’s Excess Losses after depletion of the capital. This allows calculating the changes in simulated losses affecting the different stakeholders with respect to a baseline where the same two activities are conducted within the same entity.

3.3. Calibration of market behavioural responses

To quantify the effects of separation, a two-step approach is taken. In the first step, simulations are used to estimate Gross Losses, Losses in Excess of Capital and Recapitalization needs[199] for the undivided universal banks as well as separated TEs and DTBs in a systemic crisis of the same magnitude as the recent one.[200] The sum of the two distributions of gross losses against the total capital of the two entities represents the baseline situation where the two entities are joined in a single universal bank. Comparison with the results obtained by considering only capital in each separated entity against its own losses give the effect of effects of separation before behavioural response of the TE. This allows estimating the shift in “simulated losses in case of a crisis” across stakeholders as the equity and the bail-in capacity (LAC) can no longer be pooled. Under this approach, the separation does not directly change the overall gross losses in the system but leads to changes in the allocation of losses across stakeholders. Five groups of private stakeholders are considered (DTB shareholders, TE shareholders, DTB bail-inable creditors, TE bail-inable creditors), as well as two safety nets (DGS, RF and public finances)).

These shifts in risks affect risk premia and bank funding costs by changing expected losses of the different stakeholders as activities of the TE are more risky than those of the undivided bank and as capital and bail-in capacity can no longer be pooled with the DTB.. As a result, the separated banks will have incentives to change their balance sheets. These behavioural responses are calibrated such that the riskiness of the TE matches that of the original universal banks before separation.[201] Note that DTBs cannot adjust their balance sheets as they are constraint by minimum regulatory requirements on the MCR (>10.5% RWA) and the LAC (>8% TA). Three alternative ways by which the TE can reduce the riskiness to its shareholders and creditors are considered in the analysis:

- Increasing regulatory capital, maintaining the risk weights and total assets at the same level.

- Reducing riskiness while keeping total assets and the regulatory capital constant. In the SYMBOL model this is simulated through a reduction in risk weights (RWA).

- Reducing riskiness through a reduction in total assets, while keeping the risk weight per unit constant and the regulatory capital the same (i.e. reducing leverage).

In the second step, losses are re-simulated taking account of these behavioural responses affecting the balance sheet of the TEs and the effects of separation including behavioural responses can be calculated. Note that behavioural responses are not only redistributing losses but also cause a change in the total amount of losses that need to be absorbed by the system.

In this part, an implicit assumption is made that capital and debt necessary will be available from the markets to recapitalize the separated entities after separation in case they show a shortfall. Assumptions preventing substitution of capital for bail-in capacity are also introduced, in order to keep risk for creditors of the trading entity in line with those of creditors of the universal bank.[202]

3.4. Costs and benefits

Costs and benefits of the reform are then calculated based on changes (with respect to the baseline) in estimated losses for different stakeholders in case of a financial crisis and on changes in the Weighted Average Cost of Capital (WACC).[203] Possible reduced revenues of the Trading Entities due to behavioural responses are also estimated using an average return on risk-weighted assets, and the reduction in risk-weighted assets due to behavioural responses.[204] The related changes at economy-wide level such as pass-through of variations of DTB’s WACC to the cost of capital for borrowing firms and reduced value added in the economy (e.g. market liquidity) are not explicitly calculated.

Distributional effects are estimated by measuring changes in estimated losses for different stakeholders in case of a financial crisis before and after structural separation.

For the changes in the WACC two scenarios are developed, thus providing a range for the likely WACC effect. One in which risk premia on equity and bail-inable bonds do not reflect changes in expected losses. In this case, changes in WACC are estimated exclusively based on changes in the capital structure of entities following separation. In the second scenario, risk premia are allowed to vary reflecting changes in expected losses in a financial crisis after separation. In this part, implicit assumptions are made that changes in default risk and expected loss evaluated in the case of a financial crisis can be used to estimate changes in risk premia, and that changes in simulated default risk and expected loss (i.e. excluding other factors which are not included in the simulation model employed) are a sufficient proxy to estimate changes in risk premia. Moreover, as we cannot observe and do not have a reliable estimate of risk premia after the introduction of the recovery and resolution framework (which are already partly reflected in market prices of subordinated and senior debt), available data on current risk premia will be used as baseline which may imply a small underestimation of the WACC impact.

4. Balance sheet allocation upon separation

Each bank in the sample is split into a DTB and TE. First, the total assets of the banks are split on the basis of data from SNL (in particular its classification of financial assets and liabilities). In addition, RWA, total regulatory capital and the minimum capital requirement are split between the two entities as described below.

4.1. Allocation of assets

In line with the definitions of trading activities introduced in the abovementioned JRC report, assets allocated to TEs are: Securities Held for Trading (STA) (excluding loans and derivatives) and Derivative Assets Held for Trading (DTA). This split of the balance sheet does not depend on the particular type of trading activity effectively undertaken by each bank.[205]

The DTB is instead allocated all loans to banks, loans to customers, all other assets held at amortized cost, securities held to maturity, available. for sale securities and securities held at fair value (see Table 1).

Remaining total assets (including other assets such as cash) are allocated between the TE and DTB, proportionally to the already allocated SNL financial assets.

Derivatives held for hedging purposes are allocated proportionally to the allocation of RWA based on explicitly allocated classes (see next section).

Table 1 - SNL Financial Assets considered for the allocation following separation.

SNL Balance sheet Items || TE || DTB

Net Loans to Banks (NLB) || || X

Net Loans to Customers (NLC) || || X

Securities Held at Amortised Cost (SAC) || || X

Securities Held to Maturity (STM) || || X

Securities Held for Trading (STA) (excluding loans and derivatives held for trading) || X ||

Derivative Assets Held for Trading (DTA) || X ||

Securities Held at Fair Value (SFV) (excluding loans held at fair value) || || X

Available for Sale Securities (SAFS) (excluding loans available for sale) || || X

All other assets not explicitly allocated (excluding derivatives held for hedging purposes) || Divided proportionally to explicitly allocated financial assets

Average Share of Financial Assets || 31% || 69%

Note: Derivatives Assets Held for Hedging Purposes are split proportionally to Risk Weighted Assets calculated based on other classes (see section 4.2)

Figures 1 and 2 show the average TE and DTB assets as a share of total assets following the banks split of assets.

The table of Appendix B shows the result of the allocation of total assets for each individual bank in the sample.

Figure 1 - Average TE assets as a share of total assets following structural separation

Figure 2 - Average DTB assets as a share of its total assets following structural separation

Source: SNL database and JRC estimates

4.2. Allocation of Risk Weighted Assets and Capital

RWA and capital are allocated between the DTB and the TE following the methodology described in the Commission Services report “Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements”.

A fixed effects regression model has been estimated allowing assigning risk weights to the different banking activities, as shown in Table 2.

Table 2 - Estimated risk weights across banking activities (EU sample)

|| NLB || NLC || SAC || STM || SAFS[206] || SFV || STA || DTA+DTL[207] || DHA+DHL

RWA coefficient || .18 || .52 || .31 || 0[208] || .39 || .23 || .099 || .024 || -1.95

Note: NLB = Net Loans to Banks; NLC = Net Loans to Customers; SAC = Securities Held at Amortised Costs; STM = Securities Held to Maturity; SAFS = Available For Sale Securities (excluding loans); SFV = Securities held at Fair Value; STA = Securities (excluding loans) Held For Trading as Assets; DTA= Derivatives Held For Trading as Assets; DTL = Derivatives Held For Trading as Liabilities

These estimated risk weights for the different balance sheet components are used to estimate how the RWA of the original undivided bank should be split between the TEs and the DTBs.[209] The total regulatory capital of the two entities under Basel 2 is then apportioned using the obtained split of RWA.

The split Basel 2 RWAs are adjusted to take into account future changes introduced by Basel 3 to RWA definitions and requirements. Average EU results of the 2011 Quantitative Impact Study (QIS)[210] are employed for the adjustments, as detailed in the table below. The changes are allocated between the DTB and the TE as described below.

Table 3 - Average EU (weighted on total assets) corrections for RWA and regulatory capital from EBA as of 30/06/2011

|| G1 Banks || G2 Banks

Relative Change in RWA for the whole bank (%) || 21.20 || 6.90

Relative Change in Regulatory capital for the whole bank (%) || -34.35 || -7.76

Source: EBA

Note: In this exercise G1 - Tier 1 Capital > 3 bn€, G2 - Tier 1 Capital <3 bn€

The Basel 3 increase in the RWA is allocated based on a breakdown of the changes in RWA published by EBA[211], reported here in Table 4. In particular, the Table shows the part of the total percentage increase in RWA due to:

a) the change in RWA following changes in the ‘definition of capital’[212], which is split proportionally to the share of total assets allocated to the TE and the DTB.

b) counterparty credit risk, which is allocated to TE for the share due to Credit Valuation adjustment (CVA) and to the DTB for the part due to the higher asset correlation parameter included in the IRB formula.

c) securitization in the banking book, which is fully allocated to the DTB.

d) to market risk (including securitisation in the trading book) is fully allocated to the TE.

Table 4 - EBA split of the increase in RWA due to Basel 3 (average %-increase)

Type || Total relative increase in RWA || Part due to definition of capital || CCR banking book || CCR trading book || Securitisation banking book || Trading book

G1 || 21.2 || 7.9 || 1.2 || 6.9 || 1.0 || 4.2

G2 || 6.9 || 3.4 || 2.9 || 0.2 || 0.4

Source: EBA

Results are presented in Table 5 both for Basel 2 and for Basel 3.

Table 5 - Allocation of total RWA between the TEs and the DTBs under Basel 2 and Basel 3

|| Basel 2 || Basel 3

DTB || 91% || 79%

TE || 9% || 21%

The new Basel 3 definition of the quality of capital affect both entities, thus the decrease in the regulatory capital is split proportionally to the capital allocated to the two entities. For banks with adjusted regulatory capital below 10.5% of RWA, the capital is topped up to meet the Basel 3 minimum required capital including the capital conservation buffer (10.5% RWA).

Appendix B shows the result of the allocation of RWA for each individual bank in the sample.

5. Results: costs and benefits 5.1. Simulations of separation before behavioural responses

This section presents the SYMBOL simulation results. As the very severe crisis (with excess losses at the 99.95th percentile) shows most similarity to real state aid figures during the last crisis in terms of banks' losses, all figures and calculations in the tables reflect simulations of a crisis of that severity (see Appendix D).

Table 6 shows gross losses (in row 1) and excess losses (i.e. losses in excess of capital) plus recapitalisation needs[213] in row 3, before (column A) and after (column B to D) bank structural separation, before taking into account market behavioural responses.

Table 6 – Effects of separation on distribution of losses for bank sample before market behavioural responses – bn€

Losses || Undivided Bank || Structural Separation before Market Behavioural Responses*

Sum TE+DTB || TE || DTB

|| A || B=C+D || C || D

1 - Gross Losses and recapitalisation needs || 629 || 625 || 242 || 383

2  - Losses on equity || 380 || 339 || 108 || 231

3 – Excess Losses and recapitalisation needs encumbering bail-in and resolution || 250 || 286 || 134 || 152

4 - Bail-in losses (LAC = 8% TA) || 201 || 196 || 94 || 102

The total amount of bank gross losses plus recapitalisation needs are simulated at EUR 629 bn before separation. They are virtually the same after separation EUR 625 bn. These need to be borne by equity, bail-in, safety nets and the public sector. Row 2 and 4 show that losses on equity and – to a lesser extent – losses bail-in losses are lower after separation than for the undivided bank. The simulation results of Table 6 are used as input for Table 7 that shows in more detail the effect on the different bank creditors of the separation.

5.2. Behavioural responses

As expected, gross losses (plus recapitalisation needs) over total assets are higher for the TE than for the undivided bank and for the DTB they are lower (Table 7). As losses would increase by about 20% (from 2.8% to 3.4% as a share of total liabilities plus equity) for the TEs with respect to the original undivided banks in the absence of any response, it is foreseeable that TEs will be required by the markets to change the structure of their balance sheets in response to separation. The effect of separation is particularly significant on equity, as – compared to the undivided bank- losses on equity in case of a very severe financial crisis would increase by more than 1/3 for the TE, while they would reduce by almost ¼ for the DTB. The effect of separation on bail-in losses is smaller. 

Table 7 – Losses for the bank sample for the undivided bank and after separation before market behavioural responses

|| Gross losses and recapitalization needs as a share of total liabilities plus total equity || Losses on equity as a share of total equity || Bail-in losses as a share of LAC minus equity

1 – Undivided Bank || 2.8% || 40.4% || 23.0%

2 - DTB || 2.5% || 31.1% || 20.5%

3 – TE || 3.4% || 55.2% || 24.6%

Note: The figures in this table reflect averages across all banks. They should not be compared to the calibration of the required LAC in the BRRD of 8% TA .

As detailed in section 4.2, the TE’s adjusted balance sheets are calibrated so as to pose the same probability of default to creditors as the initial undivided bank posed. It should be noted that also LAC is subject to a behavioural response. LAC is calibrated such that it is at least 8% TA for all banks and LAC on top of equity as a share of TA is at least as high as in the undivided universal bank. The latter condition is to avoid that behavioural responses increasing capital/TA would only substitute bail-in capacity and that capital/TA would be higher than LAC/TA in a number of cases.

Table 8 presents the key data of the original and adjusted balance sheets for the banks.

Table 8 – Key balance sheet data for the undivided bank, the DTB and TE before and after market behavioural responses – bn€

|| TA || Regulatory capital (B3) || LAC on top of equity || RWA

1 – Undivided Bank || 22,653 || 939 || 874 || 8,945

2 - DTB || 15,463 || 744 || 495 || 7,089

3 – TE || 7,191 || 195 || 381 || 1,856

4 – TE with increased capital || 7,191 || 322 || 381 || 1,856

5 – TE with reduced RWA || 7,191 || 195 || 392 || 1,134

6- TE with reduced TA || 4,152 || 195 || 179 || 1,134

Note: The behavioural responses of the TE are calibrated such that the probability of default of the TE matches that of the undivided universal bank after the behavioural response.

The actual bank response can be a combination of these illustrative behavioural responses. It would depend inter alia on bank specific characteristics and market specific risk appetite and regulation.

5.3. Simulations of separation after behavioural response

Based on the adjusted balance sheets following the behavioural response, SYMBOL simulations are run to capture the combined effects of the separation on the different stakeholders and the system as a whole.

Table 9 – Losses for bank sample for the undivided banks and after structural separation after market behavioural responses – bn€

Losses || Undivided Bank || Structural Separation after Market Behavioural Responses

Sum TE+DTB || TE || DTB

|| A || E=F+G || F || G

1 - Gross Losses and recapitalisation needs || 629 ||  504-583 ||  121-200 || 383

2 - % on Equity || 60% ||  61%-63% ||  61%-67% || 60%

3 – Excess Losses and recapitalisation needs encumbering bail-in and resolution || 250 ||  192-219 || 39-67 || 152

Table 9 shows gross losses (in row 1) and excess losses (i.e. losses in excess of capital - in row 3) plus recapitalisation needs[214] before (column A) and after (column E to G) bank structural separation taking into account market behavioural responses. (They can be compared to columns B to D of Table 6 showing the effects before market behavioural responses).

Benefits from structural separation can be seen along the following metrics:

- the reduction in the sum of gross losses and recapitalisation needs following structural separation, compared to the situation prior to structural separation. Gross losses can in particular decrease from 629 bn € in column A to as low as 504 bn€ as shown in in column E in row 1. The lower gross losses reflect the reduction in gross losses of the TE following market behavioural responses and will vary according to the type of market behavioural response that will prevail. For the behavioural responses with lower risk (i.e. RWA), the lowest gross losses are generated.[215] When considering net benefits, the loss in bank revenues on these activities and the related reduced value added in the economy (e.g. market liquidity, provision of credit and capital, etc) needs also to be taken into account (see costs section below);

- the reduction in the sum of excess losses plus recapitalisation needs before application of bail-in compared to the situation prior to structural separation. Excess losses plus recapitalisation needs before bail-in can in decrease from 250 bn € in column A to 192 bn € as shown in in column E in row 3. This means that the banking system – and in particular the trading activity undertaken by the part of the banking system subject to structural separation – can be expected to become less risky for the rest of the financial system as potential financial stability risks and contagion due to bail-in are reduced. This is due, in particular, both to the reduction in generated risks (i.e. gross losses), and to the fact that a higher share of these losses are absorbed by the shareholders of the TE, as shown in row 2. This means a better alignment of the incentives driving TEs, i.e. a reduction of potential moral hazard by TEs.[216]

Focusing on the resolution process, Table 10 presents how excess losses plus recapitalisation needs before application of bail-in are split between bail-inable creditors and the safety net tools/public finances. Bail-in is assumed to take place up to a Loss Absorbing Capacity (LAC) equal to 8% of Total Assets (in line with the June 2013 Council position[217] on the BRRD). The LAC considered for bail-in can nonetheless be higher than 8% of Total Assets in case of market behavioural responses: in behavioural response 1, where TEs increase their capital, the LAC is increased by the corresponding amount so that there is no substitution between capital and bail-in-able liabilities and bail-in capacity in excess of capital is unchanged. In line with response 1, in behavioural response 2 and 3, residual LAC (i.e. bail-in capacity in excess of capital) is assumed at least as high as a share of total assets as in the original undivided bank. This ensures that a higher capital ratio to total assets in the behavioural response does not lead to a perverse automatic reduction in the bail-in capacity.[218]

Table 10 – Allocation of excess losses plus recapitalisation needs after market behavioural responses– bn€

|| Undivided Bank || Structural Separation after Market Behavioural Responses* ||

|| Losses || || Sum TE+DTB || TE || DTB

|| A || E=F+G || F || G ||

1 - Excess Losses and recapitalisation needs before application of bail-in || 250 ||  192-219 ||  39-67 || 152 ||

2 - Bail-in (LAC = 8% TA or higher as per behavioural response) || 201 ||  129-149 ||  27-47 || 102 ||

3 - Excess Losses and recapitalisation needs after application of bail-in (i.e. contingent liabilities on public finances and the safety net tools - DGS and Resolution Fund - ) || 49 ||  62-70 || 12-19 || 50 ||

The amount of losses not absorbed by bail-in and potentially affecting the safety net and/or public finances increases. On the one hand, the loss of the ability to pool capital and LAC across the entities implies that they can absorb a lower share of losses. On the other hand, gross losses and excess losses in the system are reduced due to the behavioural responses of the trading banks. While the overall safety net and public finance losses show some increase, the safety net can focus more on the risks stemming from the DTB, while the risks stemming from trading activities are mostly absorbed by the LAC through the application of bail-in.

5.4. Impact on Funding costs

Impacts on the costs of capital are evaluated by looking at variations in Weighted Average Costs of Capital (WACC) due to changes in the structure of the balance sheets of banks after behavioural responses and by (a) considering a situation where risk premia do not vary in response to changes in the expected losses in financial crises and (b) one where they do vary in response to changes in risk.[219]

In all, as far as costs are concerned, the changes in banks' funding costs post structural separation are in essence due to the following effects.

(i) Increases in the capital share of total funding by the TE due to behavioural response.

(ii) Increases in LAC over the 8% TA minimum by the TE due to behavioural response.

(iii) Changes in the risk premium on equity as markets perceive changes in riskiness and expected losses.

(iv) Changes in the risk premium on bail-inable bonds as markets perceive change in expected losses.

(v) Reduced revenues and return on equity due to reduction in TE (risk-weighted) assets.[220]

For case (a), where risk premia do not vary, we measure changes in the WACC for the TE and the DTB based on effects (i) and (ii) by multiplying the increases on capital and LAC by the currently observed premiums.

For case (b), where risk premia vary in response to changes in expected losses in case of crisis, in addition to (i) and (ii), we also estimate effects (iii) and (iv) by adjusting risk premiums proportionally to the variation in expected losses.[221]

Effect (v) is calculated based on currently observed returns on RWA and is relevant for the behavioural responses with reduced RWA.

Table 11 shows effects (i) to (iv) and the total change in the WACC for the DTB (Row 1), the TE under different behavioural responses (Row 2-4) and the range spanned by the weighted averages[222] of the impacts for the two entities across scenarios (Row 5). The first 4 columns of Table 11 (A to D) are based on calculations that are presented and explained in Tables 12 to 15 below.

Column A and B of Table 11 show, respectively, the change in WACC due to the increases in capital and LAC. Tables 12 and 13 provide the underlying calculations. Columns C and D provide the estimated changes in risk premium due to changes in expected losses on equity and first bail-inable bonds (BiB). The calculations for these are presented in Tables 14 and 15. Column E provides the total estimated impact on WACC in case the changes in expected losses on BiB and equity compared to the undivided bank do not affect risk premia. Column F provides the total effects with changes in risk premia on equity and debt due to changes in expected losses.  

Results are presented as ranges of point estimates across combinations of behavioural scenarios and cases (i.e. including variations in risk premia or not when calculating WACC). While sensitivity analysis to different starting levels of risk premiums is not included, it should be noted that relative changes would not be much affected by variations in initial risk premia, as these would imply a change both in the absolute variation and in the baseline level.

Without changes in risk premia, the average WACC in the system increases as the TEs need to hold more capital and LAC. The increase in WACC for the TEs is estimated at 9 bps in case of behavioural responses with increased capital and reduced total assets and 0 bps for the behavioural response in which the TE reduces RWA but leaves its liability structure unchanged (column E). The DTB does not need to hold capital or LAC above minimum requirements as its probability of default and expected losses decrease. However, as the DTB cannot decrease its capital and LAC below the minimum requirements, the weighted average WACC across all banks in the sample increases by 2 to 4 bps.

With changes in risk premia, the average WACC across the sample banks decreases (by 1 to 9 bps) as the gross losses go down (column F). The WACC of the DTB is reduced by 4 to 9 basis points as it no longer supports the more risky TE. The reduced expected losses on capital and LAC reduce the risk premia on equity and subordinated debt. For the TEs, the reduced riskiness for bail-inable bonds as a result of the behavioural responses in particular limit the increase in the WACC. In the behavioural response with reduced RWA, the overall WACC for TEs decreases by 8 to 9 bps. This effect is smaller for the behavioural response with reduced total assets, as the bail-in capacity (LAC-capital) is significantly reduced with the reduction of total assets.[223] Estimated reductions for the part related to bail-inable bonds in TEs are due to lower gross losses, the shifting of losses towards capital and reduced pooling (see also Table 15 and its discussion).

The reduction in the WACC of up to 9 bps for the TE in case of a behavioural response with reduced risk weighted assets should be considered together with the accompanying loss of revenues per unit of assets due to the lower RWA. In addition to the change in funding costs, this loss of revenues per unit of assets represents a further ‘cost’. While the reduced RWA concerns both behavioural responses in rows 3 and 4, the effect is particularly important for the behavioural response in row 2 with reduced risk weighted assets as the quantity of total assets and thus the overall funding quantity is unchanged. Lost revenue due to lower RWA should thus be counted in full,[224] whereas for the behavioural response in row 3, RWA, total assets and thus funding needs go down proportionally such that revenues per unit of assets should not change.

The costs estimates are in line with ex-ante expectations and the objectives of the structural separation.

Table 11 -  Estimated change in WACC in different behavioural responses due to increased capital (A) and LAC (B), changed risk premium due to changed expected losses on BIB (C) and changed risk premium on capital due to changed expected losses (D) (in bps). 

|| Change in WACC due to : || ||

|| increase in capital (Table 12) || increase in LAC above 8% TA (Table 13) || change in risk premium due to change in expected losses on equity (Table 14) || change in risk premium due to changed expected losses on BIB (Table 15) || Total  without changes in risk premia || Total with changes in market risk premia || ||

|| A || B || C || D || E=A+B || F=A+B+C+D || ||

1 - DTB || 0 || 0 || -8 to -3 || -1 || 0 || -9 to -4 || ||

2 – TE with increased capital || 7 || 2 || <1 || -7 || 9 || 2 to 3 || ||

||

3 – TE with reduced RWA* || 0 || 0 || <1 || -9 || <1* || -9* to -8 || ||

4- TE with reduced TA* || 8 || 1 || <1 || -4 || 9 || 5 to 6 || ||

5 – Range (Weighted Averages on TA) || 0 to 2 || <1 || -6 to -2 || -4 to -2 || 0* to 3 || -9* to -1 || ||

Notes:(*) The reduction in the WACC of up to 9 bps for the TE in case of a behavioural response with reduced total assets or risk weighted assets should be considered together with the accompanying loss of revenues due to the lower risk weighted assets. See the discussion on previous page and in footnote 47.

- The risk premium on senior unsecured debt is assumed at 174 bps; on subordinate debt it is at 307 bps, and the equity risk premium is 700 bps. Therefore, costs of increasing capital while keeping LAC unchanged are 393 bps per unit (difference of risk premium on subordinate debt versus equity risk premium). Costs of increasing LAC are 133 bps per unit (difference of the risk premium on senior debt and subordinate debt).[225] It should be noted that while absolute results would exhibit sensitivity to the particular levels of risk premiums chosen to base the analysis, relative results would not exhibit much variation.

Table 12 shows the estimated change in the weighted average costs of capital (WACC) due to change in the share of capital over total assets following the different behavioural responses. Column C shows capital/TA for the different scenarios. The change in capital/TA due to the behavioural response is given in Column D. Column E shows the costs of the increase in the capital share in total funding on the WACC. The estimated cost of capital is 393 bps reflecting the difference between the risk premium on equity (700 bps) and the risk premium on bail-inable bonds (307) which it replaces.[226]

The effect on WACC for the TE is estimated to be in the range of 0 to 8 bps depending on the behavioural response, while the DTB does not need to hold any capital in excess of the regulatory minimum.

Table 12 - Estimated change in WACC due to increase in capital/TA due to behavioural response           

|| Total Assets (bn) || Capital (bn) || Capital/TA (%) || Change in Capital/TA due to behavioural response* (%) || Estimated effect on WACC due to change in Capital/TA (bps)

|| A || B || C = B/A || D || E = D*393bps

1 - DTB || 15,463 || 744 || 4.8% || 0.0% || 0

2 – TE with increased capital || 7,191 || 322 || 4.5% || 1.8% || 7

3 – TE with reduced RWA || 7,191 || 195 || 2.7% || 0.0% || 0

4- TE with reduced TA || 4,152 || 195 || 4.7% || 2.0% || 8

Notes: Costs of additional capital are 393 bps per unit (replacing subordinate debt (BiB) by equity).

*Capital of the TE before behavioural response is EUR 195 bn (2.7%TA), and of the DTB EUR 744 bn (4.8%TA).

Table 13 shows the estimated change in the weighted average costs of capital (WACC) due to LAC held over the minimum required 8% TA following the different behavioural responses. Column C shows the amount of LAC (as a share of total assets) that the banks hold in excess of 8% TA under the different scenarios. Column D shows the costs of this LAC on the WACC. The estimated cost of the LAC is 133 bps reflecting the difference between the risk premium on bail-inable first debt (in the form of subordinated debt) and senior debt which it replaces.

The effect on WACC for the TE is estimated to be in the range of 0 to 2 bps depending on the behavioural response, while the DTB does not need to hold any LAC in excess of the regulatory minimum.

Table 13 -  Estimated change in WACC due to holding LAC over 8% TA due to behavioural response

|| Total Assets (bn) || LAC over 8% TA held due to behavioural response (bn) || LAC over 8% TA as a share of total assets || Estimated change in WACC due to LAC over 8% TA  (bps)

|| A || B || C = B/A || D = C*133bps

1 - DTB || 15,463 || 0 || 0 || 0

2 – TE with increased capital || 7,191 || 127 || 1.8% || 2

3 – TE with reduced RWA || 7,191 || 12 || 0.2% || 0

4- TE with reduced TA || 4,152 || 42 || 1.0% || 1

Note: Costs of additional LAC (BiB) are 133 bps per unit (replacing senior debt by subordinate debt).

Table 14 provides an estimate of changes in the WACC as result of the changed risks for bank capital (either in the form of equity or subordinate debt) after structural reform and behavioural response. In columns A to D the change in expected losses on capital compared to the undivided bank are calculated. If markets anticipate these changes in riskiness, the change in expected losses will be reflected in the risk premia. Column E and F provide a minimum estimate of the risk premium impact (on capital and on the WACC respectively) if the initial risk premium is the risk premium on subordinated debt, while column H and I show the maximum of the range based on an initial risk premium assuming all capital is plain equity.

The effect of a changed risk premium on capital on the WACC for the TE is estimated to be in the range of 0 to 1 bp, as the behavioural responses are calibrated such that the probability of default is equal to that of the undivided bank. While losses on capital increase in the behavioural responses, there is a reduction in the losses per unit of capital because of the increase in total capital. The WACC of the DTB declines due to lower expected losses on equity by 3 to 8 bps, as market risks are allocated to the TE activities and capital pooling is no longer possible.

Table 14 - Estimated change in WACC due to changes in the capital risk premia after behavioural response

Bank || Capital (% TA) || Expected losses on capital (% TA) || Expected losses on capital as a share of total capital || %-change compared to undivided bank || Change in risk premium on capital, compared to undivided bank (bps) || Change in WACC due to capital risk premium change (bps) Minimum || Estimated change in risk premium on capital, compared to undivided bank (bps) || Change in WACC due to capital risk premium change (bps) Maximum

|| A || B || C=B/A || D=(C-�)/� || E=D*307bp || F=E*A || H=D*700bp || I=H*A

1– Undivided Bank || 4.1% || 1.7% || 40% (�) || - || - || - || - || -

2 - DTB || 4.8% || 1.5% || 31% || -23% || -71 || -3 || -162 || -8

3 – TE with increased capital || 4.5% || 1.9% || 41% || 3% || 8 || <1 || 18 || <1

4 – TE with reduced RWA || 2.7% || 1.1% || 42% || 3% || 8 || <1 || 19 || <1

5- TE with reduced TA || 4.7% || 2.0% || 42% || 3% || 10 || <1 || 22 || 1

Note: Initial average risk premium on capital is 307 bps in the minimum (assuming capital is at the cost of subordinate debt) and 700 bps in the maximum (assuming plain equity).

Table 15 provides an estimate of changes in the WACC as result of the changed expected losses for first bail-inable bond (BiB) holders after structural reform and behavioural response. In columns A to D the change in expected losses on first bail-inable bonds compared to the undivided bank are calculated. Columns E and F provide the risk premium impact on capital and on the WACC respectively.

The effect of a changed risk premium on bail-inable first debt on the WACC for the TE is estimated to be in the range of -4 to -9 bps depending on the behavioural response. This result is driven by lower gross losses, the increased share of losses that are absorbed by capital and the lower pooling of the LAC (as it is no longer shared across the DTB and the TE of each “group”). The WACC of the DTB is estimated to go down by 1 bp due to the lower expected losses on DTB BiB holders.

Table 15 - Estimated change in WACC due to changes in the risk premium on first bail-inable bonds (BiB) after behavioural response  

Bank || Minimum bail-in capacity (excl. capital) (% TA) || Expected losses on BIB (%TA) || Expected losses on BIB as a share of minimum BIB || %-change compared to undivided bank || change in risk premium on BIB, compared to undivided bank (bps) || Change in WACC due to BIB risk premium change (bps)

|| A || B || C=B/A || D=(C-�)/� || E=D*307bp || F=E*A

1 – Undivided Bank || 3.9% || 0.9% || 23% (�) || - || - || -

2 - DTB || 3.2% || 0.7% || 21% || -11% || -33 || -1

3 – TE with increased capital || 5.3% || 0.7% || 12% || -46% || -142 || -7

4 – TE with reduced RWA || 5.5% || 0.5% || 10% || -56% || -173 || -9

5- TE with reduced TA || 4.3% || 0.7% || 15% || -34% || -104 || -4

Note: Initial average risk premium on subordinate debt is 307 bps.

6. Conclusions

SYMBOL simulations show that the effects of separation and the TEs possible behavioural responses lead to :

ü A reduction in gross losses plus recapitalisation needs across all banks in case of a financial crisis.

ü A reduction of the losses that can fall on bail-inable creditors in line with the BRRD, reducing potential financial stability risks and contagion.

The funding cost impact of the separation and behavioural responses depend on the extent to which changes to expected losses for the different stakeholders affect market risk premia on equity and bonds:

ü If lower expected losses on capital and bonds are not reflected in lower funding costs, the weighted average WACC across all banks in the sample increases by up to 3 bps as the Trading Entities need to hold more capital and Loss Absorbing Capacity (LAC) which increase their WACC by up to 9 bps. The Deposit Taking Banks have an unchanged WACC as they do not need to hold capital or LAC above minimum requirements as their probability of default and expected losses decrease, and they cannot reduce capital or LAC below minimum requirements to lower its WACC.

ü With changes in risk premia, the average WACC across all banks in the sample decreases (by 1 to 9 bps) as gross losses reduce and are partly allocated to the safety net and/or public finances. The WACC for the Deposit Taking bank is reduced by 4 to 9 bps as it no longer supports the more risky Trading Entity activities, while the increase in WACC for the Trading Entities is limited due to the reduced riskiness for bail-inable creditors (following behavioural responses). The reduction in the WACC should be considered together with an accompanying loss of revenues due to the lower RWA.

The analysis does not find a direct beneficial effect of structural separation on losses to public finances. This is the result of the assumption that both CRDIV and bail-in under the BRRD are considered as fully effective before the introduction of separation, thus leaving limited scope for further reductions in pass-through to the safety nets.

In a comprehensive assessment further social and private costs and benefits and macro-economic impacts that cannot be modelled with the SYMBOL model also need to be considered: e.g. loss of economies of scope and scale, legal costs, relocation costs, effects on asset pricing due to reduced liquidity effect, as well as other social and private benefits such as avoiding conflicts of interest, misallocation of resources, facilitating supervision, possible bank lending effect by the DTB.

Appendix section for Annex A10

Appendix A: List of banks

Table - List of banks in the sample in alphabetical order.

|| Institution Name || Label || Country

1 || Banca Monte dei Paschi di Siena SpA || MPS || IT

2 || Banco Santander SA || Santander || ES

3 || Barclays Plc || Barclays || UK

4 || Bayerische Landesbank || BayerLB || DE

5 || Belfius Banque SA || Belfius || BE

6 || BNP Paribas SA || BNPP || FR

7 || Commerzbank AG || Commerz || DE

8 || Crédit Agricole SA || CA || FR

9 || Danske Bank A/S || Danske || DK

10 || DekaBank Deutsche Girozentrale || Deka || DE

11 || Deutsche Bank AG || Deutsche || DE

12 || Deutsche Zentral-Genossenschaftsbank AG || DZG || DE

13 || DNB ASA || DNB || NO

14 || Groupe BPCE || BPCE || FR

15 || HSBC Holdings Plc || HSBC || UK

16 || ING Bank NV || ING || NL

17 || KBC Group NV || KBC || BE

18 || Landesbank Baden-Württemberg || LBBW || DE

19 || Landesbank Hessen-Thüringen Girozentrale || LBHT || DE

20 || Mediobanca - Banca di Credito Finanziario SpA || Mediobanca || IT

21 || Nordea Bank AB || Nordea || SE

22 || Portigon AG || Portigon || DE

23 || Royal Bank of Scotland Group Plc || RBS || UK

24 || Skandinaviska Enskilda Banken AB || SEB || SE

25 || Société Générale SA || SocGén || FR

26 || Svenska Handelsbanken AB || Svenska || SE

27 || Swedbank AB || Swedbank || SE

28 || UniCredit SpA || UniCredit || IT

29 || Standard Chartered Plc || StdCh || UK

Appendix B: Sample statistics before and after structural separation

Table - Summary of split of assets, RWA and capital for the sample in bn€, data as of 2011. Source: Bankscope, SNL and Commission elaboration.

|| Universal || DTB || TE

Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant || Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant || Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant

Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€) || Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€) || Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€)

1 || Banca Monte dei Paschi di Siena || 241 || 11 || 105 || 13 || 127 || 214 || 9 || 103 || 11 || 107 || 27 || 2 || 2 || 2 || 21

2 || BancoSantander SA || 1,252 || 51 || 566 || 72 || 686 || 1,062 || 42 || 547 || 59 || 566 || 189 || 9 || 19 || 12 || 119

3 || Bank DnB A/S || 274 || 11 || 137 || 18 || 174 || 254 || 9 || 127 || 15 || 138 || 20 || 2 || 10 || 4 || 35

4 || Barclays Plc || 1,870 || 50 || 468 || 59 || 566 || 970 || 38 || 411 || 45 || 426 || 899 || 12 || 57 || 15 || 139

5 || BayerischeLandesbank || 309 || 12 || 118 || 15 || 144 || 258 || 10 || 115 || 13 || 119 || 51 || 2 || 3 || 3 || 24

6 || Belfius Bank &Verzekeringen || 233 || 5 || 53 || 7 || 64 || 196 || 4 || 52 || 6 || 54 || 36 || 1 || 1 || 1 || 11

7 || BNP Paribas || 1,965 || 57 || 614 || 78 || 744 || 1,103 || 41 || 520 || 57 || 541 || 862 || 15 || 93 || 21 || 201

8 || BPCE Group || 1,138 || 30 || 388 || 52 || 499 || 970 || 25 || 376 || 43 || 412 || 168 || 5 || 12 || 9 || 86

9 || Commerzbank AG || 662 || 24 || 237 || 30 || 287 || 500 || 20 || 225 || 24 || 233 || 162 || 4 || 12 || 6 || 53

10 || CréditAgricole-SA || 1,724 || 30 || 274 || 42 || 404 || 1,214 || 23 || 250 || 33 || 316 || 509 || 6 || 24 || 9 || 88

11 || Danske Bank A/S || 461 || 14 || 122 || 15 || 148 || 327 || 11 || 113 || 12 || 118 || 133 || 3 || 8 || 3 || 30

12 || DekaBank Deutsche Girozentrale || 134 || 3 || 25 || 3 || 27 || 94 || 2 || 22 || 3 || 26 || 40 || 0.4 || 3 || 0.5 || 5

13 || Deutsche Bank AG || 2,164 || 36 || 381 || 48 || 462 || 968 || 26 || 316 || 35 || 329 || 1,197 || 10 || 65 || 14 || 132

14 || HSBC Holdings Plc || 1,975 || 87 || 931 || 119 || 1,133 || 1,516 || 69 || 860 || 94 || 895 || 459 || 18 || 71 || 25 || 236

15 || ING Bank NV || 961 || 31 || 330 || 42 || 400 || 865 || 26 || 321 || 35 || 333 || 96 || 5 || 9 || 7 || 67

16 || KBC GroepNV || 285 || 13 || 126 || 16 || 153 || 259 || 11 || 124 || 13 || 128 || 26 || 2 || 3 || 3 || 25

17 || Landesbank Baden-Wuerttemberg || 373 || 12 || 108 || 14 || 131 || 265 || 9 || 97 || 11 || 100 || 108 || 3 || 11 || 3 || 30

18 || Landesbank Hessen-ThueringenGirozentrale - HELABA || 164 || 6 || 57 || 7 || 69 || 124 || 4 || 51 || 6 || 53 || 40 || 1 || 6 || 2 || 16

19 || MediobancaSpA || 75 || 5 || 55 || 7 || 67 || 60 || 4 || 52 || 6 || 54 || 16 || 1 || 3 || 1 || 12

20 || Nordea Bank AB (publ) || 716 || 16 || 185 || 28 || 271 || 479 || 13 || 171 || 22 || 214 || 238 || 3 || 14 || 6 || 57

|| Universal || DTB || TE

Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant || Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant || Total assets (b€) || 31/12/2011 balance sheet || Basel III compliant

Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€) || Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€) || Reg. Capital (b€) || RWA (b€) || Regulatory Capital (b€) || RWA (b€)

21 || Portigon AG || 168 || 4 || 48 || 6 || 59 || 109 || 3 || 45 || 5 || 47 || 59 || 1 || 3 || 1 || 12

22 || Royal Bank of Scotland Group Plc || 1,804 || 48 || 526 || 67 || 636 || 964 || 35 || 451 || 49 || 468 || 839 || 13 || 74 || 17 || 167

23 || SkandinaviskaEnskildaBanken AB || 265 || 8 || 76 || 12 || 113 || 212 || 6 || 72 || 10 || 91 || 53 || 1 || 4 || 2 || 21

24 || SociétéGénérale || 1,181 || 27 || 349 || 44 || 423 || 736 || 21 || 306 || 33 || 317 || 446 || 7 || 44 || 11 || 105

25 || Standard Chartered Plc || 456 || 24 || 208 || 27 || 253 || 391 || 20 || 203 || 22 || 210 || 66 || 4 || 6 || 4 || 43

26 || SvenskaHandelsbanken || 275 || 8 || 57 || 14 || 134 || 253 || 7 || 56 || 12 || 113 || 22 || 1 || 1 || 2 || 21

27 || SwedbankAB || 208 || 7 || 55 || 11 || 103 || 178 || 6 || 53 || 9 || 84 || 30 || 1 || 2 || 2 || 18

28 || UniCreditSpA || 914 || 38 || 460 || 58 || 558 || 789 || 31 || 449 || 49 || 464 || 125 || 6 || 12 || 10 || 93

29 || Deutsche Zentral-Genossenschaftsbank AG || 406 || 8 || 100 || 13 || 121 || 112 || 6 || 92 || 10 || 95 || 294 || 2 || 8 || 3 || 25

|| TOTAL || 22,653 || 676 || 7159 || 937 || 8,956 || 15,442 || 531 || 6580 || 742 || 7051 || 7,210 || 140.4 || 580 || 198.5 || 1892

Appendix C: The SYMBOL model

SYMBOL (SYstemic Model of Banking Originated Losses) is a micro-simulation model[227] which makes use of individual banks’ balance sheet data to simulate banks’ losses due to the failure of its obligors and to derive the (aggregated) distribution of losses originated in the banking system. The main idea behind this model is that it is possible to estimate and average Probability of Default (PD) of the portfolio of obligors of a bank (the so called implied obligors’ PD) by inverting the Basel FIRB (Foundation Internal Ratings Based) formula for capital requirements.

The model is implemented in the following four steps:

Estimation of the implied PD for the obligors’ portfolio of any individual bank by inverting the Basel FIRB capital requirement formula. Under the FIRB approach the total capital can be written as a function of the average obligor’s PD, of the Loss Given Default LGD, of the maturity T and of the correlation R between the assets of the corresponding obligors (for more details see (De Lisa et al., 2010)):

CapRequirement=f(PD, LGD, T, R)

Being the capital requirement publicly available and being all the other parameters set to their regulatory average values, the formula can be inverted to numerically estimate the implied PD.

Generation, via a Monte Carlo simulation, of portfolio losses for individual banks. Once the implied average PD are estimated, individual bank losses are generated via a Monte Carlo simulation, taking into account the correlation between the assets of different banks due to the presence of common shocks in the economy. Banks’ losses are simulated on the basis of the loss distribution assumed in the Basel FIRB approach[228].

3. The output of the simulation is a matrix of gross losses Li,n  where i labels the bank and n the simulation run. In each run Li,n  is compared with the amount of bank’s capital Capi. If Li,n is lower than Capi the bank does not default but could need to recapitalize to a level equal to 8% of its RWA. On the other hand, if Li,n  is greater than Capi the capital the bank holds is not enough to cover the loss and this implies that the bank is in insolvency, suffering from an excess loss ELi,n:

ELi,n =( Li,n  - Capi){Li,n  - Capi>=0}

4. Aggregation of individual banks’ Li,n, ELi,n and of recapitalization needs to obtain aggregate losses at country/banking system’s level.

Appendix D: The Implied Obligor PD for the Trading Entities

As set out in Annex C, the SYMBOL model inverts the FIRB approach (Vasicek) credit risk formula for banks minimum capital requirement to estimate the probabilities of a default of bank obligors as assessed by the bank and the country's banking system regulator. Vasicek[229] shows that the model can also be used to simulate losses on a traded portfolio exposed to market and default risk, subject to the introduction of a correction term for the difference between the maturity of the traded securities and the holding horizon and to a market correlation factor. In addition Vasicek shows that the mark-to-market loss distribution of a traded portfolio would coincide with the standard FIRB loan-loss distribution if the maturity and holding horizon were to coincide.[230]

The base SYMBOL methodology, Vasicek formula, can therefore also be used to simulate losses on the portfolio of the TE, subject to some assumptions and conditions. In particular, the issues which need to addressed are: i) Differences between the confidence levels required for market risk and credit risk; ii) Assumptions and requirements regarding the holding horizon of the securities; iii) treatment of the additional “multiplicative factor” of 3 to 4 to be applied to calculated market risk according to the AIRB framework.

Regarding the first point, given that the Basel accord proposes a target MCR for the market risk which is based on a confidence level of 99%, we perform the inversion of the FIRB formula to recover the riskiness parameter at this level instead of the 99.9% required for credit risk.

Regarding the second point, we introduce the simplifying assumption that the holding period and maturity of the portfolio coincide (thus eliminating the need to consider the adjustments for market correlation and time horizon differences), as it is usually done on the credit side, and that the Basel 3 adjusted MCR is representative of the risk incurred over the whole simulation period.

Regarding the third point, the multiplicative factor of 3, could be interpreted[231] as a correction factor for model uncertainty. If this interpretation is adopted, then this can be thought as an estimator correction factor. The implied obligor probability of default should therefore be obtained based on the estimated MCR including the correction factor.

Given the above, we use the FIRB standard loss distribution also as a basis for the simulations of losses of the TE after taking into consideration the difference in confidence for calibration purposes (i.e. inverting the FIRB at 99% and not at 99.9%) and without applying any further correction to reported MCRs.[232]

It should be noted that the choice of a 99% confidence implies an increase in the riskiness of trading entities, while the choice of ignoring the difference between the holding horizon of trading securities and the simulation horizon implies a decrease. A precise quantification of the impact of these two drivers was not possible.

Appendix E:  SYMBOL simulated losses and historical losses during the recent crisis

Table A3 present the results of the SYMBOL simulations for the structural separation compared to historical losses of the recent crisis. To estimate the representative loss simulations for all EU banks, simulation results for the sample of 29 banks is divided by its share in total EU assets (about 56%). These simulations differ from those reported in Table 1, Section 1 of Annex XIII of the BRRD IA due to the different methodology required to simulate the effects of the structural separation. As the trading and banking portfolio of the banks are simulated separately, the correlation of losses across these portfolios is reduced from 1 (when they are simulated as a single undivided portfolio), to 0.5 which is the standard setting for the correlation of losses across banks in SYMBOL.[233]

As the very severe crisis (with excess losses at the 99.95th percentile) simulation is the most similar to real state aid figures during the last crisis in terms of banks' losses, also including recapitalization needs, the analysis in this report is developed for that simulation.

Table A3 - Aggregated losses in EU banking sector simulated with the SYMBOL model under Basel 3 10.5% minimum capital requirements (no contagion) and aggregated EU state aid used in recent crisis between 2008-2012  (€ billion. In brackets: values presented in the BRD IA on the basis of banks' 2009 and data and on state aid used between 2008 and 2010.

|| Severe crisis (99.90%[234]) || Very severe crisis (99.95%) || Recent crisis (Data 2008-2012)[235] || Extremely severe crisis (99.99%)

Extra-Losses (not absorbed by capital) || 44.3 (36.2) || 72.6 (79.9) || 116.8 (121.2) || 172.8 (266.7)

Extra-Losses (not absorbed by capital) + Recapitalisation funding needs to meet 8% MCR || 340.8 (295.6) || 447.8 (466.7) || 439.0+114.6=553.6 (409.0) || 750.7 (668.3)

Source: European Commission elaborations

Appendix F: Assets of large banking groups within the EU-27

Table  shows the share of total consolidated assets within the EU-27 for the largest EU banking groups included in the simulation exercise. These represent roughly 83% of the total sample’s assets. The second column reports the 2011 total consolidated assets, downloaded from Bankscope, the third column shows estimations of the percentage of these assets within EU-27 borders.

Table A4: Share of activity in the EU-27 for a number of large EU banking groups. Consolidated data from 2011

Bank Name || Country || Total Assets (b€) || Estimated % of total assets inside EU-27

Deutsche Bank AG || DE || 2,164 || 79.31%

HSBC Holdings Plc || UK || 1,981 || 41.05%

BNP Paribas || FR || 1,965 || 85.57%

Barclays Bank Plc || UK || 1,818 || 56.34%

Royal Bank of Scotland Group Plc || UK || 1,752 || 77.96%

Crédit Agricole S.A. || FR || 1,724 || 97.74%

Banco Santander SA || ES || 1,252 || 77.81%

Société Générale || FR || 1,181 || 90.11%

ING Bank NV || NL || 961 || 85.38%

UniCredit SpA || IT || 927 || 93.54%

BPCE SA || FR || 796 || 97.10%

Nordea Bank || SE || 716 || 87.70%

Commerzbank AG || DE || 662 || 97.09%

Danske Bank A/S || DK || 460 || 99.44%

Weighted Average for all banks || || || 79.05%

Source: ECB, Bankscope, European Commission elaborations (*)

ANNEX A11 Selected observations on submitted data templates

Introduction

As one element of input into the impact assessment, the Commission services have invited EU banks to submit data that illustrate the expected impact of stylised structural reform scenarios on the group's balance sheet, profit and loss account and selected other variables. The data request was part of the public consultation between 16 May 2013 and 11 July 2013 and can be retrieved here:

http://ec.europa.eu/internal_market/consultations/2013/banking-structural-reform/docs/data-template_en.xlsx

As structural reform targets large and complex too-important-to-fail banking groups, the Commission services in particular have encouraged data submissions from EU banks with the highest degree of systemic importance. Respondents have been requested as a matter of priority to complete the data request as well as to report underlying assumptions about relevant macroeconomic and other variables.

In that context, banks have been asked to provide information about balance sheet, profit and loss, and selected other variables in a number of different scenarios:

- balance sheet and profit and loss account information, as well as information about selected other variables, at year end 2012;

- simulated group balance sheet and profit and loss by end 2017, assuming in particular the implementation of the Bank Recovery and Resolution Directive and the Capital Requirements Directive IV/Capital Requirements Regulation; and

- simulated end 2017 balance sheet and profit and loss account in the stylised structural reform scenarios specified below and to be implemented by end 2017.

The two structural reform scenarios differ in the activities to be undertaken within a trading entity, as well as with respect to the strength of the separation between the deposit and trading entity.

The two reform scenarios are stylised and simplified and intend to usefully inform the assessment of the set of structural reform options being considered by the Commission services.

Stylised EU structural reform scenario 1 – main features:

Legal separation of certain trading activities from deposit taking activities within a banking group. Separation to be completed by end-2017. Only the following activities are to be excluded from the legally separate deposit entity (exhaustive list): proprietary trading, exposures to venture capital, private equity and hedge funds, and market making. The trading entity cannot be a subsidiary of the deposit entity. The deposit entity and the trading entity each have to comply with prudential requirements (including capital, liquidity, leverage and large exposure requirements) on an individual or subconsolidated level (i.e. in case there are several trading entities within the same corporate group all those entities can be consolidated for prudential requirements; similar principle applies in case there are several deposit entities within the same corporate group).

Stylised EU structural reform scenario 2 – main features:

Legal separation of certain trading activities from deposit taking activities within a banking group. Separation to be completed by end-2017. Only the following activities are to be excluded from the legally separate deposit entity (exhaustive list): All investment banking activity (please specify your definition used). The trading entity cannot be a subsidiary of the deposit entity. Completely independent funding and capitalisation of deposit and trading entity. Lending and asset sales between the deposit entity and the trading entity need to take place on a commercial and arm's length basis. The deposit entity and the trading entity each have to comply with prudential requirements (including capital, liquidity, leverage and large exposure requirements) on an individual or subconsolidated level (i.e. in case there are several trading entities within the same corporate group all those entities can be consolidated for prudential requirements; similar principle applies in case there are several deposit entities within the same corporate group). The deposit entity is not allowed to have exposures to financial institutions (including to any trading entity within its own corporate group), except for treasury functions, payments services, and letters of credit. No waivers should be granted. Independent risk management for deposit entity. Independent treasury management for deposit entity. Separate financial and supervisory disclosure requirements should be applied to the deposit entity and the trading entity.

These scenarios are stylised, relatively restrictive and prescriptive and do not correspond to concrete structural reform options and do not prejudge the policy choice to be made by the European Commission at a later stage.

The selected structural reform scenarios are restrictive in terms of exhaustively listing the banking activities that are to be performed by the legally, economically and operationally separate trading entity (limited flexibility to decide about in which entities activities can be performed), to ensure maximum comparability of results across banks.

The specified structural reform scenarios have been designed as being sufficiently distinct, so as to allow the Commission services to appreciate the incremental costs of different structural reform design issues.

High-level findings inferred from submitted data templates

Very few banking groups submitted data templates. Only six banking groups submitted templates that are sufficiently complete so as to allow detailed cross-bank comparisons. All but one of these six banks have total assets exceeding 1000bn EUR. The outlier bank has total assets around 250bn EUR. The limited sample of banking groups that provided information does not seem representative and hence does not allow to infer firm conclusions. The estimated impact on balance sheet size varies a lot across banking groups. Whereas one bank reports a small balance sheet increase (+1%), most others report single-digit balance sheet percentage reductions, with one outlier bank that reports an expected balance sheet decrease of 25%. The estimated size of the trading entity ranges widely across banks within a given structural reform scenario. The trading entity size ranges between 6% and 66% of the group’s total balance sheet for scenario 1 and between 7% and 71% for scenario 2. Only three banking groups reported the relative importance of proprietary trading. The reported share of proprietary trading in overall trading revenues ranges from a low of 0% to 4% of total trading revenues, which in turn is only a fraction of total revenues. The estimated impact of the proposed separation on the profit before taxes bank responses can be divided in three groups. For a first group of banks, the estimated impact amounts to at most a 10% profit reduction. For a second group of banks, the profit reduction is estimated to be around 40%. The impact on profit for a third group (bank) is an outlier, as a loss of 6 to 12 times the annual 2012 profitability is reported. Note: The ICB impact estimate of private costs is 33% of pre-tax profit of UK banks. Except for one bank, the pre-tax Return on Equity (RoE) of the deposit-taking entity (DE) is positive and surprisingly high, ranging between 8.4% and 17.2%. The pre-tax RoE of the outlier deposit-taking entity is estimated to be minus 10%. Despite this negative RoE, the DE rating for the outlier bank is estimated to be unaffected by the structural reform scenario. The pre-tax RoE of the trading entity (TE) is positive for three banking groups and negative for the other three banking groups. The range hovers between 5.2% and 11% for the first group and between minus 5.3% and minus 35.6% for the second group. The rating of the DE is unaffected for the 4 banks that reported rating estimates. The rating of the TE is estimated to go down for the four banks that reported rating estimates. Again, results vary a lot across banks. For the first scenario, the downgrades go from 1 notch to 5 notches[236]. For the second scenario, the range is 2 notches to 4 notches. The impact on revenues goes from 0% to minus 6% for the first scenario and from 0% to 16% for the second scenario. The importance of investment banking activities other than proprietary trading and market-making is surprisingly small, ranging between 0% and 10% of the group’s balance sheet. Proprietary trading and market-making make up for 90% of the balance sheet size of investment banking. Impacts on total costs also vary widely across banks. Total costs are estimated to go up for three banking groups. They are estimated to go down for two other banks. The first group reports a range of cost increases of 1% to 9%, whereas the latter group reports decreases of total costs of 2% to 11%. Funding costs are estimated to go up with approximately 35% and 20% for two banking groups. Other banks did not report in a format that allows inferring funding cost impact. There are numerous inconsistencies in the submissions, both within a given submission and across banks.

Annex A12 – Estimating the impact of a banking sector funding cost increase on GDP

1. Introduction and summary

This note presents the results of an exercise that quantifies the economic costs to society of an increase in the average funding cost of the EU banking sector, following bank structural reform. Three shocks of different magnitude to banks’ funding costs are introduced in a dynamic stochastic general equilibrium (DSGE) model.

At the outset, it should be stressed that any quantitative modelling intrinsically involves simplifications and assumptions, which yields results that are uncertain and imprecise. Given the inherent complexity of modern banking, and given that many social benefits and costs are dynamic in nature (often related to unobservable incentives), the results presented in this Annex should be treated cautiously.

With this caveat in mind, the results show that economic output, as measured by the level of GDP, would decline in the long run by between 0.04% and 0.1% after applying the funding cost increases. These effects are conservative estimates in the sense that the funding cost shocks are assumed to be completely passed through to customers.

2. Model description

The model used in this exercise is a DSGE model with a banking sector. The model belongs to a branch of applied general equilibrium theory that is influential in contemporary macroeconomics. The DSGE methodology attempts to explain aggregate economic phenomena, such as economic growth, business cycles, and the effects of monetary and fiscal policy, on the basis of macroeconomic models derived from microeconomic principles. The model distinguishes between borrowers (entrepreneurs) and savers. Savers divide up their financial wealth into government bonds, bank equity and deposits. In this model government bonds pay a risk free rate. The deposit rate is lower than the risk free rate because of a liquidity premium, i.e. banks charge depositors to make funds available on request. The rate of return on equity includes a risk premium, providing investors with a return above the risk free rate.

The banking system transforms savings of households into loans for entrepreneurs. Decision rules (demand for deposits, demand for capital, loan rates, etc.) are derived from maximising the value of the bank – the present value of bank dividends – subject to a capital requirement constraint. The banks buy labour services from the households, which are partly fixed and partly flexible in proportion to lending activities. Furthermore, the bank pays dividends to its shareholders.

The model shows the macroeconomic effects of permanently increasing (or decreasing) the risk premia for bank capital and/or increasing (or decreasing) the funding cost for deposits. The banking sector is represented by one stylised bank with a simple balance sheet, where total assets consist of loans (L) and liquid assets (government bonds: B). Total liabilities are the sum of deposits and wholesale funding (D) and bank equity (). The banking sector faces a capital requirement constraint, which is formulated in terms of risk weighted assets:

(1)                   

The concept applied is that of a consolidated balance sheet for the banking sector, which provides information about the stock of loans to the non-financial sector, and yields an estimate of the borrowing costs of non-financial corporations (NFC).

The main transmission channel of increasing funding costs is via higher lending rates for firms. An increase in funding cost increases marginal costs for banks. These costs are assumed to be shifted completely onto loan rates (there is a zero mark-up). In addition there is some tightening of the collateral constraint, since the value of capital of NFCs declines, because of the expected decline of dividends.

Higher lending rates and the tightening of the collateral constraint decrease investment and future consumption, and have a marginal effect on employment (since real wages adjust in the long run). In case of an increase in the return on equity (RoE) for banks, domestic households receive a higher return, which affects consumption positively in the short run. However, in the medium to long run this effect is dominated by the increase in the cost of capital. The increase in the deposit rate is modelled via a reduction in the supply of deposits of private households. Reduced savings increases consumption, which also dampens the negative macroeconomic effect in the short run.

3. Calibration and scenarios tested

Before applying any type of shock to the model, the model has to be calibrated to observed data. All parameters describing the non-banking part of the model are taken from Ratto et al. (2009). For the EU Banking sector, the following parameters are applied:

§ ratio of tier 1 capital to risk weighted assets: 6%;

§ risk weights for the two asset classes are 55% for loans and 5% for government bonds, respectively;

§ loan rate: 4.1%;

§ return on equity: 10%;

§ deposit and wholesale rate: 2.5%.

Three different funding cost shocks are applied to the model. These shocks are based on a collection of information of private costs to banks. The information stem from a survey, where banks were asked to estimate the resulting costs of stylised structural reform scenarios, and model-based estimates of funding cost increases. The latter model estimates come from the SYMBOL model, initially developed to assess the consequences of bank failures in the EU (see Annex A10). To this information was also applied a certain amount of own judgement, in order to reconcile some estimates with others, as the surveyed banks supplied a rather wide range of cost estimates, which often were internally inconsistent (see Annex A11).

The first two shocks are realistic, but still conservative measures of the economic impact of structural reform, as the model assumes that shocks are entirely passed through to customers, with maximum impact as a consequence. The first shock involves increasing the funding cost (the deposit and wholesale funding rates taken together) faced by banks that may fall under the regulation by 5 basis points. The second shock is an increase in the funding cost of 12 basis points, which represents an upper bound, as derived from other analysis in the Impact Assessment on bank structural reform.. This second scenario reflects the upper bound for the cumulative effect of introducing the Bank Recovery and Resolution Directive and implementing structural reform of the banking sector.

For illustrative purposes, the third shock is an extreme event where banks affected by structural reform faces an increase in their funding cost of 25 basis points. Obviously, this is not a realistic scenario. However, it serves to illustrate the linearity of the results within the ranges of increasing funding costs.

These shocks to funding costs only apply to those banks that are affected by the reform. It is assumed that banks affected by structural reform constitute approximately 55% of total banking assets in the EU. This is an estimate based on a threshold calibration exercise, which is part of the Commission’s Impact Assessment on bank structural reform. Yet again this is a conservative estimate, as it implies that the increasing costs apply also to some banks that basically have no or very little trading activity. Thus the shocks that are fed to the model are 2.75, 6.6 and 13.75 basis points, respectively.

4. Results

The increase in the funding rate (weighted average of deposit and wholesale rates) is introduced in the model via a reduction in the supply of funding. The amount of withdrawn funding is calibrated to generate the appropriate long-term increase in the funding rate. Thus, the three shocks to the funding rate are visible in the rightmost columns in Table 1 (2.75 basis points), Table 2 (6.6 basis points), and Table 3 (13.75 basis points). The increase in funding rate is entirely passed through onto the banks’ lending rates (the last line in each table).

Table 1:   Increase of funding rate: 5 bp (for 55% of bank assets)

Year || 1 || 2 || 3 || 4 || 10 || 50 || Long Run

GDP || -0.02 || -0.02 || -0.02 || -0.03 || -0.04 || -0.09 || -0.10

Capital stock || 0.00 || -0.02 || -0.03 || -0.04 || -0.09 || -0.21 || -0.24

Loan stock || -0.04 || -0.10 || -0.11 || -0.11 || -0.14 || -0.23 || -0.25

Funding rate || -2.60 || -1.69 || 5.04 || 2.67 || 2.45 || 2.70 || 2.75

Loan rate || -2.43 || -1.53 || 5.19 || 2.83 || 2.61 || 2.84 || 2.89

Employment || -0.02 || -0.02 || -0.01 || -0.02 || -0.01 || -0.01 || -0.01

Note:   Interest rates are deviations from baseline in basis points. The remaining variables are % deviations from baseline.

The long-run impact of the funding cost increase is a decline in the long-run level of GDP of about 0.1%, which is illustrated in the top row of Table 1. The shocks imply a decline in the capital stock and the loan stock. Employment is also affected negatively, but only marginally. From the table one can also infer what the result would be if the shock was lower, which there are indications of. For example, if the shock is instead 2 basis points, the lowest estimate, the economic impact is approximately 0.04%.

Table 2:                Increase of funding rate: 12 bp (for 55% of bank assets)

Year || 1 || 2 || 3 || 4 || 10 || 50 || Long Run

GDP || -0.04 || -0.04 || -0.05 || -0.06 || -0.10 || -0.22 || -0.24

Capital stock || -0.01 || -0.04 || -0.06 || -0.10 || -0.22 || -0.51 || -0.57

Loan stock || -0.10 || -0.24 || -0.27 || -0.26 || -0.34 || -0.55 || -0.59

Funding rate || -6.23 || -4.05 || 12.09 || 6.41 || 5.88 || 6.48 || 6.60

Loan rate || -5.83 || -3.67 || 12.47 || 6.79 || 6.26 || 6.83 || 6.95

Employment || -0.06 || -0.04 || -0.03 || -0.04 || -0.03 || -0.02 || -0.02

Note:   Interest rates are deviations from baseline in basis points. The remaining variables are % deviations from baseline.

The cumulative effect of introducing both the BRRD and bank structural reform, implying additional private costs, would amount to a decline in GDP of 0.2%. This result is presented in in the last cell of the top row in Table 2. The effect on employment remains marginal. The estimated impact of the cumulative effect of both introducing BRRD and bank structural reform does not change much as the impact of BRRD seems to dominate that of bank structural reform.

Table 3:                Increase of funding rate: 25 bp (for 55% of bank assets)

Year || 1 || 2 || 3 || 4 || 10 || 50 || Long Run

GDP || -0.08 || -0.08 || -0.10 || -0.13 || -0.22 || -0.45 || -0.50

Capital stock || -0.02 || -0.08 || -0.13 || -0.20 || -0.45 || -1.06 || -1.19

Loan stock || -0.22 || -0.50 || -0.57 || -0.54 || -0.71 || -1.14 || -1.23

Funding rate || -12.98 || -8.44 || 25.18 || 13.36 || 12.26 || 13.50 || 13.75

Loan rate || -12.14 || -7.64 || 25.97 || 14.15 || 13.04 || 14.22 || 14.47

Employment || -0.12 || -0.08 || -0.07 || -0.08 || -0.07 || -0.05 || -0.05

Note:   Interest rates are deviations from baseline in basis points. The remaining variables are % deviations from baseline.

Given the initial values and the calibration, the outcome of the model is to a large extent linear, which is illustrated in Table 3. When the cumulative shock is five times larger, also the impact on GDP is five times larger. The information in the tables therefore allow for interpolating the results of other shocks within the considered range. Beside the decline in output, both the capital stock and loan stock declines. Also in this more extreme scenario the effect in employment is very limited.

5. Comparison of results

In this section the derived results are compared with those of the Impact Assessment (IA) published by the UK government at the introduction to Parliament of the Financial Services (Banking Reform) Bill. Although the UK proposal is different from the Commission’s, the UK Impact Assessment may serve as a consistency check, both in terms of the size of the shocks and the outcome. The information in the UK IA can be translated to figures that are comparable to those used and derived in this note. However, this requires making use of additional data sources and making some more or less restrictive assumptions.

Table 3:    Breakdown of private costs of ring-fencing, UK proposal

On-going costs, per year || Low || High

Capital || £1.3bn || £2.6bn

Funding || £70m || £890m

Operational || £150m || £530m

Depositor preference || £200m || £380m

Total on-going costs, per year || £1.7bn || £4.4bn

According to the modelling approach of the UK government, total private costs for UK banks will range between GBP 1.7-4.4 billion, of which capital costs are in the range of GBP 1.3-2.6 billion, and wholesale funding costs are in the range of GBP 0.07-0.89 billion. The UK IA also considers two other on-going costs: operational costs and costs due to depositor preference, i.e. creditors preferring insured deposits to senior unsecured debt (see Table 3).

As the balance sheet in the in the DSGE model is simple and scaled down, all costs other than capital costs are bundled together to calculate a comparable shock to be applied in the DSGE model. This bundling of almost all costs and applying them as funding costs in the DSGE framework is of course a crude simplification, which allow for conceptual errors in terms of how the different costs affects a bank and are passed on to customers. Nevertheless, the result is that funding costs would increase in a range of GBP 0.42-1.8 billion, which translates into a funding cost increase between EUR 0.49-2.1 billion (with an average exchange rate for 2013 of 0.853 GBP/EUR).

According to the ECB's consolidated banking statistics, interest expenses were EUR 68.44 billion. Given the assumptions, these would increase to between EUR 68.9-70.5 billion. Applying this range of costs to the total amount of deposits and debt certificates (including bonds), yields a funding cost increase for the domestic banks in the UK in the order of 1-4 basis points. Adjusting the figures by the weight of the affected banks (0.55), imply a chock between 0.52-2.24 basis points. With the DSGE model, this range of shocks yields a decline in GDP of between 0.02%-0.08%.

The Commission’s estimates and the UK estimates are comparable, both in terms of magnitude of private costs and the impact on the real economy. The UK IA estimates a reduction in long-run GDP level of 0.04%-0.16%. Note that the UK makes use of a different model to generate their results. They use the NiGEM model, developed by the National Institute of Economic and Social research. NiGEM is an estimated model, which uses a ‘New-Keynesian’ framework. The model is structured around the national income identity, can accommodate forward looking consumer behaviour, and has many of the characteristics of a DSGE model. Unlike a pure DSGE model, NiGEM is based on estimation using historical data. Even though different estimation techniques and different models have been used, the results are very close to each other.

Annex A13 – Shadow banking - Securities finance transactions and transparency

1. introduction

The 2008 crisis was global and financial services were at its heart, revealing inadequacies including regulatory gaps, ineffective supervision, opaque markets and overly-complex products. The crisis highlighted the need to improve regulation and monitoring not only in the traditional banking sector but also in the area of non-bank credit activity, called shadow banking. The shadow banking system can broadly be described as “credit intermediation involving entities and activities (fully or partially) outside the regular banking system”. In practice it includes entities which raise funding with deposit - like characteristics, perform maturity and/or liquidity transformation, allow credit risk transfer or use direct or indirect leverage.

Shadow banking features high on the international agenda. G20 Leaders have asked the Financial Stability Board (FSB) to look into shadow banking in order to identify the main risks and make recommendations. The overarching aim, as reaffirmed on several occasions by the G20, is to eliminate the dark corners in the financial sector that have a potential impact on systemic risk or merely result from regulatory arbitrage and extend regulation and oversight to all systemically important financial institutions, instruments and markets.

Because of its size and close links to the regular banking sector, the shadow banking sector poses a systemic risk. The first factor is size. The latest studies indicate that the aggregate shadow banking assets are about half the size of the regulated banking system. Despite the fact that shadow banking assets have decreased slightly since 2008, the global figure at the end of 2012 was €53 trillion[237]. In terms of geographical distribution, the biggest share is concentrated in the United States (around €19.3 trillion) and in Europe (Eurozone with €16.3 trillion and the United Kingdom with around €6.7 trillion). The second factor which increases risks is the high level of interconnectedness between the shadow banking system and the regulated sector, particularly the regulated banking system. Any weakness that is mismanaged or the destabilisation of an important factor in the shadow banking system could trigger a wave of contagion that would affect the sectors subject to the highest prudential standards.

The response to the crisis has been international and coordinated through the G20 and the FSB. The FSB has suggested that as long as such activities and entities remain subject to a lower level of regulation and supervision than the rest of the financial sector, reinforced banking regulation could drive a substantial part of banking activities beyond the boundaries of traditional banking and towards shadow banking. For this reason the FSB under the lead of the G20 initiated at the end of 2011 several work streams aimed at identifying the key risks of the shadow banking system. These work streams include: (i) the interaction between banks and shadow banking entities; (ii) the systemic risks of Money Market Funds (MMFs); (iii) the regulation of other shadow banking entities like hedge funds; (iv) the evaluation of existing securitisation requirements and; (v) the use of Securities Financing Transactions (SFTs) like securities lending and repurchase agreements (repos).

Before the G20 and the FSB looked at the shadow banking system, the hedge funds were singled out as an area of grave concern. In April 2009, the G20 called for the hedge funds and their managers to be registered and properly supervised. Particular attention was given to their use of leverage and counterparty exposures. This is for this reason that the Commission proposed as early as April 2009 a directive on Alternative Investment Fund Managers (AIFM), including managers of hedge funds. The shadow banking regulatory agenda of the Commission has been set out in a Communication adopted in September 2013.

The recent financial crisis has also shown how critical funding liquidity risks can be in shaping the fate of individual institutions and in transmitting contagion across the financial markets. The procyclicality of funding liquidity created by private financial players, especially shadow banking entities, can be disruptive. It helped to fuel the financial bubble with liquidity generated by several forms of asset inflation. The rehypothecation of the collateral to support multiple deals, in particular securities lending and repurchase agreements, allowed for increased liquidity as well as the build-up of hidden leverage and interconnectedness in the system. When confidence in the value of assets, safety of counterparties and investor protection collapsed it created wholesale market runs leading to a sudden deleveraging and/or public safety nets (central bank facilities, etc.).[238] In this context, trust and funding liquidity evaporated and it became impossible for even the biggest and strongest banks to access either short or long-term funding.[239]

The FSB recommendations on shadow banking of 29 August 2013 have been formally endorsed at the G20 summit in St Petersburg.[240] They cover large areas of the financial system, notably the wide spread use of securities lending and repurchase agreements, also called securities financing transactions (SFTs), and of rehypothecation. These techniques are used by almost all actors in the financial system, be they banks, securities dealers, insurance companies, pension funds or investment funds. SFTs use assets belonging to an entity to obtain funding from or to lend them out to another entity. The main purpose of SFTs is therefore to obtain additional cash or to achieve additional flexibility in carrying out a particular investment strategy. The FSB and G20 have concluded that SFTs have the propensity to increase the built-up of leverage in the financial system as well as to create contagion channels between different financial sectors. This international work on shadow banking comes at a time where the banks are being subject to more stringent rules, including the proposed rules on the structural separation of banking activities. Confronted with this new legislative framework, there is no certainty that banks will not shift parts of their activity into less regulated areas as shadow banking. In order to closely follow market trends regarding entities whose activities qualify as shadow banking, in particular in the area of SFTs, it is necessary to implement transparency requirements that could inform the regulatory authorities about the next steps that would be required to deal with these issues. In this regard, the work undertaken by the FSB gives some precision about the transparency level that is required.

In 2013, the FSB adopted 11 Recommendations to address the risks inherent to securities lending and repurchase agreements. This impact assessment will only touch upon the issues related to the transparency of the SFTs markets and rehypothecation, i.e. Recommendations 1, 2, 5 and 7.

When assessing the transparency of the SFTs markets and rehypothecation, three main themes emerge: (1) the monitoring of the build-up of systemic risks related to SFT transactions in the financial system; (2) the disclosure of the information on such transactions to the investors whose assets are employed in these transactions; and (3) the contractual transparency over rehypothecation activities. EU regulatory authorities lack the necessary data to better monitor the use of SFTs and the risks and the vulnerabilities for the stability of the financial system that they imply. Access to this information will give the possibility to regulators to better design and apply their macro-prudential tools. At the same time the investors are not properly informed whether and to what extent the investment fund, in which they have invested or plan to invest in, has encumbered or intends to encumber investment assets by means of engaging in SFTs and other equivalent financing structures that would create additional risks for the investors. Finally, insufficient contractual transparency makes clients uncertain about the extent to which their assets can be rehypothecated, or about the risks posed by rehypothecation.

2. Securities financing markets in the EU

The securities financing markets represent one part of the shadow banking universe. According to the FSB definition, SFTs are considered as a shadow banking activity irrespectively of the entity that is performing such a SFT activity. The following analysis therefore covers all entities that use SFTs.

Economic context

Example of the securities lending case, Source: International Securities Lending Association (ISLA)

SFTs consist of any transaction that uses assets belonging to the counterparty to generate financing means. In practice, this mostly includes lending or borrowing of securities and commodities, repurchase (repo) or reverse repurchase transactions, or buy-sell back or sell-buy back transactions. All these transactions have similar, even identical, economic effects. The two main differences between repos and securities lending are their different transaction structure and their different purpose. In terms of transaction structure, securities lending occurs when an institutional investor agrees to lend out its securities to another party in return for a fee and collateral, while a repo is the sale of securities together with an agreement for the seller to buy back the securities at a later date. In terms of transaction purpose, securities lending is being driven significantly by a demand to borrow securities (for short selling purposes, trade settlements, etc.) whereas repo is more often driven by a desire to either borrow or lend cash. Market participants rely heavily on bilateral repos for financing. SFTs can be conducted on a bilateral basis, using a triparty agent, via an agent lender or being centrally cleared.

While not being directly associated with a SFT, other financing structures may produce equivalent effects. Those other financing structures include, for example, total return swaps (TRS), collateral swaps or liquidity swaps that are often used interchangeably with classic SFTs by investment funds. For the purpose of reporting to investors, those other financing structures will be included in the scope in addition to the SFTs. With regard to the reporting to competent authorities, the reporting requirement will be limited to SFTs because derivative contracts are already covered by European Market Infrastructure Regulation (EMIR)[241] reporting obligations.

The EU repo market tripled between 2001 and 2011 from EUR 0.9 trillion to EUR 3.1 trillion.[242] The June 2013 market survey[243] of the International Capital Market Association estimated the value of the outstanding repo contracts of the participating 65 institutions at EUR 6.1 trillion and the growth of the market over the preceding six months at 8.6%.[244]

The 2013 ECB's Euro Money Market survey[245] does not provide absolute outstanding levels but provides some statistics relating to the European repo market denominated in euro. There is a relatively high level of concentration in the euro repo market as the top 20 banks represent more than 82% of the total repo activity of the 161 credit institutions surveyed. It is also important to note that 71% of all bilateral euro repo transactions were cleared by central counterparties, compared to 56% in 2012.

There is no publicly available data on securities lending transactions in the EU. Several private data vendors, however, conduct private market surveys on securities lending. According to International Securities Lending Association, global securities lending stands at EUR 1.4 trillion.

Financial institutions responded to collateral scarcities resulting from the shift from unsecured lending by engaging in more collateral management, including optimising the use of available collateral to enhance liquidity, e.g., collateral swaps. An important element in this collateral management is rehypothecation. A Data Explorers survey (from investment banks) is the only data source available that covers the total global amount of rehypothecation. It estimates that securities of EUR 0.4 trillion were subject to rehypothecation in the EU in 2011, while globally it reached EUR 1.2 trillion in 2011. The IMF estimates that globally hedge funds provide an additional EUR 0.6 trillion of securities. This implies a total of EUR 1.8 trillion of collateral available in 2011 (i.e. EUR 0.6 trillion plus EUR 1.2 trillion). Thus, rehypothecation contributed about EUR 2.8 trillion to globally available collateral (EUR 4.6 trillion according to IMF estimates) in 2011. The players facilitating rehypothecation and the beneficiaries of the funding liquidity it gives are concentrated. The 14 major investment banks (G14) accounted for 86% of the rehypothecated collateral at the end of 2011 through their shadow banking activities with MMFs.[246] This translated into the financing of more than 30% of the financing of the total liabilities of shadow banking entities between 2007 and 2010.[247]

Regulatory context

The EU rules on capital requirements[248] require supervisory reporting of aggregate data on repo transactions of credit institutions, but existing requirements are not detailed and frequent enough for the purposes of monitoring of financial stability. SFTs involving entities such as investment funds, pension funds and insurance companies are rarely covered by existing supervisory reporting requirements despite the fact that they can give rise to financial stability concerns. Although the Markets in Financial Instruments Directive (MiFID)[249] includes a reporting obligation on any transaction in financial instruments, level 2 implementing measures[250] exempt securities financing transactions from this reporting requirement (notwithstanding the record-keeping of client orders and transactions).

The provisions regarding the reporting to investors on the use of SFTs and other financing structures are scattered in different places in the Directive on undertakings for collective investment in transferable securities (UCITS) [251] directive and in the ESMA guidelines on ETFs and other UCITS issues[252]. Neither SFTs nor other financing structures are defined in the UCITS or Alternative Investment Fund Managers (AIFM)[253] directive but these activities are covered through other definitions and obligations. Overall there is a lack of harmonisation and granularity in the existing reporting standards.

The EU’s current regulatory framework does not take account of the systemic issues posed by shadow banking, such as the rehypothecation of collateral. The Financial Collateral Directive[254] lacks clarity on the operational processes that should be followed where collateral takers decide to reuse securities collateral given using a security interest. Client asset protection is a key feature of the MiFID. Currently under revision,[255] it requires the investor’s consent for the intermediary’s use of its assets, investors can be left unprotected where an intermediary uses a title transfer to use the investor’s securities. In essence, the legal framework governing how securities are held and used is currently left to Member States' law. It is composed of a patchwork of national laws. Nevertheless, each of these instruments has a different limited personal and material scope. Together, these measures cover only some of the aspects relevant to how securities are used by financial markets and leave some important gaps and inconsistencies in the regulatory framework, in particular in relation to shadow banking activities.

3. Problem definition Driver 1: lack of comprehensive (frequent and granular) data on securities financing transactions

SFTs display structural similarities with banking activities as they can lead to maturity and liquidity transformation and increased leverage, including short-term financing of longer-term assets. They are, therefore, considered as shadow banking activities by the FSB. During the financial crisis, the authorities responsible for the monitoring of financial stability encountered significant difficulties to anticipate the emergence of systemic risks due to the lack of timely and comprehensive data on trends and developments in securities financing markets. Moreover, existing information gaps and lags prevent regulators from identifying the built-up of financial stability risks that would prove detrimental in times of a credit or liquidity crisis. The absence of data also prevents regulators from promptly taking the measures necessary to mitigate the negative effects of a potential crisis. Moreover, the lack of frequent and granular data made it impossible to develop a comprehensive picture across the full range of market participants in these markets, especially on the interactions of the regulated banking sector with shadow banking entities.

The FSB recently published a summary of the available data to regulators on securities lending and repos showing the lack of frequent and granular data on EU securities financing markets.[256] Moreover, in a paper published by the European Systemic Risk Board (ESRB), the authors concluded that the information available to EU regulatory authorities was not sufficient for the purpose of monitoring the systemic risks that may arise from SFTs.[257] Existing industry data or data collected in other publicly available surveys displays weaknesses in relation to the level of granularity, coverage of instruments and of institutions and their geographic coverage across Member States. This makes it particularly difficult to compare and use the data from different surveys for prudential purposes.

The lack of transparency is most acute with respect to bilateral transactions. In these areas no frequent and granular market data is readily available. Data on securities lending is very limited or not available at all, while information on key indicators for monitoring of the financial stability risks such as haircut levels, remaining maturity of collateral, reuse of collateral, is not typically available.

Driver 2: SFTs create conflicts of interests between the fund managers and the investors

With a volume of assets under management around EUR 9 trillion, investment funds are heavily engaged in SFTs and other financing structures in Europe. The majority of the European investment funds, representing around 70% of the assets under management, operate under the rules of the UCITS Directive while all other funds operate under the AIFMD. Generally, investors choose an investment fund according to its publicly stated investment strategy. This strategy comprises the asset classes in which the fund intends to invest but also the investment techniques that the fund intends to employ. SFTs are currently not presented as being an integral part of an investment strategy. Asset managers argue that SFT are not part of their core strategy and, at best, play an ancillary role. SFTs may, however, have significant impacts on the performance and the risk profile of the fund as they lead to expose the funds to additional risks.

A common feature of all SFT techniques is that they involve exchanging assets belonging to the investment fund with an external counterparty. While sharing similar structures, these techniques are not necessarily used for the same purpose.

1. Securities lending involves a fund lending its investment assets against a "lending fee". The lending proceeds can be used to enhance the return of the fund or to decrease the management fees. Securities borrowing is used by investment funds mainly to cover short positions (mostly relevant for AIFs). Liquidity or collateral swaps often take the same form as classic securities lending transactions.

2. Repos are used by investment funds to generate cash. This cash is then used to finance additional investments of the fund. Repos are used to increase the leverage of the fund. Reverse repos are used by investment funds to lend cash on a secured basis mostly to credit institutions.

3. Other financing structures include for example TRS. With a TRS an investment fund collects the investor's cash and passes it on to a TRS counterparty, usually an investment bank. Although not explicitly mentioned in the FSB recommendations, TRS create the same type of risks as securities lending and repo. TRS are used by managers because they offer exposures to strategies that would be difficult or too costly to implement otherwise.

SFTs are used because they offer managers economic interest or management flexibility. Their use is entirely subject to the manager's discretion which is driven by a motivation that might not necessarily be aligned with the interests of the investors. This could raise principal-agent problems. This agency dilemma may incentivise managers to act to increase their own profit, before and above considerations linked to the interests of their investors.

Conflict of interest may appear in the context of security lending transactions where the fund manager lends investment assets of the fund for a fee and receives collateral as a guarantee in case of default of the borrower.  It is common practice for the fund manager to retain part of the fee. This practice creates a conflict between the interests of the manager and the investors, because as more assets of the fund are lent out or more of the collateral received as guarantee is of bad quality, the higher the lending fees that the manager can expect. On the other hand, high levels of securities lending and collateral of inferior quality increases investors' exposure to risk.

This conflict may be worsened when the manager and its counterparty managing the securities lending program, the agent lender, belong to the same corporate group. In that case, the fund manager may have little ability to negotiate favourable terms for the interests of the fund investors, as it will be confronted with a parent or sister entity that has controlling power over the management company as well (i.e. a banking entity). TRS also involve possible conflict of interests when they are concluded on an intra-group basis. A conflict exists because the banking counterparty has an interest in posting collateral of lower quality as a way to remove these assets from its balance sheet in order to avoid regulatory requirements.

Conflicts of interests are inherent between funds’ managers and funds’ investors but are more likely to occur in the relative opacity that surrounds the use of SFTs. Especially fund investors are insufficiently informed about the existence of this financing tool or the extent of assets that are encumbered by SFT transactions. The relevant sectorial legislation, UCITS and AIFM directives, do not treat SFTs alike traditional investments of the fund. But like the primary investments of the funds, SFTs change the risk profile of the fund and will often change the fund's investment profile. These changes constitute material information that is needed to assess the risk and reward profile of the fund. However, this information is not properly disclosed to investors or at least not with the sufficient degree of granularity. Fund investors, even the institutional investors like the insurance groups or the pension funds, have therefore little means of assessing whether these transactions are in their interests or not.

Driver 3: Rehypothecation creates risks for clients and for engaging counterparties

For the purpose of this report, “rehypothecation” is defined as any pre-default use of assets collateral by the collateral taker for their own purposes. Rehypothecation is used in bilateral transactions between commercial market participants (i.e. dynamic rehypothecation) and between intermediaries and their clients (i.e. static rehypothecation). When the intermediary or the counterparty exercises its rehypothecation right, the ownership right is replaced with a contractual right to the return of equivalent securities. This is not protected as MiFID only protects a client's ownership rights. Thus, rehypothecation works until bankruptcy; if an intermediary defaults, a client with a contractual claim is an unsecured creditor, whose assets are tied to the insolvency estate and they have to line up with other unsecured creditors.

Static rehypothecation has declined since Lehman’s collapse as clients demanded segregation of their securities from the ones owned by their intermediaries or limited the amount of securities that intermediaries could take as collateral. Data from August to November 2008 shows a sudden drop in rehypothecable assets  at Morgan Stanley (fall of 69 %), Merrill Lynch (51 %) and Goldman Sachs (30 %).[258] However, anecdotal evidence indicates a trend reversing its decline as confidence returns that governments will not allow another major intermediary to fail.

There are also drivers incentivising dynamic rehypothecation of collateral received from counterparties. In a survey, 76% of respondents agreed that entering or expanding rehypothecation is a key lever to increase trading revenues related to collateral utilisation in the context of collateral scarcity.[259] But too much rehypothecation (whether static or dynamic) has the potential to threaten financial stability. Rehypothecation allows securities collateral to be used to create multiple obligations that interconnect different parties. The resulting obligations amount to a multiple of the value of the securities, creating concerns as the chains are opaque and hidden from participants and regulators. Maturity and liquidity mismatches allow for hidden leverage and risks to build that increase as the chain lengthens and more deals are secured using the same collateral.[260] In contrast, overall available collateral has declined post-Lehman and the length of the collateral chains has also shortened.[261]

The opacity and legal certainty as to a legal position can undermine confidence in counterparties and magnify a financial crisis. These problems are compounded by the fragmentation and barriers posed by diverging legal frameworks in the internal market. Rehypothecation can deprive clients of their investment and prevent counterparties from exercising rights attached to securities. This is even truer in an internal market context as clients may be unaware of the consequences of rehypothecation.

Problem 1: Regulatory authorities are unable to effectively monitor the use of SFTs

Regulatory authorities have experienced difficulties in monitoring, in a timely and granular manner, the developments in securities financing markets and the relevant risks. These markets are complex, evolve rapidly and involve a variety of participants. Participants range from regulated credit institutions to insurance companies, broker-dealers, asset managers and pension administrators. The recent financial crisis showed that securities financing markets are vulnerable to bank-like runs and fire sales of the underlying collateral, especially when the value of the assets is decreasing (e.g. 2008). Moreover, the assumption that securities financing is always robust even in stressed market conditions proved to be flawed as this led to the formation of interconnections among markets and market participants and the propagation of contagion.

Since SFTs are structured in a variety of ways, it can be difficult to identify the real risks individual market participants incur or pose to financial stability. According to the FSB, the lack of appropriate market transparency left regulatory authorities repeatedly dealing with relatively late-stage market developments that sparked the transmission of systemic risk during the financial crisis. Authorities had a limited overview of the maturity, liquidity and credit risk transformation taking place through SFTs. It was very difficult for them to detect the accumulation of risks and anticipate the consequences of the failure of a systemically important player such as Lehman Brothers International.

In general, SFTs provide funding and thus liquidity to financial markets. Consequently, some of the inherent risks of repos and securities lending transactions are similar to the risks of bank intermediation. As many shadow banking entities actively participate in SFTs markets, while this activity is not apparent to regulatory authorities, SFTs can contribute to the build-up of leverage and lead to significant systemic risks. For example, SFTs can cause significant maturity and liquidity transformation to take place outside the regulated banking system. This happens when a non-bank is financing long-term assets with short-term SFTs or collateralising less liquid assets in order to obtain liquid funding. In stressed market conditions, the risks related to maturity and liquidity transformation can lead to the default of shadow banking entities, which in turn can negatively impact the banking sector.

In the run-up to the crisis, leverage in securities financing markets was increased because of the low levels of collateralisation (i.e. no or low haircut requirements) combined with often inadequate valuation of the underlying collateral.[262] However, after the outbreak of the crisis, market participants started to require higher amounts of collateral given the falling prices of the collateralised assets and funding was withdrawn from the market. Once the value of collateral decreases, SFTs that rely on this collateral may no longer be sustainable and may need to be unwound quickly. Thus, SFTs led to a brusque deleveraging and put additional pressure on leveraged credit institutions relying on SFTs markets. SFTs can further amplify systemic risk through the interconnectedness between financial institutions, especially between banks and shadow banks.  

The dynamics of SFTs markets can be impacted by significant risks of fire sales i.e. selling of the collateral received by the borrower following its default. Moreover, fire sales have a pro-cyclical effect on market prices in case of distressed market conditions as they further exacerbate the downward trend. Many of these risks can significantly impact asset quality, counterparty credit risk and the availability of funding in securities financing markets, thus impacting the activities of the regulated banking sectors.

Problem 2: SFTs are used at the detriment of the investor

In an investment fund, investors expect to be exposed to traditional market risk linked to the investment of the fund. The gains or losses that investors realize are expected to be generated by the investment policy that is pursued in accordance with the stated investment strategy of the fund as communicated to and agreed with the investors. SFTs, however, expose the investors to counterparty risk, leverage and liquidity risk that are not part of the stated investment strategy of the fund.

Should the counterparty to a SFT default, the fund is fully exposed to the loss, e.g. the collateral posted with that counterparty or the securities loaned to that counterparty. For example, when a substantial portion of the assets of a fund have been lent out to a single counterparty, the default of this counterparty might put in peril the viability of the fund. It is not uncommon that funds rely on a single counterparty when managing their entire securities lending program. In this case, the default of that counterparty may entail severe losses for investors.

In the case that the cash investments by a fund manager subsequent to a repo transaction cause losses, the fund investor will bear the losses – which will manifest themselves in a lower performance of the fund. In addition, as SFTs encumber the liquidity of the fund, investors might not always be able to redeem their investments as promised in the funds redemption policies. Fund managers may be forced to suspend redemptions for the period necessary to unwind the SFT transaction. It has been demonstrated that some funds may encumber up to 100% of their Net Asset Value in such transactions[263], thus undermining their liquidity. According to a study on major providers of exchange traded funds (ETF, a type of "listed" investment fund), 25% of ETFs can have more than 50% of their assets loaned out and 3% had, in 2011, on average more than 90% of their assets on loan.

Securities lending generates additional revenues for the fund through the fee that is earned. These revenues are not always distributed back to fund investors and may be used at the sole discretion of the fund manager. According to the above study realized on European ETFs, the portion of revenues returned to the fund could range from 45% to 70% of gross revenue, with the fund manager and the securities lending agent retaining the balance. Only a few fund managers return 100% of the net revenue. The agent lenders are charging fees amounting to 10% to 40% of the gross revenue. When the fund performs securities lending activities, the investor bears 100% of the risks related to these activities but receives only a fraction of the return that is generated.

Problem 3:  Rehypothecation shifts the legal and economic risks in the market

Collateral scarcity and the need for funding liquidity is encouraging rehypothecation of assets instead of simply keeping collateral as insurance against a default. Although the increased use of rehypothecation eases this scarcity, it shifts the legal and economic risks in an already complex financial system from regulated sectors to shadow banking by involving non-financial institutions through repos and securities lending. These risks centre on the difficulty of identifying who is exposed to whom. The failures of Lehman Brothers and Bear Stearns also revealed the risk of runs by investors on large intermediaries due to asset protection concerns about the safety of their property.

This increased reliance on rehypothecation of collateral can contribute to systemic risk, since in times of market stress it motivates counterparties to withdraw their assets if they fear the insolvency of their counterparty or intermediary - events that significantly contributed to the failures of Lehman and Bear Stearns[264]. This is inter alia since there is the lack of contractual transparency of rehypothecation activities. When assets are rehypothecated, the collateral provider's ownership rights are weakened so that they have only a claim for equivalent securities to be returned by the collateral taker. The failure of a major intermediary caused by a run can then set off a domino effect across the financial system where highly interconnected counterparties are unable to mitigate their exposures to each other or are unable to access the securities they need to secure funding because they are trapped in the insolvent estates of failed intermediaries. Mounting distrust between investors also causes sudden deleveraging as the liquidity given by churn disappears.

Rehypothecation can be seen as a developing threat that will become a major problem again if it remains unaddressed. Despite the dangers demonstrated by Lehman and MF Global, short-term incentives and procylicality are increasing again, heightening the risk that it could trigger another crisis. The extension of funding liquidity that it gives to the market is dependent on confidence and susceptible to sudden deleveraging with consequences for the real economy.

4. Analysis of subsidiarity

Based on the nature of the problems outlined in the above analysis, several major justifications that meet the principle of subsidiarity for action at the EU level become apparent. The majority of SFTs as well as rehypothecation activities are performed on a cross-border basis between entities that often do not have their seats in the same jurisdiction and involve assets and currencies issued in different jurisdictions. Acting at the European level is the minimum to cover SFTs to the greatest extent possible and to allow regulatory authorities at national and EU level to have a comprehensive overview of the SFTs markets across the entire EU. The effectiveness of remedies implemented in an autonomous and uncoordinated way by individual Member States would likely be very low as such remedies would be able to capture just a portion of the market. Furthermore, given the systemic impact of the problems, uncoordinated action may even prove counterproductive because of the risk of data fragmentation and incoherence. Only aggregated data at the European level can give the necessary macroeconomic picture that is required to monitor the use of SFTs.

As regards investment funds, the European fund industry has an important cross-border dimension. The share of cross-border assets for the European investment funds industry as a whole (UCITS and non-UCITS AuM) has risen from 21% in 2001 to 45% in 2012[265]. This means that around one of two investors buy a fund that is not domiciled in their country of residence. It is therefore important that the investor protection standards are applied evenly across the EU in order to ensure that all European investors benefit from the needed transparency over the use of SFTs..

5. objectives

The general objectives are to:

(1) Ensure financial stability in the internal market by preventing the build-up of systemic risks;

(2) Increase the protection of investors and clients.

Reaching these general objectives requires the realisation of the following more specific policy objectives:

(1) Ensure that the systemic risks of the SFT markets are adequately monitored;

(2) Ensure that SFTs profit to investors first;

(3) Limit the potential risks for clients and counterparties linked to rehypothecation.

The specific objectives listed above require the attainment of the following operational objective:

(1) Make frequent and granular information on SFT markets available to regulatory authorities;

(2) Increase the transparency toward the fund investors over the use of SFTs and other financing structures;

(3) Reduce the uncertainty about the extent to which assets have been rehypothecated.

6. Policy options

In order to meet the first operational objective, the Commission’s services have analysed different policy options, such as relying on existing initiatives, improving the scope and frequency of existing market surveys, enhancing regulatory reporting and requiring a SFT reporting to a trade repository.

In order to meet the second operational objective, the Commission’s services have analysed policy options related to the implementation of disclosure requirements in various periodical reports that UCITS and AIFs have to produce as well as in the pre-contractual documents such as the fund prospectus.

In order to meet the third operational objective, the Commission’s services have analysed four policy options, ranging from no action at EU level, to contractual transparency and to the introduction of a rehypothecation cap.

7. Stakeholder consultation

The FSB conducted a public consultation[266] in November 2012 on relevant problems in SFT markets, inter alia, the lack of transparency. There was broad support for more transparency in the securities lending and repo markets, while many respondents suggested taking into account existing reporting requirements and other available market data. The contributions received in response to the Commission's public consultation on shadow banking and the European Parliament's own-initiative-report also highlight the importance of appropriate measures in this area.[267]

A public consultation on different UCITS issues was conducted in 2012 and stakeholders were notably asked about the need to increase the transparency requirements. Many industry stakeholders felt that for UCITS funds the transparency issues are adequately addressed by the ESMA guidelines. Many however, also stated that a codification of such transparency rules would facilitate their harmonized implementation.

8. analysis of impact of different policy options

Operational objective 1: options aimed at reporting frequent and granular information to regulatory authorities

Option 1.1 Rely on existing initiatives at national level or from the industry side (no action)

This option relies on existing initiatives as well as possible actions at national or industry level. Although some initiatives such as ICMA’s semi-annual survey on EU repo market, the ECB’s annual euro money market survey or the survey on credit terms and conditions in euro-denominated securities financing markets by the Committee on the Global Financial System and the ESCB, have led to improved SFT transparency, there are still important gaps in the reported indicators and data continuity. These gaps have been identified by the FSB and ESRB, both of which call for a more comprehensive data collection on SFTs.

Indeed, the current initiatives are not broad enough in their coverage to allow for effective monitoring of the systemic risk linked to SFTs: ICMA's survey covers around 61 institutions in 15 European countries, the ECB's survey mainly covers transactions by a constant panel of 104 banks in euro denominated interbank repos. Because of their semi-annual or annual frequency, existing surveys lack continuous data and, in certain circumstances, the data may represent an inaccurate or even misleading source of information (e.g. outdated data, "window-dressing" discrepancies). Moreover, participation in the surveys is voluntary and there are data protection issues with certain indicators (e.g. counterparty identity) which prevent their reporting. Given these critical gaps in the data and as seen during the recent crisis, regulators' ability to understand the risks and react in stressed market conditions is significantly undermined.

Option 1.2 Improve the scope and frequency of existing market surveys

On the benefits side, this option would permit to close, at least partially, some of the gaps in the scope of the existing surveys, thus slightly increasing the transparency compared to Option 1.1. It may also increase the standardisation and streamlining of the processes used by market participants as more actors and transactions will be covered.

Using market surveys, however, makes it difficult to ensure that all relevant participants report the data needed, thus not achieving a sufficient level of transparency of the market activity. As market surveys are essentially periodic, they do not provide a continuous flow of information and increasing their frequency will ultimately increase the reporting costs for participating firms. Although the market surveys could be adapted from one period to another to include new indicators, the successive adaptations will further raise reporting costs and affect comparability of data. Finally, the use of improved market surveys will still allow for some "window dressing" and will not provide regulatory authorities with up-to-date data. Therefore, market survey cannot be used for frequent and timely monitoring purposes.

Option 1.3 Enhance regulatory reporting

Regulatory reporting is a key tool to monitor regulated entities' activities and assess the level of risks linked to SFTs (see also section 3 on regulatory context). This option has a number of benefits compared to Option 1.1 and Option 1.2. It will solve some of the existing shortcomings related to the lack of continuous data, the data protection issues and the voluntary nature of market surveys since a frequent standardised reporting would be required by law. This option, however, will only cover supervised entities and the extent of its scope will mostly depend on the ability of regulatory authorities to obtain information from non-supervised entities. If not all market participants are captured, the reporting coverage will be incomplete. In any case, the market coverage would be higher than with market surveys. This means that the overall level of transparency would be significantly improved compared to Options 1.1 and 1.2.

There are, however, a number of disadvantages as this option will lead to the fragmentation of reported information among regulatory authorities. This problem would be particularly acute for market regulators and systemic risk regulators, but could also prove significant for prudential regulators of groups of financial entities active in multiple jurisdictions. In addition, regulatory authorities will need to increase their resources to deal with the reported information which would mean a cost increase for the authorities concerned. Moreover, regulated firms will also face increased reporting costs due to the required granularity and higher frequency of the data reporting but these costs are expected to decrease over time because of on-going automation and standardisation of reporting processes and templates.

Option 1.4 Require SFTs reporting to trade repositories

The reporting to trade repositories (TRs) will lead to a substantial increase in the transparency of securities financing markets. Since the information will be collected in a central database, this will facilitate regulators' access to the data and avoid the need to compile individual information from different regulators. It would allow for complete and timely information to be reported (e.g. principal amount, currency, type and value of collateral, the repo rate or lending fee, counterparty, haircut, value date, maturity date), therefore making it possible for regulators to perform a well-timed comprehensive monitoring of the market developments, which is not the case for Options 1.1, 1.2 and 1.3. This option will completely close the current data gaps and the reporting obligation will cover all market participants, regulated or unregulated. The periodic publication of aggregate data by TRs can be an additional benefit as it will improve the overall data available to investors but also for research projects.

There could be one-off investment costs of creating SFTs trade repositories which could be owned and run either by private entities (e.g. existing TRs) or public bodies. These costs could be, however, minimised by using existing structures such as registered TRs, matching facilities, tri-party agents, central counterparties. In terms of operating costs, market participants will incur the cost of handling the SFTs reporting processes to the TRs as well as the fees to the TRs for services provided. This option will therefore incur higher total costs than the ones under Options 1.1 and 1.2 but costs can be reduced to a certain extent by leveraging experience and facilities created through the existing obligation to report derivative instruments to TRs under EMIR. The experience from derivatives reporting shows that the fees might not be particularly high and mainly depend on the biggest firms which will report large numbers of transactions to TRs, thus allowing for economies of scale. It is expected that the Member States with the biggest SFTs markets will be more affected by these measures but they will benefit, as well as the entire Union, from greater monitoring of the systemic risks linked to SFTs markets. This option will however allow for less flexibility to modify the reporting content and it will incur higher costs than in options 1.1 and 1.2.

The overall costs of this measure can be mitigated by allowing smaller market participants to delegate the reporting to their counterparties or third parties. In most cases the delegation will be to bigger institutions with whom small market participants usually have SFTs, and who are better placed (e.g. scale economies) to bear the reporting costs. In order to allow for an accurate reporting, a certain level of standardisation will also be needed, which can be provided by level 2 measures, building on existing initiatives and studies (e.g. EMIR, FSB).

Option 1.5 Require reporting to trade repositories or, if that is not possible, directly to regulators

This option takes into account the possibility that a trade repository may not exist or, in case it does exist, that it may either not be willing or not be able to accept the information reported by the counterparties. This means that, if it is impossible to report SFTs to a trade repository, then the counterparties should report them directly to the relevant regulator.

The cost-benefit analysis for this option is largely identical to the one for the previous option on requiring SFTs reporting to TRs. There will, however, be additional costs for regulators (e.g. ESMA, national competent authorities) to handle the information reported but, at the same time, market participants would not have to pay fees to the services of the TRs. The key additional benefit of this option is that it ensures the effective reporting of SFTs in any event, thus guaranteeing that regulators acquire a comprehensive picture of SFTs markets.

Summary

The preferred policy option that has been chosen is to request that all SFTs are reported to a trade repository, or, if that is not possible, directly to regulators. This option is the most efficient in order to allow early detection of risks building up in the SFT market as it allows gathering data with a higher level of granularity and frequency.

This option is therefore the most indicated to answer to the objective of this regulation, i.e. ensuring financial stability by preventing the build-up of systemic risks in SFT markets.

Each option is rated between "---" (very negative), ≈ (neutral) and "+++" (very positive) based on the analysis in the previous sections. The benefits are, however, not quantified in monetary terms, as this is not possible on an ex ante basis. The costs should be understood in a broad sense, not only as compliance costs but also as all the other negative impacts on stakeholders and on the market. This is why we have assessed the options based on the respective ratio of costs to benefits in relative terms. The assessment highlights the policy options which are best placed to reach the related objectives.

Policy options || Impact on stakeholders || Effectiveness || Efficiency

1.1 No action || 0 || 0 || 0

1.2 Improve the scope and frequency of existing market surveys || (+) Some reporting gaps of existing surveys could be closed, providing regulatory authorities with more  market information (+) Market participants would not be subjected to reporting obligations   ||  (-) No continuous flow of information. (--) Sufficient coverage (in terms of number/significance of market participants) cannot be guaranteed (-) Scope for 'window dressing'. (-) Risk of insufficient standardisation || (--) Regulatory authorities will not avail of comprehensive up-to-date data matching their requirements for frequent and timely monitoring

1.3 Enhance regulatory reporting || (++) Regulatory authorities would obtain continuous data from a wider pool of market participants (-) Increased costs for regulatory authorities (-) (Only) regulated entities have to provide information and bear the reporting costs. || (++) Frequent and standardised data flow (-) Coverage limited to supervised entities. (-) Fragmentation of information among regulatory authorities. || (+) Improved data availability counterbalanced by limitations of coverage of entities, fragmentation of data among authorities and cost increases for authorities and regulated firms.

1.4 Require SFTs reporting to trade repositories || (+++) Regulatory authorities obtain complete and timely market information (--) Market participants have to bear reporting costs. (+) Investors and other stakeholders benefit from the periodic publication of aggregate data  || (+++) Complete closure of current data gaps  (+) Easy access for regulators || (++) Regulators are enabled to perform well-timed comprehensive market monitoring; there are possibilities to minimise investment costs and reporting costs for small market participants

1.5 Require reporting to trade repositories or, if that is not possible, directly to regulators || (+++) Impact as above in 1.4, but (-) where reporting to trade repositories would not be possible, additional costs for regulators from handling the information reported directly to them. || (+++) As above in 1.4, additionally: (+++) Effective reporting ensured, where reporting to trade repository is not possible || (+++) As above in 1.4., with the additional key benefit that effective reporting is ensured under all circumstances.

Operational objective 2: options aimed at increasing the transparency toward fund investors

Option 2.1 Rely on existing transparency requirements[268]

Under this option, the initiative would be left to National Competent Authorities (NCA) of the Member States and to the European Securities and Markets Authority (ESMA) to create transparency in the SFT area. NCAs under the umbrella of ESMA have already taken action in this field with the entry into force of guidelines on UCITS funds. Those guidelines require the manager to publish in the UCITS annual report information on the exposures; identity of counterparties as well as type and amount of collateral for SFTs. In addition managers are required to publish the revenues and the costs related to the use of securities lending and repos. These reporting obligations come on top of what is required by the UCITS directive, notably by the requirements contained in the annexes of the directive.

The reporting requirements toward investors are much less developed in the AIFMD context. The directive itself contains few specifications on SFTs whereas no additional guidelines have been developed up to date. Under the AIFMD the competent authorities have access to a larger amount of information on SFTs than investors.

Impact on investors: The existing UCITS requirements give some general information to the investors but are not granular enough to be able to give a detailed picture. For example the investor has no information on the amount of securities subject to SFTs as a proportion of the fund’s Assets under Management (AuM). As such investors are not able to assess the degree of involvement of the fund in this activity and therefore cannot judge the degree of risk that this fund entails. Other information is missing such as concentration data on the collateral or data on the re-use and re-hypothecation of the collateral. The revenue and costs of those transactions are disclosed indistinctively as a gross amount; it is therefore impossible to analyse the breakdown for each type of activity and the source of the costs. The comparability between different investment funds is therefore not optimal.

This issue is further reinforced by the absence of SFT reporting for AIFs. The AIFM directive contains several data that fund managers should disclose to their investors but almost none of them cover the use of SFTs. There is a risk that national legislations diverge in that respect and thus that investor protection standards diverge between Member States.

Impact on managers: The absence of an EU coordinated approach entails the risks that some Member States decide to implement SFT reporting requirements individually and thus in a diverging manner. Fund managers operating cross-border would then need to apply different rules which would potentially increase their reporting costs. More generally ESMA guidelines do not benefit from the same enforcement quality as primary EU legislation because Member States have the option not to apply ESMA guidelines.

The respondents to the UCITS consultation supported the initiative taken by ESMA to increase the transparency of SFTs. The majority of them, being from the industry side or the public side, considers that the ESMA guidelines are in this regard sufficient to address the issue of transparency. Other respondents consider that some legal codification (e.g. through a legal initiative or technical standards) would be necessary in order to ensure a harmonized implementation in the EU.

Option 2.2 Incorporate SFTs and equivalent financing structures reporting into existing ex-post documentation, such as periodical reports required under the UCITS directive or the AIFMD

According to the UCITS Directive, UCITS funds have to produce annual reports containing different financial statements as well as information on the different investments undertaken by the fund. UCITS funds must also produce half-yearly reports containing information on assets and liabilities and more generally on the composition of its portfolio. According to the AIFM Directive, information on AIFs should be disclosed to investors at least on an annual basis or more frequently according to the fund rules. The existing UCITS and AIF reports already do provide some information on the use of SFTs and could be easily used to incorporate also more detailed regular reporting on the use of SFTs. The data to be reported would correspond to the list proposed by the FSB, such as global data, concentration data, counterparty disclosure or policy on acceptable types of collateral. To increase the awareness of the investor to the revenue that is generated by the SFT and equivalent activity, the reports will have to include detailed information on the costs and returns. In order to ensure complete information of the investors, these information requirements should cover financing structures equivalent to SFTs.

Impact on investors: Ex-post documentation gives insight to the investors on the transactions that the fund has been involved in over the previous reporting period. It is a means for the investors to check the performance of the fund and other indicators regarding the risks or costs. More generally, it gives the possibility to verify that the fund's investment strategy has evolved as announced in the prospectus. Including more detailed data on SFTs, presented in a structured way will enable investors to gain a more comprehensive understanding of these transactions and in particular their implications on the fund risk and reward profile. This will enable the investors to understand whether these transactions create a value added to them and will facilitate the comparison with other similar investment funds. Investors will also be able to identify the proportion of SFT revenue that is returned to the fund and assess the amount of costs related to this activity. In practice it is to be expected that the institutional investors will most probably be interested to have access to this information whereas doubts exist regarding the capabilities of retail investors to grasp the significance of this information. This information could for example be used by institutional investors such as pension funds or insurance companies as part of their due diligence activity when selecting investment funds. These investors could then screen and compare the different investment fund targets at the light of new criteria linked to the SFT activity.

Impact on managers: SFTs are currently labelled by fund managers as an ancillary activity. SFT are not revealed in detailed mostly to avoid disclosing their inherent risks (default of a SFT counterparty, insufficient collateral) but also the pecuniary benefit that these transactions entail for the manager. By introducing enhanced reporting requirements, SFT would become more visible and would de facto be assimilated to the fund's investment strategy. Managers will have to incur some costs for computing this information and passing it on to the investors. Some stakeholders that responded to the FSB consultation[269] warned that the additional disclosure will increase the reporting costs which ultimately will increase the fees that investors pay. The costs should not however be overestimated since the additional data will be reported through existing UCITS and AIF reports. In addition part of those data already has to be disclosed under existing rules applying to UCITS and AIFs.

For the funds that are not active in SFTs, the enhanced reporting requirements will have no impacts, whereas the reporting costs will generally increase with the degree of use of SFTs. However, funds that use SFTs very actively usually do so in a standardised manner – in the case of securities lending with agent lenders – for whom reporting is a relatively straightforward task; accordingly recourse to agent lenders will decrease the cost of providing this information. The disclosure of revenue and costs attached to the SFT activity may have an influence on the manager's remuneration. The transparency to investors might increase the pressure on the manager to rebate a larger part of the revenue from SFTs to the investors.

Generally, the industry stakeholders that responded to the FSB consultation and to the Commission consultation on UCITS issues supported the goal of increased disclosure to fund investors. Some stakeholders expressed that transparency standard of SFTs should be harmonised upwards in order to inform investors in an efficient, standardised and comparable way. Within the UCITS context, many stakeholders expressed that there would be merit in making the ESMA guideline requirements mandatory to facilitate their harmonized implementation.

The FSB consultation respondents questioned the usefulness of detailed disclosure on SFTs as they argued that such detailed information may rather confuse than inform retail investors. It should however be recalled that UCITS funds are mostly purchased by institutional investors. It is estimated that around 90% of the UCITS funds AuM are held by institutional investors. Those professional investors have the capability to understand the impacts (such as on the counterparty risk or the liquidity risk) of using SFTs and as indicated previously this new information will be useful in selecting a fund.

In the case of UCITS funds, they did not see the need of such detailed information as SFTs cannot be used to increase fund leverage due to regulatory limitations. Even though the UCITS directive and the ESMA guidelines pose certain limits to the leverage of UCITS funds, there are other major risks created by these measures – in particular counterparty risk - that make disclosure of such measures necessary.

Finally the argument that disclosing SFT information risks providing confidential information to competitors about the positions of the fund has little merit since investment funds are already obliged to disclose on a regular basis all investments and positions they take directly in the market.

Option 2.3 Implement SFT and equivalent financing structures reporting through ex-ante documentation, in the prospectus or equivalent AIF report according to the article 23 of the AIFM directive

Every investment fund, being a UCITS fund or an AIF, is required to produce a prospectus setting out the fund rules and the rules of incorporation. The fund rules usually contain all the information related to the investment strategy that the fund intends to pursue. Those fund rules represent the “contractual obligation” of the fund manager towards the investor. Once the fund is set-up according to those rules, the manager is not allowed to deviate from them. The supervisory authorities, with the help of the depositary of the fund, are responsible to control that managers are acting according to the predefined rules. Under this option, managers would be required to also include the use of SFT as part of the investment strategy they intend to pursue. This could include for example information on the total amount of assets that can be on loan at any point in time, the reasons and goals behind the use of SFTs, the policy of the manager regarding the valuation and management of collateral that is exchanged as part of an SFT, including its re-use or re-investment, the policy regarding the quality and identity of the counterparties and how the revenues and the costs related to those transactions will be shared.

Impact on investors: Investors would have knowledge, prior to their investment, on whether SFTs form part of the investment strategy pursued by a fund. They will be able to measure the expected risk and reward profile linked to this activity. Their ability to compare the investment proposition of different investment funds will increase. In addition, investors will receive increased assurance that managers will not use to a greater extent than announced in the fund rules. In the case that managers lose money because their SFT transactions exceed the pre-announced limits, investors would be able to invoke the manager's liability for breach of contract. Enhanced disclosure of SFT will also mitigate potential conflict of interests between the fund manager and the fund investors because investors will be reluctant to invest in funds where their interests are neglected.

Impact on managers: Stating in the fund rules the planned activity related to SFTs will restrict the manager’s flexibility in using those techniques. Managers will be bound by the predefined strategy enacted in the fund rules. For example they could be limited in their use of repo which could in turn have an effect on the degree of leverage that managers are permitted to use. In practice if the manager wants to change the rules on the use of SFTs, the manager will be first obliged to seek the agreement of the investors. Therefore the manager will no longer have the entire discretion over the use of SFTs. Since the prospectus or other equivalent pre-contractual documents have to be produced only once at the creation of the fund, the introduction of SFT information will not increase the reporting costs. The costs of setting up a prospectus could be slightly increased but this should be seen in conjunction with all the other information that has to figure in the document. As such the cost impacts will be marginal.

Generally, the industry stakeholders felt that fund prospectus should inform the investor in the most objective, transparent and impartial way. A major industry association also suggested that a standard template for securities lending disclosures should be developed for UCITS to incorporate into the prospectus and with a requirement that anything outside of these parameters should be disclosed separately.

Option 2.4 Implement specific reporting requirements for retail investors

Under the Packaged Retail Investment Products (PRIPS) initiative[270], fund managers will have to produce a short document aimed at retail investors only. Irrespective of being a structured product, insurance product or an investment fund, investors will have the same information about the products. The information should be presented in a clear, simple and comparable manner so that all retail investors could understand it. Under this option the Key Information Document (KID) for investment funds would contain additional information on the use of SFTs.

Impact on investors: Retail investors would have access to the SFT information through a pre-contractual document that is easy to access and to read. This would raise their awareness that SFTs exist and that they increase the riskiness of the fund. It would however be challenging to present the main risk and reward profile linked to SFTs in a KID that is supposed not to be longer than 2-3 pages and that should already include all the other information related to the fund. The information would need to be reported in a very concise way that would not help the retail investor to understand it. Moreover it could end up being misleading should the information be reported in a too short manner. In addition there is a risk that the core objective of the PRIPS initiative to make all packaged investment products comparable is not reached because SFT reporting would concern investment funds only and not structured products or life insurance investment products.

Impact on managers: to the extent that managers already have to produce a KID, this should not create major impacts. Because the KID is equivalent to a pre-contractual document such as the prospectus, the impacts on the manager’s flexibility would be the same as under option 2.3.

The option to introduce SFT disclosure in the KID was not directly tested in the different consultations but some stakeholders expressed doubts as to the usefulness of SFT information for retail investors. The interest of retail investors to know the details of the SFT activity performed by the investment funds is also put into question. It is therefore doubtful that a KID contained SFT information would be of any added value.

Summary

Policy options || Impact on stakeholders || Effectiveness || Efficiency

2.1 No action || 0 || 0 || 0

2.2 Implement SFT reporting through ex-post documentation || (+) investors benefit from full transparency (-) managers must incur disclosing costs || (+) transparency increases the protection of the investor (+) coherence with FSB || (++) increased transparency at little cost for the manager

2.3 Implement SFT reporting through ex-ante documentation || (+) investors benefit from pre-contractual disclosure (--) managers become fully accountable over the use of SFTs || (+) investors benefit from contractual protection in the case of misuse of SFTs (+) coherence with FSB || (+) increased protection at the price of making SFT a normal activity for the manager

2.4 Implement specific reporting requirements for retail investors || (+) retail investors benefit from preeminent SFT information (-) retail investors might lack interest and understanding (≈) little impact in terms of cost for the manager || (-) risk of unclear SFT information (-) risk of undermining comparability of the KID || (--) risk of confusion not compensated by increased benefit for retail investors

Option 1 cannot be retained as it would not address the problem of conflict of interest and unequal treatment of investors. Investors will continue to have less information than the manager, to have no influence on the fund’s activity and to be neglected in revenue sharing arrangements.

Option 2 has the merit to disclose all the SFT activity over the last reporting period so that investors can follow it. Such information will increase their awareness of the use of SFTs. Under option 3, investors have pre-contractual information on the use of SFT so that they can choose each fund according to its stating strategy. Options 2 and 3 are not superior or inferior to each other but are complementary. Option 2 is important for on-going transparency whereas option 3 is important for pre-contractual transparency. As such both options should be retained.

Option 4 has the merit to recognize that retail investors need a particular treatment in the access to the information. But the KID is not the right place to address this concern because it risks undermining the objective to make the KID simple and comparable between all packaged investment products. Moreover it is doubtful that retail investors will have any interest or understanding for this new information. As such this option cannot be retained.

In order to ensure that investors have sufficient information over the use of SFTs, it is therefore necessary to require disclosure before the investors invest in the fund as well as after the SFTs have been used. The use of SFT, its extent and the eligible counterparties should be identified in both the periodical reports and in the fund's prospectus or equivalent AIF report according to the article 23 of the AIFMD. These information requirements should apply to financing structures equivalent to SFTs in order to ensure complete information of the investors.

Operational objective 3: option aimed at reducing the uncertainty about the extent to which assets have been rehypothecated

Option 3.1 No action

Under this option no specific rules on rehypothecation would be introduced. In this case, the systemic risks related to the uncertainty about the extent to which assets have been rehypothecated would not be fully addressed throughout the EU and the contractual and operational transparency of such activities would remain low, effectively preventing counterparties from properly managing their risk exposure. This implies that when the same securities are being rehypothecated several times, the failure of a counterparty could lead to a race to secure the collateral by multiple parties, as happened in the Lehman and Bear Stearns collapses. This option would not achieve minimum harmonisation of rehypothecation rules within the Union which could allow for better transparency towards clients and engaged counterparties in their cross-border activities.

Option 3.2 Oblige contractual transparency on rehypothecation

This option consists of specific transparency requirements to be met by contractual agreements on rehypothecation as well as requiring the prior express consent to rehypothecation by the counterparty providing collateral. The FSB has developed a similar policy recommendation on sufficient disclosure to clients in relation to rehypothecation of assets.[271] All Member States also support the need for more transparent rehypothecation.[272]

This would fully meet the specific objective 3 by requiring the prior express consent of the providing counterparty to the rehypothecation of the financial instruments it has provided as collateral, ensuring that it is fully aware of the potential risks involved, in particular in the event of default of the receiving counterparty. Furthermore, according to this option, prior to the actual rehypothecation the financial instruments received as collateral have to be transferred to an account opened in the name of the receiving counterparty, which would also help prevent a future crisis scenario, where investors are uncertain about their rights, thus contributing to financial stability. Such rules are consistent with existing market practice in major securities markets in the EU (e.g. rehypothecation undertaken by prime brokers based in the UK).

The rule concerning the providing counterparty's express consent to rehypothecation would be along the lines of Article 19, MiFID Implementing Directive 2006/73/EC. This approach would be broader than MiFID as it would apply to all counterparties engaging in rehypothecation. It would also cover the providing counterparty who has only a mere contractual right to have equivalent securities returned after rehypothecation ends (MiFID protects only clients' ownership rights). Currently an investment firm may book the securities as its own on its records and, in such a case, they would not be covered by the MiFID client asset protection rules (e.g. rules that impose segregation between the client's and the firm's assets and require that the investment firm uses securities only with the client's express consent). In the MiFID review, the Commission proposed to address this by banning title transfer collateral arrangements with retail clients for the purpose of securing or covering clients' present or future, actual, contingent or prospective obligation.[273] The Alternative Investment Management Association, which represents global hedge funds community, agrees with a prior consent rule as they want to be able to authorise counterparties to rehypothecate their assets in some well-defined circumstances.[274]

In terms of effectiveness, the proposed option would also ensure that counterparties are enabled to fully manage their exposure. By clarifying that express consent is needed for rehypothecation to take place, the proposed option would prevent non-authorised rehypothecation of assets. The requirement that financial instruments received as collateral have to be transferred to an account opened in the name of the receiving counterparty prior to rehypothecation would supplement the Financial Collateral Directive and is analogous to the FSB Recommendation 7. This would not only rebalance the position of collateral providers, but would reduce systemic risk by preventing excessive rehypothecation and make the rehypothecation chain transparent, thus contributing to the general objective to ensure financial stability.

Option 3.3 Introduce a rehypothecation cap

A cap on the level of rehypothecation would constitute a clear direct restriction on rehypothecation. Such a cap could be introduced along the lines of the cap in the US[275] where a limit on the maximum leverage and the amount of collateral that could be rehypothecated could be introduced, e.g., intermediaries could only be allowed to: (1) lend 50% of the purchase price of securities to be deposited as collateral with the respective intermediary; and (2) to rehypothecate clients’ securities they hold as collateral up to 140% of the value of the client's liabilities towards the intermediary.

This would prevent excessive levels of rehypothecation by setting it at a constant level and, thus, reduce the uncertainty about the extent to which assets have been rehypothecated . A transparent rehypothecation cap that is constant would, in addition, allow authorities and regulators to monitor the level of endogenous liquidity that is generated by market practices. It would also provide a level playing field within the EU, thus improving the functioning of the Single Market, and globally, notably with the US. The alignment with the US 140% rule would close the regulatory gap between the USA and the EU that leaves market participants the opportunity for arbitrage. For example, when Lehman Brothers went bankrupt many US hedge funds found themselves with significant exposure to Lehman Brother International Europe as their prime brokerage agreements were structured to permit client-asset transfer to the prime brokerage's UK subsidiary.

In terms of fundamental rights, this option could have a negative impact on the right to property (Article 17 Charter of Fundamental Rights), as the rehypothecation cap would also cover collateral provided under title transfer arrangements. In this situation, the collateral taker becomes the securities owner for the duration of repo/securities lending transaction and as such the collateral taker is entitled to dispose of its ownership right, which the rehypothecation cap would limit. The Alternative Investment Management Association considers that it would be inappropriate to introduce harmonised rules which set a limit on rehypothecation. [276] However, individual rights would have to be considered against the cumulative impact of the collective behaviour of market participants if a majority engages in rehypothecation, leading to a build-up of hidden leverage and generating chains of contagion in the system that can threaten the overall economy. In these circumstances, it may be justified to impose limits.

This option also has a potential negative macro-economic impact. Despite the fact that a constant maximum would prevent rehypothecation from being pro-cyclical and prevent future liquidity bubbles being created by this market practice, it could lead to a reduction in an important source of liquidity and funding for financial intermediaries. The reduction of the liquidity available in the EU financial system in terms of collateral available from 2007 to 2011 was the result of two different dynamics: (1) a reduction in high quality collateral due to the EU debt crisis; and (2) a reduction of the velocity of collateral i.e. the number of times the same collateral has been reused in the system. A rehypothecation cap could negatively impact the repo and securities lending markets, which are an important tool for funding, collateral management and secured lending and could adversely affect economic growth. Finding a sustainable velocity rate wold require better data on levels of rehypothecation in the market. In the current absence of this data, retaining this option would lead to an arbitrary cap being chosen and could impact the markets and economy disproportionately. Therefore this option should not be retained at present.

Option 3.4 Introduce a duty for intermediaries to offer a contractual rehypothecation cap

This option consists of allowing rehypothecation for assets taken as collateral for the value of the client's or counterparties actual obligation plus a reasonable haircut contractually agreed by the parties. This option would not align EU rules with those of the US but it might give the parties the ability to negotiate a haircut equivalent to the 40% one used in the US. By prescribing that the parties could not overrule the obligation to set a cap by contract, transparency around rehypothecation and thus client asset protection would be enhanced. As discussions with stakeholders have shown, such contractual rehypothecation arrangements are likely to be ineffective in reducing the uncertainty of rehypothecation as a contractual rehypothecation cap is common market practice and usually oscillates between 100% and 150% of client's indebtedness.  This option would therefore have a very limited impact.

Summary

The proposed policy option is option 3.2, i.e. specific transparency requirements to be met by contractual agreements on rehypothecation, prior express consent by the counterparty providing collateral and a requirement to transfer financial instruments received as collateral to an account opened in the name of the receiving counterparty.

This approach is the most efficient and proportionate in order to make counterparties fully aware of the potential risk exposure of their collateralised assets, to enable them to fully manage risk exposure and make efficient use of their assets. This option is therefore the most appropriate to answer to the objective of this regulation.

Policy options || Impact on stakeholders || Effectiveness || Efficiency

3.1 No action || 0 || 0 || 0

3.2 Oblige contractual transparency on rehypothecation || (++) Collateral  providers are aware of their exposure in case of rehypothecation (++)  Counterparties benefit from greater transparency || (+) Consent rule helps prevent non-authorised rehypothecation (+) Transparency enables counterparties to fully manage their risks (+) Coherence with FSB and Financial Collateral Directive || (++) Enhanced  protection of collateral provided and limitation of opaque collateral chains.

3.3 Introduce a rehypothecation cap || (-) Intermediaries' scope for providing collateralised loans and for rehypothecating collateral obtained is limited || (+) A constant cap would effectively limit rehypothecation (+) The regulatory gap with the US would be closed and a level playing field would be provided. (-) The introduction of a cap limits fundamental rights (property rights) (--) The cap could adversely affect macro-economic liquidity and funding, if not set at the right level. || (-) In the absence of reliable data, at present the benefits are outweighed by possible risks to economic growth.

Option 3.4 Introduce a duty for intermediaries to offer a contractual rehypothecation cap || (+) The parties are enabled to negotiate a haircut on collateral || (≈)  Effect depends on result of negotiations. || (≈) Very limited impact

9. The retained policy options and their impacts

The retained policy options

In order to increase the transparency over the use of SFTs, a combination of different measures is necessary.

In order to enhance the supervisory monitoring over the use of SFTs, each counterparty will have to report SFTs to a trade repository (option 1.5). The use of SFTs raises also questions linked to investor protection standards. Investors are often unaware of the use that is made of these techniques and even less of the risks that are created. In order to increase the awareness of the investors in investment funds, the fund reporting will have to be enhanced: at the level of the periodic reporting and at the level of the pre-contractual documents (options 2.2 and 2.3). To complete the policy toolkit, it is also necessary to implement new rules on rehypothecation. These new rules should ensure that all counterparties providing collateral have given their prior express consent to rehypothecation of the financial instruments concerned and that these financial instruments are transferred to an account opened in the name of the receiving counterparty before rehypothecation can take place (option 3).

All retained options taken together will ensure that the shadow banking activity of using SFTs is properly supervised and regulated. The SFT practice will not be prohibited nor limited by specific restrictions but be more transparent. As such the retained options are not expected to create structural impacts on the SFT market. The retained options will increase the reporting costs for the counterparties but this increase will be outweighed by the benefits of having greater transparency for the competent authorities, clients, investors and society at large.

Impact on SMEs

The requirement to report SFTs to trade repositories is not expected to have any impact on SMEs as they do not participate in SFT markets. The indirect costs related to additional reporting costs for financial companies would be negligible.

The strengthening of the provisions to better deal with the risks of investment funds will increase investor protection standards for all investors, including SMEs. SMEs, as retail investors and other corporates of larger size, may use investment funds to realize certain placements. Enhancing the transparency requirements could help SMEs to select investment funds that minimize their use of SFTs and therefore their overall risk. As such the risks that SMEs face in investing in investment funds could decrease.

SME-clients of financial intermediaries will benefit from the rules on express consent and increased transparency to be met by rehypothecation agreements, which increase the protection of the financial instruments they may provide as collateral.

Social impact

To the extent that the proposed policies will help contain the effects of future financial crises on the real economy, they will also help reduce the social costs of those crises (e.g. unemployment).

Regarding the impacts on the asset management sector’s employment, should the assets under management be maintained at current levels, no further impact would be expected.

Environmental impact

Nothing would suggest that the proposed policy will have any direct or indirect impacts on environmental issues.

Impact on Member States

The reporting of SFTs to trade repositories is expected to be similar to the existing reporting framework for trade repositories for OTC derivatives under EMIR. This would limit the impact to Member States, national competent authorities and ESMA.

The creation of new rules will require the national competent authorities (NCA) of the Member States to check their implementation. As some of those rules are already applied through the ESMA guidelines, this should not require any additional substantial work regarding the UCITS funds. Additional supervision work will however be needed as regards the AIFs but no material impact is expected since those funds are already under close scrutiny through the application of the AIFM Directive and its reporting requirements.

Some additional burdens might impact ESMA that could be required to harmonize the supervision process. ESMA will also have to be involved in the usual complaint resolution that arises in the application of single market law. Regarding the opinion of Member States on possible issues on compliance with any new requirement, no specific views have yet been expressed.

The rehypothecation requirements would not significantly impact Member States. National competent authorities would be required to enforce these rules at national level.

Impact on third countries

No impacts on third countries are expected for the obligation to report SFTs to trade repository. Only entities based in the EU would be subject to this obligation.

The new reporting requirements in the asset management sector will have to be implemented by managers domiciled in third countries when they market or manage AIFs in the EU. The AIFM directive has introduced the same reporting requirements to all AIFs that are marketed in the EU territory so the third country managers have to respect the EU rules when marketing non-EU AIFs in the EU. This principle will be the same as regards the new SFT reporting requirements. Because UCITS funds are domiciled and managed in the EU, no impact is expected for third countries.

The rehypothecation requirements would cover also counterparties established in third countries, when they rehypothecate collateral provided by an EU entity. Therefore, third country counterparties would have to respect these rules.

10. Monitoring and Evaluation

Ex-post evaluation of all new legislative measures is a priority for the Commission. Evaluations are planned about 4 years after the implementation deadline of each measure. The forthcoming legislation will also be subject to a complete evaluation in order to assess, among other things, how effective and efficient it has been in terms of achieving the objectives presented in this report and to decide whether new measures or amendments are needed.

The following indicators can be used to monitor the first and third specific objectives on increased transparency of SFTs towards regulatory authorities and reduced uncertainty on rehypothecation: [1] size of different segments of SFT markets, level of interconnectedness and market concentration, average maturity of SFTs and leverage; and [2] size of rehypothecation activities and collateral velocity. It is important to note that data on rehypothecation encompasses SFTs and other collateral-based activities such as collateral provision for derivative contracts.

 In terms of indicators and sources of information that could be used to monitor the second objective of increasing fund’s transparency, data collected by the NCAs as part of the authorisation process and their ongoing supervision task can be used. It is also possible to access directly the different periodical reports and prospectuses on internet. As such a sample of different funds can be assembled to perform a detailed monitoring exercise. This analysis could assess how the funds are communicating to their investors and how important is their use of SFTs.

As regards the international dimension of the policy measures, the FSB plans to conduct a peer review of the implementation of their recommendations in the different jurisdictions. The European Commission will closely monitor this review in order to ensure that the recommendations have been evenly applied by all G20 Member States.

11. Additional information

Overview of existing reporting requirements

The below table provides a general overview of the different reporting obligations that exist in the financial legislation on SFTs and equivalent measures.

|| Reporting to competent authorities || Reporting to investors || Reporting to trade repositories

CRR Credit institutions || Scope: SFTs. Data elements and frequency: aggregate data and annual or semi-annual or quarterly frequency. || ||

MIFID Investment firms || Scope: any transaction in financial instruments, except SFTs. Data elements and frequency: highest level of granularity and frequency. || ||

EMIR Counterparty of a transaction. || || || Scope: any OTC derivative transaction, including total return swaps. Data elements and frequency: highest level of granularity and frequency.

UCITS Mutual funds || Equivalent as investor reporting except for the use of derivatives where the types, risks and applicable limits must be reported || Annual and semi-annual reports Scope: general scope including broadly derivatives and other techniques such as SFT, no granular disclosure Data elements (included in ESMA guidelines): exposure, identity of counterparties, collateral, revenue Prospectus Scope: same as above Data elements (included in ESMA guidelines): intention to use them, revenue agreement, risk description ||

AIFMD Alternative investment funds || Scope: covers partially some SFTs and some other financing transactions Data elements: information on sources of leverage, including names of counterparties, value of collateral or reuse of assets Data frequency: depends on the size and varies from quarterly to annually || Annual report Scope: securities lending Data elements: exposure, revenue Prospectus Scope: all assets and techniques Data elements: collateral and reuse arrangement, risk description ||

Detailed overview of existing and proposed fund reporting requirements

UCITS directive

The provisions regarding the reporting to investors over the use of SFTs and other financing structures are scattered in different places in the UCITS directive and in the ESMA guidelines on ETFs and other UCITS issues.

Neither SFTs nor TRS are defined in the UCITS directive. What generally covers securities lending and repos is the broad concept of Efficient Portfolio Management (EPM) technique. EPM techniques are defined according to their objectives: reduction of risk, reduction of costs or generation of additional income (Art. 11 Eligible Asset Directive). This implies that the scope of EPM techniques is flexible and depends on the interpretation that every manager makes of the concept. In practice it is however commonly admitted that securities lending and repos form an integral part of the EPM techniques. Practices that are economically equivalent to securities lending or repos, the other financing structures, can be considered as well as EPM techniques. This includes for example the liquidity swap (change the liquidity profile of the fund) or the collateral swap (exchange of fund’s assets).

UCITS funds may also invest in all kind of Financial Derivative Instruments (FDI) that have equivalent characteristics as SFTs. These instruments have in common that they are not used for the primary purpose of investing the assets of the funds but for pursuing some other objectives. For example in the case of Total Return Swap (TRS) the objective is to get exposure to strategies that would be difficult or too costly to implement. For that reason certain FDI could also fall under the scope of “other financing structures”.

The new reporting requirements that are proposed are indicated in italic in the below table.

Periodic report || Propesctus

Exiting requirements UCITS Directive, Annex I Schedule B, Point VII: Details, by category of transaction within the meaning of Article 51 carried out by the UCITS during the reference period, of the resulting amount of commitments. ESMA guidelines: · the exposure obtained through EPM · the identity of the counterparty(ies) · the type and amount of collateral · the revenues together with the direct and indirect operational costs and fees incurred. Proposed add-ons Specification of the details that have to be reported. This will include the information contained in ESMA guidelines plus all the following information: · global data: amount of securities on loan and total amount of assets engaged in each type of SFT · concentration data: top 10 counterparties for each type of SFT · aggregate transaction data: type and quality of collateral, maturity tenor, currency of the collateral, country of domicile, settlement and clearing (tri-party, central counterparty, bilateral) · data on re-use: share of collateral that is re-used, cash collateral re-investments, information on any restrictions · safekeeping methods: number of custodians, legal chain (segregated accounts or pooled accounts) · data on returns and costs: breakdown between the fund manager, the investors and the agent lender In addition these provisions will cover equally all transactions that have the same economic profile, namely the SFTs and the other financing structures such as the TRS, the collateral or the liquidity swaps. || Exiting requirements UCITS Directive, Annex I Schedule A, Point 1.15: Indication of any techniques and instruments or borrowing powers which may be used ESMA guidelines: · Information of investors over the intention to use EPM, including a description of the risks (counterparty risk, conflict of interest) and the impacts on the UCITS performance · Disclosure of the policy regarding the costs/fees that may be deducted from the revenues and the identities of the entities to which the costs/fees are paid · information on the underlying strategy and composition of the investment portfolio or index in the case of TRS Proposed add-ons Specification of the details that have to be introduced in the prospectus. The prospectus is a pre-contractual document that is binding to the manager. Therefore it gives confidence to the investors that the fund will respect the pre-defined investment limits. In that regard the idea is to treat SFTs and other equivalent structures as a normal investment activity in the prospectus: · General description of the SFTs and the rationale for their use · Reporting on the max proportion of the portfolio that can be subject to such techniques and the types of assets that can be subject to it · Criteria used to select counterparties · Policy on collateral valuation and safekeeping · Description of the incurred risks, including risks linked to collateral management, operational, liquidity, counterparty custody, legal risks · Revenue sharing agreement between fund manager, investors and third parties

AIFM directive

As under UCITS, neither SFTs nor other financing structures are defined, directly or indirectly, in the AIFM directive. General provisions are set out in the main directive and in the delegated regulation. AIFs are otherwise subject to national rules.

Periodic report || Propesctus

Exiting requirements AIFM delegated regulation, Art. 104 · Amounts due to counterparties for collateral on return of securities loaned · Fee income from securities loaned Proposed add-ons Same information as UCITS || Exiting requirements AIFM Directive, Art. 23 Types of assets, techniques, associated risks, applicable investment restrictions, sources of leverage, collateral and asset reuse arrangement Proposed add-ons Same information as UCITS

FSB recommendations

|| FSB Recommendations || EU response

1 || Authorities should collect more granular data on securities lending and repo exposures amongst large international financial institutions with high urgency. Such efforts should to the maximum possible extent leverage existing international initiatives such as the FSB Data Gaps Initiative, taking into account the enhancements suggested in this document. || Creation of a Trade Repository

2 || Trade-level (flow) data and regular snapshots of outstanding balances (position/stock data) for repo markets should be collected. Regular snapshots of outstanding balances should also be collected for securities lending markets and further work should be carried out on the practicality and meaningfulness of collecting trade-level data. Such data should be collected frequently and with a high level of granularity, and should also capitalise on opportunities to leverage existing data collection infrastructure that resides in clearing agents, central securities depositories (CSDs) and/or central counterparties (CCPs). National/regional authorities should decide the most appropriate way to collect such data, depending on their market structure, and building on existing data collection processes and market infrastructure where appropriate. Trade repositories are likely to be an effective way to collect comprehensive repo and securities lending market data. Regulatory reporting may also be a viable alternative approach. || Creation of a Trade Repository

3 || The total national/regional data for both repos and securities lending on a monthly basis should be aggregated by the FSB which will provide global trends of securities financing markets (e.g. market size, collateral composition, haircuts, tenors). The FSB should set standards and processes for data collection and aggregation at the global level to ensure consistent data collection by national/regional authorities and to minimise double-counting at the global level. || To be implemented by the FSB

4 || The Enhanced Disclosure Task Force (EDTF) should work to improve public disclosure for financial institutions’ securities lending, repo and wider collateral management activities, taking into consideration the items noted above. || To be followed by the EDTF

5 || Authorities should review reporting requirements for fund managers to end-investors against the FSB’s proposal, and consider whether any gaps need to be addressed. || To be implemented now within the investment fund frameworks set out in the UCITS and AIFM Directives

6 || Regulatory authorities for non-bank entities that engage in securities lending (including securities lenders and their agents) should implement regulatory regimes meeting the minimum standards for cash collateral reinvestment in their jurisdictions to limit liquidity risks arising from such activities. || To be followed later following more assessment

7 || Authorities should ensure that regulations governing re-hypothecation of client assets address the following principles: • Financial intermediaries should provide sufficient disclosure to clients in relation to re-hypothecation of assets so that clients can understand their exposures in the event of a failure of the intermediary; • In jurisdictions where client assets may be re-hypothecated for the purpose of financing client long positions and covering short positions, they should not be re-hypothecated for the purpose of financing the own-account activities of the intermediary; and • Only entities subject to adequate regulation of liquidity risk should be allowed to engage in the re-hypothecation of client assets. || Contractual transparency to be implemented now. The other provisions to be followed later following more assessment

8 || An appropriate expert group on client asset protection should examine possible harmonisation of client asset rules with respect to re-hypothecation, taking account of the systemic risk implications of the legal, operational, and economic character of re-hypothecation. || Expert group to be set up

9 || Authorities should adopt minimum regulatory standards for collateral valuation and management for all securities lending and repo market participants. || To be followed later following more assessment

10 || Authorities should evaluate, with a view to mitigating systemic risks, the costs and benefits of proposals to introduce CCPs in their inter-dealer repo markets where CCPs do not exist. Where CCPs exist, authorities should consider the pros and cons of broadening participation, in particular of important funding providers in the repo market. || To be followed later following more assessment

11 || Changes to bankruptcy law treatment and development of Repo Resolution Authorities (RRAs) may be viable theoretical options but should not be prioritised for further work at this stage due to significant difficulties in implementation. || To be considered at a later stage

Annex 14: Glossary of terms

ABS - Asset Backed Security - Asset backed securities are securities backed by a pool of receivables. Investors generally only bear the risk arising from these receivables and are generally insulated from the credit risk of the respective (former) owner of the assets (originator/seller). The receivables of the underlying portfolio that is securitised generate interest and principal payments. These payments as well as potential losses that may occur in case the underlying obligors of the securitised assets do not serve their obligations, are distributed to investors according to certain rules (« the structure »). Hence, the investors in ABS have to focus on both the underlying risk of the securitised portfolio and the rules that determine which consequences investors have to face in case a certain event occurs. Typically, the securitised assets are referenced by various notes with different risk profiles, and hence, ratings. The fact that different notes have different risk profiles, though they all reference the same underlying portfolio, is based on the respective aforementioned transaction structure. This in principle can enable investors to satisfy their individual risk appetite and needs. ABS allows for a broad band of flexibility in terms of asset classes being securitised and structures being applied.

BU - Banking Union - The Banking Union in the broad sense includes a single legal and regulatory framework for all EU banks, a single bank supervisor, a single bank recovery and resolution mechanism (authority and fund), including provisions to bail in creditors efficiently and effectively when needed, and a single deposit guarantee scheme. The justification for a banking union in a monetary union are financial stability, efficiency of financial intermediation, and the effective and uniform transmission of monetary policy throughout all member states. Banking Union implies that the creditworthiness of a national sovereign be decoupled from the creditworthiness of the banks in its jurisdiction. It is designed to break the link between ailing banks and indebted governments. The goal is to avoid the strain which is put on public finances when banks need rescuing, and, at the same time, to reduce banks’ exposure to increasing risks in public debt. Banking Union is based on a single rulebook for financial regulation, common to all 28 members of the Single Market.

BCBS – The Basel Committee on Banking Supervision is the primary global standard-setter for the prudential regulation of banks and provides a forum for cooperation on banking supervisory matters. It is a committee of banking supervisory authorities that was established by the central bank governors of the Group of Ten countries in 1974. It provides a forum for regular cooperation on banking supervisory matters.

BRRD - Bank Recovery and Resolution Directive - is a Commission's proposal for a Directive on crisis prevention, management and resolution that assigns to the EBA the task to develop a wide range of Binding Technical Standards, Guidelines and reports on key areas of recovery and resolution, with the aim of ensuring effective and consistent procedures across the European Union, in particular with respect to cross-border financial institutions.

CRD IV/CRR - Capital Requirements - EU rules on capital requirements for credit institutions and investment firms putting in place a comprehensive and risk-sensitive framework and to foster enhanced risk management amongst financial institutions.

Collateral - Collateral is an asset or third party commitment that is used by the collateral provider to secure an obligation to the collateral taker. Collateral arrangements may take different legal forms. Collateral may be obtained using the method of title transfer or pledge. It may be forfeited in the event of a default. It includes all sorts of legal arrangements giving additional security to a creditor, e.g. pledge, lien, repo.

Collateral management - granting, verifying, and giving advice on collateral transactions in

order to reduce credit risk in unsecured financial transactions.

Competent authority - Any organization that has the legally delegated or invested authority,

capacity, or power to perform a designated function. In the context of  structural reform, it refers to the body which is in charge of bank supervision.

Default - An event stipulated in an agreement as constituting a default. Generally, such events relate to a failure to complete a transfer of funds or securities in accordance with the terms and rules of the contract in question. A failure to pay or deliver on the due date, a breach of agreement and the opening of insolvency proceedings may all constitute such events.

Directive - A directive is a legislative act of the European Union, which requires Member States to achieve a particular result without dictating the means of achieving that result. A Directive therefore needs to be transposed into national law contrary to regulation that have direct applicability.

DFA - Dodd Frank Act - The Dodd–Frank Wall Street Reform and Consumer Protection Act became law in the United States in 2010, introducing reforms to financial Regulation.

EBA – European Banking Authority - The European Banking Authority is an independent EU Authority which works to ensure effective and consistent prudential regulation and supervision across the European banking sector. See ESA

ECB - European Central Bank is the central bank for the euro and administers the monetary policy of the Eurozone, which consists of 18 EU member states. It is one of the world's most important central banks and is one of the seven institutions of the European Union (EU) listed in the Treaty on the Functioning of the European Union (TFEU). The capital stock of the bank is owned by the central banks of all 28 EU member states.

EIOPA – European Insurance and Occupational Pensions Authority is part of the European System of Financial Supervision consisting of three European Supervisory Authorities (ESA) and the European Systemic Risk Board (ESRB). It is an independent advisory body to the European Parliament, the Council of the EU and the European Commission. EIOPA’s core responsibilities are to support the stability of the financial system, transparency of markets and financial products as well as the protection of insurance policyholders, pension scheme members and beneficiaries. See ESA.

EESC – European Economic and Social Committee is a consultative EU Committee established in 1958. It is a consultative assembly composed of employers (employers' organisations), employees (trade unions) and representatives of various other interests.

ESAs – European Supervisory Authorities - European Securities and Markets Authority (ESMA); European Banking Agency (EBA); and European Insurance and Occupational Pensions Authority (EIOPA) - created in January 2011 with a mandate to contribute to financial stability and improve the functioning of the internal market by creating an integrated supervisory framework.

ESMA - European Securities and Markets Authority - Successor body of CESR, continuing work in the securities and markets area as an independent agency and also with the other two former level three committees. See ESA.

ESRB - European Systemic Risk Board - European Systemic Risk Board was set up in response to the de Larosière group's proposals, in the wake of the financial crisis. This independent body has responsibility for the macro-prudential oversight of the EU.

Financial instrument - A financial instrument is an asset or evidence of the ownership of an asset, or a contractual agreement between two parties to receive or deliver another financial instrument.

FSB - Financial Stability Board - Established to coordinate at the international level the work of national financial authorities and international standard setting bodies and to develop and promote the implementation of effective regulatory, supervisory and other financial sector policies. It brings together national authorities responsible for financial stability in significant international financial centres, international financial institutions, sector-specific international groupings of regulators and supervisors, and committees of central bank experts.

G8 – The countries of Canada, France, Germany, Italy, Japan, Russia, the United Kingdom, and the United States.

G20 – The Group of Twenty Finance Ministers and Central Bank Governors. The G-20 is made up of the finance ministers and central bank governors of 19 countries: Argentina, Australia, Brazil, Canada, China, France, Germany, India, Indonesia, Italy, Japan, Mexico, Russia, Saudi Arabia, South Africa, Republic of Korea, Turkey, United Kingdom, United States of America. The European Union, who is represented by the rotating Council presidency and the European Central Bank, is the 20th member of the G-20.

Government intervention – can take place through amongst others taxation (subsidisation), regulation, or the setting up of institutions. Basic justifications for government intervention are (i) to correct market failures, e.g. those due to positive or negative externalities, asymmetric information, and coordination failures; (ii) to guard against abuse of market power, thus keeping markets competitive (e.g. competition policy); (iii) to redistribute income, through taxes and subsidies; (iv) to keep the system honest by creating and enforcing rules of the game (e.g. Libor); and (v) to protect taxpayer interests when public money is spent or put at risk.

Hedging - Hedging is the practice of offsetting an entity's exposure by taking another opposite position, in order to minimise unwanted risk. This can also be done by offsetting positions in different instruments and markets.

HFT - High frequency trading - High frequency trading is a type of electronic trading that is often characterised by holding positions very briefly in order to profit from short term opportunities. High frequency traders use algorithmic trading to conduct their business.

IFRS –International Financial Reporting Standards are designed as a common global language for business affairs so that company accounts are understandable and comparable across international boundaries. They are a consequence of growing international shareholding and trade and are particularly important for companies that have dealings in several countries.

Leverage - The “leverage ratio” is defined in Article 4(86) on the CRR IV proposal as the relative size of an institution's assets, off-balance sheet obligations and contingent obligations to pay or to deliver or to provide collateral, including obligations from received funding, made commitments, derivatives or repurchase agreements, but excluding obligations which can only be enforced during the liquidation of an institution, compared to that institution’s own funds.

Liquidity –

Liquidity defies a simple definition. Market liquidity is generally referred to as the ability to buy or sell an asset at short notice with little impact on its price. Funding liquidity describes the ability to raise cash either by borrowing or via the sale of an asset (which again depends on market liquidity). Market liquidity is a complex concept that is used to qualify market and instruments traded on these markets. It aims at reflecting how easy or difficult it is to buy or sell an asset, usually without affecting the price significantly. Market liquidity is a function of both volume and volatility. Market liquidity is positively correlated to volume and negatively correlated to volatility. A stock is said to be liquid if an investor can move a high volume in or out of the market without materially moving the price of that stock. If the stock price moves in response to investment or disinvestments, the stock becomes more volatile. Financial institutions provide funding liquidity through, for example, interbank lending, and they provide market liquidity to securities markets, for instance through market-making activity. The conditions under which these intermediaries can fund their balance sheets, in turn, depend on the willingness of other market participants to provide funding or market liquidity. Thus, liquidity is to a large degree endogenous. A funding shortage (illiquidity) arises when it is prohibitively expensive to (i) borrow more funds (low funding liquidity) and/or (ii) sell assets (low market liquidity). Funding liquidity may dry up due to maturity mismatch, high margins/haircut, roll-over risk, redemption risk, etc. Market liquidity may be low due to fire sale discounts or depressed sales prices. Funding illiquidity and market illiquidity interact in times of crisis and explain why liquidity can suddenly evaporate (cause a funding shortage). The two liquidity concepts do not exist in a vacuum. They are influenced by the level of confidence and risk appetite in the financial system, and by the financial soundness of other financial firms. 

Long Position - A “long position” refers to the buying of a security with an expectation that the security will rise in value.

Market efficiency - Market efficiency refers to the extent to which prices in a market fully reflect all information available to investors. If a market is efficient, then no investors should have more information than any other investor, and they should not be able to systematically predict price changes better than other investors.

Market failures – refer to economic situations when the market (made up of private actors and when left on their own without government intervention) does not provide a good or service efficiently even though economic benefits outweigh economic costs. This happens when the private benefits (or costs) are not equal to the public benefits (or costs). Market failure is an economic concept, not a political one. Market failures justify state intervention. The main types of market failures are (i) externalities/spill-overs: Positive (R&D, training) or negative (environmental pollution, bank failures); (ii) imperfect/asymmetric information (SME financing, financial market freezes, etc.); (iii) coordination failures (standard setting, subsidy races, depositor runs); (iv) abuse of dominant positions; and (v) public good provisioning (defence, legal system, etc.).

Market fragmentation - Market fragmentation typically refers to the fact that rules and/or market conditions vary across countries and/or markets for similar services. It is the opposite of market integration. 

Market integrity - Market integrity is the fair and safe operation of markets, without misleading information or inside trades, so that investors can have confidence and be sufficiently protected.

Market maker - A market maker is a firm that will buy and sell a particular security on a regular and continuous basis by posting or executing orders at publicly quoted bid-ask prices.

Market making - Market making is the purchase and sale of financial instruments (government bonds, corporate bonds, equities, derivatives, etc.) for own account at prices defined by the market maker, on the basis of a commitment to provide market liquidity on a regular and on-going basis.

MiFID - Markets in Financial Instruments Directive is a EU law that provides harmonised regulation for investment services across the 31 member states of the European Economic Area (the 28 Member States of the European Union plus Iceland, Norway and Liechtenstein). The main objectives of the Directive are to increase competition and consumer protection in investment services. As of the effective date, 1 November 2007, it replaced the Investment Services Directive.

MTF - Multilateral Trading Facility is an electronic system which facilitates the exchange of securities between counterparties. The securities may include derivatives and instruments which do not have a main market, as well as traditional securities.

Negative externalities - A negative externality in economics and finance is usually a cost incurred by a party that is outside of the decision or transaction of another party. For example, pollution and traffic jams are externalities of a person’s private decision to drive a car to work. Negative externalities are an important type of market failure which justify government intervention (e.g. taxation). 

OECD – Organisation for Economic Co-operation and Development is an international economic organisation of 34 countries founded in 1961 to stimulate economic progress and world trade. It is a forum of countries committed to democracy and the market economy, providing a platform to compare policy experiences, seek answers to common problems, identify good practices and co-ordinate domestic and international policies of its members.

OTC - Over the Counter - Over the counter, or OTC, trading is a method of trading that does not take place on an organised venue such as a regulated market or an MTF. It can take various shapes from bilateral trading to trading done via more organised arrangements (such as systematic internalisers and broker networks).

Principle of proportionality - Similarly to the principle of subsidiarity, the principle of proportionality regulates the exercise of powers by the European Union. It seeks to set actions taken by the institutions of the Union within specified bounds. Under this rule, the involvement of the institutions must be limited to what is necessary to achieve the objectives of the Treaties. In other words, the content and form of the action must be in keeping with the aim pursued. The principle of proportionality is laid down in Article 5 of the Treaty on European Union. The criteria for applying it is set out in the Protocol (No 2) on the application of the principles of subsidiarity and proportionality annexed to the Treaties.

PE - Private equity is an asset class consisting of equity instruments provided to firms that are not publicly traded on an exchange. Private equity is about buying stakes in businesses, transforming business and then realising the value created by selling or floating the business.

Procyclicality - A condition of positive correlation between the value of a good, a service or an economic indicator and the state of the economy. The value of the good, service or indicator tends to move in the same direction as the economy, growing when the economy grows and declining when the economy declines. The term is generally used to refer to the mutually reinforcing mechanisms through which the financial system can amplify business fluctuations and possibly cause or exacerbate financial instability. These 'positive feedback' mechanisms are particularly disruptive and apparent during an economic downturn.

Proprietary trading - Proprietary trading is the purchase and sale of financial instruments for own account with the intent to profit from subsequent price changes.

Regulation - A regulation is a form of legislation that has direct legal effect on being passed in the European Union.

Regulatory arbitrage - Regulatory arbitrage is exploiting differences in the regulatory situation in different jurisdictions or markets in order to make a profit.

Regulated market - Regulated market is a multilateral system which brings together or facilitates the bringing together of multiple third-party buying and selling interests in financial instruments in a way that results in a contract. Examples are traditional stock exchanges such as the Frankfurt and London Stock Exchanges.

Rehypothecation - Any pre-default use of financial collateral by the collateral taker for its own purposes.

Repo/Repurchase agreement - A 'repurchase agreement' is economically similar to a secured loan, with the 'repo buyer' (effectively the money lender) receiving securities as collateral to protect him against the default by the 'repo seller' (effectively, the borrower of money). Legally, a 'repo,' can be defined as a collateral arrangement in which the 'repo seller' transfers ownership of securities sold to the 'repo buyer' for an amount of cash (the purchase price) at moment T, while the 'repo buyer' agrees to sell and transfer equivalent securities at a future moment T+x for a certain amount of money, including an interest component (the repurchase price).

Risk premium - The risk premium is the return that investors require above the amount that a similar but otherwise 'risk-free' asset promises. A risk-free asset is a theoretical asset that would never default. So the risk premium is the amount that an investor wants to be paid for taking credit and/or market risk.

Sanction - A penalty, either administrative or criminal, imposed as punishment.

SFT - Securities financing transactions - This is the general terms for financing transactions backed by securities collateral such as repo, securities lending or any transaction having an equivalent economic effect and posing similar risks, in particular buy/sell back transactions.

Securities lending/Securities lending agreement - In securities lending, the economic purpose is to temporarily obtain a specific security for certain purposes, e.g. to facilitate settlement of a trade or to facilitate delivery of a short sale. Legally, a ‘securities lending agreement’, can be defined as a transaction in which the 'securities borrower' borrows specific assets from 'securities lender' in exchange for a fee and collateral at moment T. The parties agree to return the lent securities and the collateral on maturity.

Securitisation - Asset backed securities are securities backed by a pool of receivables. Investors only bear the risk arising from these receivables and are generally insulated from the credit risk of the respective (former) owner of the assets (originator/seller). The receivables of the underlying portfolio that is securitised generate interest and principal payments. See ABS.

SIFIs - Systemically important financial institution is a bank, insurance company, or other financial institution whose failure might trigger a financial crisis.

Single rulebook - The single rulebook is the concept of a single set of rules for all Member States of the union so that there is no possibility of regulatory arbitrage between the different markets.

SMEs - Small and medium sized enterprises - On 6 May 2003 the Commission adopted Recommendation 2003/361/EC regarding the Small and medium sized enterprise definition. While 'micro' sized enterprises have fewer than 10 employees, small have less than 50, and medium have less than 250.

Shadow banking system - Defined by the FSB as 'the system of credit intermediation that involves entities and activities outside the regular banking system' or, in short, 'non-bank credit intermediation'. Experience from the financial crisis demonstrates that capacity for some non-bank entities and transactions to operate on a large scale in ways create bank-like risks to financial stability (long-term credit extension based on short-term funding and leverage).

Short Position - refers to the selling of a security with an expectation that the security will fall in value.

SSM - Single Supervisory Mechanism is a mechanism through which, per the European Commission's proposal, the European Central Bank (ECB) shall assume ultimate responsibility for specific supervisory tasks related to the financial stability of the biggest and most important Eurozone based banks.

SRM - Single Resolution Mechanism has been designed to complement the Single Supervisory Mechanism (SSM) which, once operational in late 2014, will see the European Central Bank (ECB) directly supervise banks in the euro area and in other Member States which decide to join the Banking Union. SRM would ensure that – not withstanding stronger supervision - if a bank subject to the SSM faced serious difficulties, its resolution could be managed efficiently with minimal costs to taxpayers and the real economy.

Subsidiarisation – Subsidiarisation in the context of structural reform requires the trading entity and the deposit entity to maintain self-standing reserves of capital and of loss-absorbing debt, as well as to comply with other prudential and legal requirements on an individual, sub-consolidated or consolidated basis. Subsidiarisation provides a degree of independence and to some extent also insulates the deposit entity from shocks and losses.

Systemic failure - A systemic failure refers either to the failure of a whole market or market segment, or the failure of a significant entity that could cause a large number of failures as a result.

TEC – Treaty of the European Community

TFEU – Treaty on the functioning of the European Union

Trade repository - Means a legal person that centrally collects and maintains the records of certain transactions.

Trading venue - A Trading venue is an official venue where securities are exchanged; it includes MTFs and regulated markets (e.g. typical stock exchanges).

UCITS – The Undertakings for Collective Investment in Transferable Securities, Directive 2001/107/EC and 2001/108/EC (UCITS) are a set of European Union Directives that aim to allow collective investment schemes to operate freely throughout the EU on the basis of a single authorisation from one member state. In practice many EU member nations have imposed additional regulatory requirements that have impeded free operation with the effect of protecting local asset managers.

Underwriting - Securities underwriting is a typical investment banking activity in which banks raise investment capital from investors on behalf of corporations and governments that are issuing securities (both equity and debt securities) in return for a fee. It is a way of selling newly issued securities, such as stocks or bonds, to investors.

Venture capital - Venture capital (VC) is that part of private equity that entails finance provided to early-stage, high-potential and possibly, high-growth start-up companies. This commonly covers the seed to expansion stages of investment.

Volatility - Volatility refers to the change in value of an instrument in a period of time. This includes rises and falls in value, and shows how far away from the current price the value could change, usually expressed as a percentage.

Volcker Rule – The "Volcker Rule" refers to Section 619 of the Dodd-Frank Wall Street Reform and Consumer Protection Act. See Dodd Frank Act.

[1] In 2011 the Basel Committee on Banking Supervision supplemented the trading book framework with an incremental risk capital charge, which includes default risk as well as migration risk, for unsecuritised credit products. http://www.bis.org/publ/bcbs158.htm

[2] http://ec.europa.eu/internal_market/bank/docs/high-level_expert_group/report_en.pdf

[3] See e.g. remarks from Wayne Bayres – Secretary General of the Basel Committee in Banking Supervision “Regulatory reforms – incentives matter” to the Bank of Portugal conference on the 24 October 2012.

[4] There is clearly not a one to one correspondence between commercial/trading activities and declared split of RWAs between credit and market risk. Not only market risk can also be originated by activities considered as commercial banking (i.e. assets held at fair value) and credit risk can also be generated by trading activities, but there seem to be no clear-cut definition of commercial banking activities and trading activities in the literature or in policy practice

[5] The analysis is conducted on a sample of 255 banks located in the EU and across years 2006 to 2011. Data are sourced from the SNL commercial database http://www.snl.com. See next section for details.

[6] The measurement of aggregate profits is subject to a high degree of uncertainty, (see e.g. “Measuring and analysing profit developments in the Euro area” http://www.ecb.europa.eu/pub/pdf/other/pp63-73_mb200401en.pdf - ECB). We are here however not interested in exact measures in absolute terms as we are more concerned with measuring levels of income per unit of capital in order to measure incentivation effects.

[7] These 255 banks are available in the SNL database (http://www.snl.com), have a total assets data point available in 2011 and are located in EU27 countries. 183 banks are effectively used to estimate the preferred model. The reduction of the sample is only due to missing data.

[8] The exercise was repeated also on a sample including all EEA EFTA countries (i.e. plus Norway, Iceland, Lichtenstein) EU27 and all EFTA countries (i.e. EEA EFTA plus Switzerland). Results were substantially unvaried. See the following section for additional details.

[9] At the time of the first drafting of this report Croatia still had to join the EU, bringing membership to 28 states. The current analysis refers to the 27 members of the EU as of January 2012. 14 financial institutions are available in SNL for Croatia; all of them are small-sized: their respective total assets are under 30 bn EUR. Some of these banks are local subsidiary of European financial institutions (Raiffeisenbank Austria for example). None of these would have been proposed as candidates for structural reform.

[10] http://www.ecb.int/stats/money/consolidated/html/index.en.html contains EU consolidated banking data for 2011 divided between Domestic credit institutions, with a sum of total assets of 35’901 bn EUR, and Foreign-controlled subsidiaries and branches, with a sum of total assets of 8’916 bn EUR.

[11] In their report “Trading activities and bank structural separation: possible definitions and calibration of exemption thresholds” Commission Services have proposed different definitions in order to identify institutions conducting different trading activities, such as market making and proprietary trading. The sample analysed in this report is selected based on definition 3, focusing on market and counterparty risk, see page 10 of the cited report for a more in-depth discussion. Although the sample of selected banks varies according to the definitions used for selection purposes, definitions 2, 3 and 4 of that report would all involve the same allocation of trading assets and liabilities for selected banks, so that a single analysis can be considered to cover all these three cases.

[12] It should be noted that two banks from the candidates’ sample are identified as outliers when calculating results, and they are thus dropped from the sample effectively used in section 5.

[13] It should be noted that income statement data are less populated than balance sheet data used in the panel regression, thus the sample used for the calculation and presentation of results might differ from the sample used in the regression analysis.

[14] http://www.eba.europa.eu/capitalexercise/2011-EU-Capital-Exercise.aspx

[15] Only an extremely limited number of banks publish RWAs/MCRs for the banking and trading books. However, the definition of banking and trading book might not coincide with the definitions of trading and commercial banking activities here investigated.

[16] See e.g. S. Avramova and Le lesle, V., (2012), “Revisiting Risk-Weighted Assets: Why Do RWAs Differ Across Countries and What Can Be Done About It?”, IMF working papers, WP/12/90, International Monetary Fund, Washington (DC).

[17] See Appendix A for more details on the econometric analysis development.

[18] In fact, predicted RWAs based on these coefficients are re-normalized to sum to total RWAs declared by banks (i.e. estimated RWAs are in fact used to attribute to each activity line a share of RWAs as declared by banks). A series of corrections based on expected effects of changes of rules and definitions when moving to Basel III or alternative regulatory scenarios are also considered in the final allocation. See Annex F for details.

[19] Alternative breakdowns were also tested, in particular other liability side items and net values of assets and liabilities. See (Reference to be inserted) for additional details of categories’ definitions.

[20] The exercise was also repeated including all European Economic Area countries: all coefficients remain identical except for the coefficient on Securities Held to Maturity, which halves and becomes even less significant.

[21] This variable has a coefficient of 0.20 before the introduction of Basel 2.5 regulation changes in 2011.

[22] This coefficient is basically indistinguishable from zero, given its extremely low statistical significance.

[23] The stars denote statistical confidence according to the usual 10%/5%/1% scale

[24] The number of cases used in the regression is lower than the full sample due to the fact that not all variables are available for all banks in the sample and all years.

[25] Different regulatory scenarios will imply different corrections to the predicted RWAs and the final value of the MCRs. See Annex A5 for details.

[26] During preparation of the report, an alternative assumption related to the alternate allocation of assets held at fair value or AFS was also considered. Intermediate calculations referring to additional definitions considering this element are available upon request.

[27] This is done by calculating the share of predicted RWAs of each activity on total predicted RWAs and multiplying it by total RWAs reported in the balance sheet.

[28] The European Banking Authority conducted a Quantitative Impact Study (QIS) to estimate the impact of new requirements to raise quality and level of capital base. Results presented in Annex A5 that shows that the RWAs would increase. For the purposes of this document, as precise data on fulfillment of criteria 2 and 3 are not available, group 1 banks are those with a capital in excess of 3bn. The increase in RWAs is calculated separately for trading activities and for commercial banking activities. This methodology for the correction of RWAs is consistent with the treatment of RWAs applied in the JRC cost benefit analysis via the SYMBOL model. QIS corrections by country as available in 2011 are used a factor of correction of RWA independently of the years (i.e. the same correction is applied from 2006 to 2011).

[29] While RWAs are increased in Basel III scenarios, MCRs are decreased in Basel II scenarios, to represent the fact that part of what is declared and allowed as capital will not be considered as such under the more stringent future rules.

[30] Alternative definitions have been tested for the allocation of income. Results can be provided by the authors.

[31] This departure from the use of earnings before taxes is mainly due to the fact that under alternatives definitions of trading activities, which were tested in other versions of this work

[32] For interest revenues not referring to loans, this is equivalent to an assumption that the interest rate revenue per unit of commercial or trading assets are equivalent: the hypothesis of equal returns has been confirmed via statistical tests conducted on the part of the sample for which more detailed data is available. For interest expenses this is equivalent to assuming that all activities would face the same funding cost: this assumption is not based on results from data, but constant funding costs across activities could be justified on the grounds that all activities within the same institution would be facing the same funding WACC. A set of regression models was tested to obtain a more detailed split of operational expenses across activity lines, without satisfactory results.

[33] Results where operational expenses are not included are available from the authors.

[34] Graphs are based on all available observations in the entire 6-years period; thus the same bank can be considered several times in the construction of the plot.

[35] The definitions of outliers employed here is the original introduced by Tukey: 1.5 times the inter-quartile range above or below the 3rd and 1st quartile, respectively.

[36] It should be noted that, as some banks have a “whole bank” density which is 1 or very close to it (as per their balance sheet), the RWA density for the commercial activities will come out as larger than one if the density for the trading activities will be lower than the density for the “whole bank”. These points are excluded from the graph.

[37] Three banks are excluded from the sample from this point of the document due to their results being outliers by several orders of magnitudes with respect to the rest of the data.

[38] In a previous version of the study, additional scenarios based on changes in the risk weights coefficients for all years (i.e. counterfactuals for B2 weights post-2010 or for B2.5/B3 weights pre-2011) were also included. Given the fact that the new regressions do not highlight a large effect from the changes in the sample considered, and the complication of several interacting scenarios and definitions, this has not been included in the current analysis.

[39] The main difference with MCR is therefore that, while MCR is calculated using the exact capital adequacy ratio, this proxy of equity will take into account any eventual surplus capital and will be set always to be equal to or larger than the estimated MCR.

[40] In 2011 the Basel Committee on Banking Supervision supplemented the trading book framework with an incremental risk capital charge, which includes default risk as well as migration risk, for unsecuritised credit products. http://www.bis.org/publ/bcbs158.htm

[41] www.snl.com

[42] The main advantage of this definition of income is that it allows to consider unrealized gains and losses in available for sale activities at the same time as corresponding gains and losses in activities accounted for at fair value.

[43] Alternative definitions of trading activities including one or both of these classes were also tested, results are available from the authors

[44] These hypotheses have been confirmed via statistical tests conducted on the part of the sample for which more detailed data is available. The assumption of proportional sharing of funding costs is instead not borne by the data, but could be justified on the grounds that all activities within the same institution would be facing the same funding WACC.

[45] This implies that when a value is missing, the repartition is applied to a upper level of the income hierarchy. In this case if at least one value between ‘Interest Earned on Customer Loans’ (1.1.a1) and ‘Interest Earned on Trading Assets’ (1.1.a2) is missing, α will be applied to the ‘Interest income’ (1.1.a)

[46] Other operating income = Equity accounted results + dividends from equity instruments + rental revenue + lease and rental revenue + other non-interest income

[47] Appendix D provides a set of figures for all banks in the original sample.

[48] A. Pagano, J. Cariboni, M. Marchesi, N. Ndacyayisenga, M. Petracco-Giudici, H. Joensson; (201) “Trading activities and functional structural separation: possible definitions and calibration of a de minimis exemption rule”, JRC (European Commission Internal).

[49] In this exercise, Group1 banks are those banks whose Tier Capital is over 3 bn EUR. Group 1 RWA will be increased by +21.2%, Group 2 by +6.9%.

[50] In this exercise, Group1 banks are those banks whose Tier Capital is over 3 bn EUR. Group 1 RWA will be increased by +21.2%, Group 2 by +6.9%.

[51] Due to data issues (excess missingness, outlier behaviour, excess variation), LLänsförsäkringar, ING Bank, Belfius Banque, ESF Group data are not considered. Some other banks are also partially considered (Dexia, Lloyds Group, Rentenbank, Caixa Económica Montepio Geral, Cajas Rurales Unidas, HSH Nordbank, SNS REAAL).

[52] http://www.eba.europa.eu/documents/10180/87706/EBA-BS-2012-037-FINAL--Results-Basel-III-Monitoring-.pdf/778804a5-8e3e-4073-83df-afd1be0b626e

[53] See footnote 5 for complete reference.

[54] These effects are not to be confused with those linked to the stricter definition of the quality of capital introduced by Basel 3.The change in the RWA due to the change in the definition of capital measures: (i) the effects of lower RWA for exposures that are included in RWA under Basel 2 but receive a deduction treatment under Basel 3; (ii) the increase in RWA applied to securitisation exposures deducted under the Basel 2 that are risk-weighted at 1250% under Basel 3; (iii) the increase in RWA for exposures that fall below the 10% and 15% limits for CET1 deduction.

[55] It should be noted that total ROA referring to the income and total assets items used in this paper, which differ from the after-tax profit reported by the firm, can be obtained as  where ,  refer to the Trading Activity and Commercial Banking Activity.

[56]     Many of the leading UK banks have told the UK Parliamentary Commission on Banking Standards that they do not engage in proprietary trading at all. The same message was given by NL banks to the Members of the Commission on the structure of Dutch banks. The French and the German structural reform proposals propose to subsidiarise proprietary trading (see Annex A1 for more details). Their cost-benefit analysis findings have not been made public, but BNP Paribas corporate-banking and investment-banking revenues are estimated to be impacted by the government plans by less than 2%. An internal Febelfin survey provides evidence that proprietary trading amounts to 2% of trading revenues for Belgian banks in 2012 (down from 13%, 11% and 8% in 2009, 2010, and 2011 respectively). In turn, trading revenues are estimated to amount to 9% of overall revenues in 2012 (first semester). 

[57]     Jérôme Kerviel at Société Générale, Nick Leeson at Barings Bank, Kweku Adoboli at UBS are only some of the traders that caused multiple billion trading related losses, some of which effectively bringing down their employer banks.

[58]      Very few banks submitted quantitative evidence in the public consultation. None of them reported that proprietary trading accounted for more than 4% of trading revenues, which in turn is only a fraction of total revenues.

[59]     The treasury function of a bank needs to engage in trades to manage excess liquidity or hedge the risk from for example selling fixed-rate mortgages while being funded with floating rate borrowing. Over time the treasury functions in some banks have become more aggressive traders with strategies that could be seen as resembling proprietary trading. In some cases, Treasury operations no longer merely manage the natural dynamics of the balance sheet arising from customer activity, but increasingly perform a set of trading activities in themselves and become pure profit centres.

[60]     A Febelfin 2013 commissioned survey suggests that market making (excluding short term ALM transactions) accounts for 66% of Belgian bank trading revenues in 2012 (first semester), whereas market making (including short term ALM transactions) would account for 56% of all trading revenues for all Western European banks. Few banks submitted quantitative evidence during the public consultation about the importance of their market making activities, but those that did reported that market making accounted for 25% to 100% of their trading revenues.

[61]     There are two types of market makers. Voluntary Market Makers (VMM) and Designated Market Makers (DMM). VMM act on their own initiative and earn no compensation, but the profit they make by charging the bid-ask spread. DMM are appointed and have contractual obligations to maintain two way prices and volumes for a specified period of the trading day. DMM are contractually required to give customers the best bid or ask price for each market order transaction. This ensures a fair and reasonable two-sided market. In return for fulfilling these obligations, DMM are often offered a range of potential benefits by exchanges, including reduced trading fees, monthly stipends and a share of net trading revenue. DMM exist in most stock exchanges of the major industrialised countries. The presence of DMM varies across markets. The market for foreign exchange and for some classes of derivatives rely mainly on VMM. Other markets, such as equity markets give rise to both VMM and DMM. For some rarely traded instruments, DMM are the main actors. Most transactions done in OTC markets are with a MM. Designated market makers provide these benefits also in bad times and therefore dampen negative cyclical effects. Without designated market makers liquidity would dry up each time there is a market downturn. Of course, their ability to assume this role is limited, given that they expose their balance sheet to losses. Losses can put all intermediaries in distress, and given that several designated market makers are universal banks in Europe the losses on designated market making activity can put the traditional banking activities at risk such as the provision of payment system services or of loans to non-financial corporates.

[62]     Market makers often have signed on to a voluntary code of conduct, which already considers manipulative practices by banks with each other or with customers to be “unacceptable trading behaviour”. However, the multiple financial scandals in the years since signing these voluntary agreements cast doubt on their effectiveness.

[63]     Zingales (2012): “With the repeal of Glass-Steagall, investment banks exploded in size and so did their market power. As a result, the new financial instruments (such as credit default swaps) developed in an opaque over-the-counter market populated by a few powerful dealers, rather than in a well regulated and transparent public market. The separation between investment and commercial banking also helps make the financial system more resilient. After the 1987 stock market crash, the economy was unaffected because commercial banks were untouched by plummeting equity prices. During the 1990-91 banking crisis, securities markets helped alleviate the credit crunch because they were unaffected by the banking crisis. By contrast, in 2008 the banking crisis and the stock market crisis infected each other, pulling down the entire economy.”

[64]     For example, benefits of market liquidity should become smaller with the degree of market liquidity. The additional benefits of the extra liquidity derived from high-frequency trading must be of negligible (or negative) value compared to the benefits from having a market which is reasonably liquid on a day-by-day basis. Moreover, ever greater market liquidity may give rise to destabilising momentum effects, such as cycles of undervaluation and overvaluation. In addition, voluntary market making may not occur when it is most needed, i.e. during troubled market conditions. Even dedicated market makers are typically only allowed to post quotes during 90% of the trading period and of course they may decide to breach their contractual obligations if they deem that fulfilling them would threaten their solvency.

[65]     Zingales (2012): “The third reason why I came to support Glass-Steagall was because I realised it was not simply a coincidence that we witnessed a prospering of securities markets and the blossoming of new ones (options and futures markets) while Glass-Steagall was in place, but since its repeal have seen a demise of public equity markets and an explosion of opaque over-the-counter ones.  To function properly markets need a large number of independent traders. The separation between commercial and investment banking deprived investment banks of access to cheap funds (in the form of deposits), forcing them to limit their size and the size of their bets. These limitations increased the number of market participants, making markets more liquid. With the repeal of Glass-Steagall, investment banks exploded in size and so did their market power. As a result, the new financial instruments (such as credit default swaps) developed in an opaque over-the-counter market populated by a few powerful dealers, rather than in a well regulated and transparent public market.”

[66]     When facilitating client business a bank is likely to try and hedge most of its risks. Hence, genuine market making should entail limited market risk. However, the actual exposure to risk may vary across time depending on the liquidity of the instruments, on changes in market volatility and on significant variation in the sizes of positions that market making clients may wish to acquire or liquidate. Moreover, there may be a mismatch between the position and the hedge (basis risk) and the hedge will need to be rebalanced over time as market moves alter risk profiles. Furthermore, market makers are still exposed to high counterparty risk and the concrete functioning of market making can vary in relation to different financial instruments and market models.

[67]     Ellis et al.  (2000) and Kang and Liu (2007)

[68]     Kroszner and Rajan (1994), Kroszner (1998), Hebb and Fraser (2003) and Stiglitz (2010)

[69]     Kroszner and Rajan (1994), Puri (1994), Benston (1990), Hebb and Fraser (2002), Hebb and Fraser (2003)

[70]     Hartzell, Kallberg, Liu (2008) and Kovner (2010).

[71]     Diamond (1991), Rajan (1992), Saunders and Walter (1994), and Stein (2002), for example.

[72]     Kroszner and Rajan (1994), Kanatas and Qi (2002), Drucker and Puri (2005).

[73] Mortgage securities are created when these loans are packaged, or “pooled,” by issuers or servicers for sale to investors. As the underlying mortgage loans are paid off by the homeowners, the investors receive payments of interest and principal.

[74] In a simple CDO, the underlying pool consists of bonds; whereas in a synthetic CDO the underlying pool consists of Credit default Swaps.

[75] Here the underlying pool consists of commercial loans.

[76]     Note that several measures have been taken after the crisis with respect to securitisation, such as CRD IV requirements on risk retentions and due diligence duties, CRA III regulation and transparency obligation to be introduced in MIFID 2. However, the concerns of TBTF and TITF remain.

[77]      Popov and Smets (2013) analyse the role of direct intermediation through financial markets with the indirect intermediation through levered banks. They argue that less deep financial markets in the EU relative to those of the US are, to a large extent, responsible for the smaller increase in productivity and slower pace of industrial innovation. They also compare the liquidity spirals, asset fire sales, and interbank market freezes of the recent financial crisis with the much more orderly burst of the dot-com bubble. They argue that the credit boom of the 2000s was driven by debt finance, while the dot-com bubble was mostly driven by an expansion in equity ownership, and equity is not held in levered portfolios.

[78]     A Frontier Economics report (2013) estimates that the market value of patents granted to private equity backed firms in 12 large European countries between 2007 and 2011 is 1.5 times the total amount of private equity investment during that period. Also they find that the value added of financing in private equity can be significantly more effective in promoting innovation (measured by the number of patents granted) than non-private equity finance. Mollica and Zingales (2007) show that the causality runs from private equity (in particular venture capital) to innovation and not the other way around (that is private equity selecting more innovative firms). However, Gilligan and Wright (2012) consider that the evidence is ambiguous on whether private equity performance outweighs alternative forms of investments such as quoted shares. They explain that while industry studies shows better performance of private equity funds, once controlling for risk and correcting for sample bias evidence shows that average fund performance is very close to the S&P 500 (from slight over-performance to 3% per annum underperformance).

[79]     Not all investment banking activities are transaction-oriented activities. Not all commercial banking activities are relationship-oriented activities.

[80]     The classic motivation (Diamond and Dybvig, 1983) for banks to offer deposits derives from the existence of random liquidity shocks faced by depositors and the need for depositors to be insured against these liquidity shocks. The law of large numbers implies that aggregating over these idiosyncratic liquidity shocks leads to exploitable diversification benefits.

[81]             What subsidiarisation means in practice is breaking up complex financial institutions, including branches that cross borders – into distinct subsidiaries. This is often called “ring-fencing,” a term that makes clear that the goal of subsidiarisation is to define robust boundaries between different corporate operations to keep the "sheep on one side and the wolves on the other."

[82]             See as explained on http://www.ifrs.org.

[83]          Directive 2013/34/EU of the European Parliament and of the Council on the annual financial statements, consolidated financial statements and related reports of certain types of undertakings, amending Directive 2006/43/EC of the European Parliament and of the Council and repealing Council Directives 78/660/EEC and 83/349/EEC, OJ L 182, of 26 June 2013, page 19. In this context, it is also relevant to mention the specific Directive that applies to banks and other financial institutions in combination with the Accounting Directives, namely: Council Directive 86/635/EEC on the annual accounts and consolidated accounts of banks and other financial institutions, OJ L 372, of 31 December 1986, page 1.

[84]             See Article 7 of Regulation (EC) No 1606/2002 of the European Parliament and of the Council on the application of international accounting standards, OJ L 243 of 19 July 2002, page 1 ("IFRS Regulation").

[85]             See Article 8 of the IFRS Regulation.

[86]             See the Directive of the European Parliament and of the Council on the access to the activity of credit institutions and the prudential supervision of credit institutions and investment firms and amending Directive 2002/87/EC and repealing Directives 2006/48/EC and 2006/49/EC (the "CRDIV"), OJ L176, of 27 June 2013, page 338.

[87]             Directive 2009/111/EC of the European Parliament and of the Council amending Directives 2006/48/EC, 2006/49/EC and 2007/64/EC as regards banks affiliated to central institutions, certain own funds items, large exposures, supervisory arrangements, and crisis management, OJ L 302, of 16 September 2009, page 97.

[88]             For purposes of this Annex, "insured deposits" are defined as deposits repayable by the guarantee scheme under national law.

[89]             An Investment firm is any legal person whose regular occupation or business is the provision of one or more investment services to third parties and/or the performance of one or more investment activities on a professional basis. Investment services and activities are listed in Annex I to the Directive 2004/39/EC of the European Parliament and of the Council on markets in financial instruments amending Council Directives 85/611/EEC and 93/6/EEC and Directive 2000/12/EC of the European Parliament and of the Council and repealing Council Directive 93/22/EEC, OJ L 145, of 30 April 2004, page 1, ("MiFID").

[90]             However, this credit institution authorization could not cover receiving insured deposits.

[91]             For a more in-depth description of on-going structural reform projects both within the EU and outside, see Annex A1 of this Impact Assessment.

[92]             Directive 2011/35/EU of the European Parliament and of the Council concerning mergers of public limited liability companies, OJ L 110, of 5 April 2011, page 1; the Sixth Council Directive 82/891/EEC based on Article 54(3) of the Treaty, concerning the division of public limited liability companies, OJ L 378, of 31 December 1982, p. 47; and Directive 2005/56/EC of the European Parliament and of the Council on cross-border mergers of limited liability companies, OJ L310 of 26 October 2005, page 1.

[93]             Directive 2012/30/EU of the European Parliament and of the Council on coordination of safeguards which, for the protection of the interests of members and others, are required by Member States of companies within the meaning of the second paragraph of Article 54 of the Treaty on the Functioning of the European Union, in respect of the formation of public limited liability companies and the maintenance and alteration of their capital, with a view to making such safeguards equivalent, OJ L 315, of 25 October 2012, page 374.

[94]             Directive 2011/35/EU of the European Parliament and of the Council concerning mergers of public limited liability companies, OJ L 110, of 5 April 2011, page 1; the Sixth Council Directive 82/891/EEC based on Article 54(3) of the Treaty, concerning the division of public limited liability companies, OJ L 378, of 31 December 1982, p. 47; and Directive 2005/56/EC of the European Parliament and of the Council on cross-border mergers of limited liability companies, OJ L310 of 26 October 2005, page 1.

[95]             See as explained in the European Commission's Green Paper on Shadow Banking, COM(2012) 102 final, Brussels 19 March 2012.

[96]             Those objectives are: reducing moral hazard (excessive risk taking); facilitate resolvability; facilitate management, supervision; reducing conflicts of interest; reducing resource and capital misallocation; and reducing efficiencies.

[97]             The Directive establishing a framework for the recovery and resolution of credit institutions and investment firms (the "BRRD") is formally not yet in force. However, for purposes of this Annex it has been assumed that it will be in force prior to the implementation of any separation measure required by financial institutions. It will therefore be included in the notion of "existing legislation" as used in this Annex.

[98]             "Management body" is defined in the CRDIV and means the body or bodies of an institution, appointed in accordance with national law, which is empowered to set the institution's strategy, objectives and overall direction, and which oversees and monitors management decision-making. This shall include persons who effectively direct the business of the institution.

[99]             See Article 91 of the CRDIV.

[100]            Regulation (EU) No 575/2013 of the European Parliament and of the Council on prudential requirements for credit institutions and investment firms and amending Regulation (EU) No 648/2012 (the "CRR"), OJ L 176, of 26 June 2013, page 1. Note that all Investment firms as defined in Article 4(1) of MiFID are subject to CRR prudential requirements as laid down in the CRDIV and the CRR. Investment firms which are not authorised to provide ancillary services or which provide only certain services and activities or which are not permitted to hold money or securities belonging to their clients are subject to specific prudential requirements which are specified in the CRR and the CRDIV.

[101]            More specifically, credit institutions and investment firms (with some exceptions) are not allowed to incur an exposure to any of their counterparties exceeding 25% of their eligible capital. Importantly, where Member States adopt national laws requiring structural measures to be taken within a banking group, intragroup exposures, where these exposures consist of exposures to an entity that does not belong to the same subgroup, competent authorities may apply a limit on a sub-consolidated basis which is set at a level below 25% but not lower than 15% until 30 June 2015 after which the limit may go as low as to 10% but not further. See Article 395(6) of the CRR.

[102]            This general rules is set out in Articles 6, 11 and 412 of the CRR.

[103]            See Articles 131 and 133 of the CRDIV.

[104]            See Articles 37 to 50 of the Directive of the European Parliament and of the Council establishing a framework for the recovery and resolution of credit institutions and investment firms and amending Council Directives 77/91/EEC and 82/891/EC, Directives 2001/24/EC, 2004/25/EC, 2005/56/EC, 2007/36/EC and 2011/35/EC and Regulation (EU) No 1093/2010 (the "BRRD"). Subject to certain specific conditions, resolution authorities may choose to apply the minimum requirement of own funds and liabilities on a consolidated basis to groups which are subject to consolidated supervision.

[105]            See Article 104 of the CRDIV.

[106]            Governance arrangements have to include a clear organizational structure with well-defined, transparent and consistent lines of responsibility, effective processes to identify, manage, monitor and report the risks that financial institutions are or might be exposed to, adequate internal control mechanisms, and remuneration policies and practices that are consistent with and promote sound and effective risk management. This translates into a requirement for the management body to approve and implement an effective risk management strategy. See Article 88 the CRDIV

[107]              See Articles 92 to 96 of the CRDIV on remuneration policies.

[108]            The term "significant subsidiary" of an EU parent institution is not defined in the CRR.

[109]            See Article 13 of the CRR.

[110]            See Article 7 of the BRRD.

[111]            See Article 11 of the BRRD.

[112]            See Articles 7 and 8 [and 10??] of the CRR regarding the derogation to the application of prudential requirements on an individual basis with regard to own funds, large exposures, leverage, liquidity and public disclosure.

[113]            Article 11(5) of the CRR already provides that competent authorities may require structurally separated institutions to comply with the large exposure obligations on a sub-consolidated basis.

[114]            Articles 16 to 22 of the BRRD provide the conditions and procedures for intra-group financial support. Currently Member States shall ensure that parents and subsidiaries may enter into agreements to help each other. Approval of such an agreement is subject to certain conditions (Article 19).

[115]            See Article 11 of BRRD that provides that group resolution plans shall identify how the group resolution actions could be financed and, where appropriate, set out principles for sharing responsibility for that financing between sources of funding in different Member States.

[116]            "Intra-group exposures" consist of exposures of a group entity or a subgroup of entities to another group entity or subgroup of group entities. Those intragroup exposures may arise from assets or off-balance sheet items attracting a capital requirement for credit or counterparty credit risk. Limiting intra-group exposures aims at reducing the concentration risk associated with the default of a group entity.

[117]            Whether a transaction is undertaken on an arm’s length basis is in principle judged according to all the circumstances of each particular transaction. The test to apply could be to consider whether a prudent person acting with due regard to their own commercial interests would have made such a transaction. Note, however, that Article 248(2) of the CRR provides that the EBA shall, in accordance with Article 16 of Regulation (EU) No 1093/2010, issue guidelines on what constitutes arm's length conditions.

[118]            Articles 16 to 22 of the BRRD provide the conditions and procedures for intra-group financial support. Currently Member States shall ensure that parents and subsidiaries may enter into agreements to help each other. Approval of such an agreement is subject to certain conditions (Article 19). As previously mentioned, in principle, an agreement to provide financial support may only be concluded if the supervisory authority considers none of the parties in breach of the rules on capital or liquidity, or is at risk of insolvency. However, it makes sense to specifically set out in legislation that for the deposit entity such agreements can by their nature jeopardize the liquidity or solvency of the deposit entity or create a threat to financial stability and therefore cannot be concluded. 

[119]            See Article 11 of the BRRD that provides that group resolution plans shall identify how the group resolution actions could be finance and, where appropriate, set out principles for sharing responsibility for that financing between sources of funding in different Member States.

[120]            This would have to be coordinated with the Commission's assessment of the appropriateness and the impact of imposing limits on exposures to shadow banking entities which carry out banking activities outside a regulated framework that need to be finalized by 31 December 2015. (See Article 395 of the CRR.)

[121]         The Commission Staff Working document from June 2 2010, SEC (2010) 669: "Corporate Governance in Financial Institutions: Lessons to be drawn from the current financial crisis, best practices." Accompanying document to the Green Paper: "Corporate governance in financial institutions and remuneration policies;" COM(2010) 284 final.

[122]            Current CRDIV rules provide that members of the management body have to possess "sufficient knowledge, skills and experiences" to perform their duties. The overall composition of the management board shall also reflect an adequately broad range of experiences. The number of directorships is limited so that board members of significant institutions cannot hold more than one executive directorship and two non-executive directorships or four non-executive directorships. Executive or non-executive directorships within the same group are counted as one.

[123]            If this rule is selected the term "sufficient" would have to be defined. The definition of "sufficient" is to some extent contextual and will depend on the strength of the fence and the overall objectives of the rule. This rule will not necessarily marry well with other rules and might even become redundant; for example in combination with b.

[124]            Any of a and b would require provisions on sanctions in case of non-compliance. I.e., Member States would have to ensure effective penalties in case of non-compliance.

[125]            Those objectives are: reducing moral hazard (excessive risk taking); facilitate resolvability; facilitate management, supervision; reducing conflicts of interest; reducing resource and capital misallocation; and reducing efficiencies.

[126]            Derogations to individual requirements could not be allowed which most likely will require some type of amendment to current rules.

[127]            The degree of control will depend on the management style of the parent company's executives and the share ownership structure.

[128]            The deposit entity may engage in certain activities that are not separated but involve a certain degree of trading; for example, underwriting which may necessitate some ancillary market making.

[129]            For sake of clarity intra-group lending and borrowing is not included in the definition of debt and can therefore take place.

[130]            With regard to ownership restrictions, see footnote 56.

[131] http://ec.europa.eu/internal_market/bank/docs/high-level_expert_group/report_en.pdf

[132] http://ec.europa.eu/internal_market/consultations/2012/hleg-banking/replies-summary_en.pdf

[133] http://ec.europa.eu/internal_market/financial-markets/derivatives/index_en.htm

[134] http://ec.europa.eu/internal_market/bank/regcapital/new_proposals_en.htm

[135] http://ec.europa.eu/taxation_customs/taxation/other_taxes/financial_sector/

[136] http://www.snl.com/. See Annex I for additional information on the sample.

[137] The use of annual reports published by individual banks is not retained since balance sheet items published in annual reports do not always use the same classification of financial instruments.

[138] 244 banks are part of EU 27 and 1 bank of Norway (DNB ASA) has been added in order to cover the EBA sample.

[139] No small banks are available in SNL, which means that these banks cannot be used for the calibration and consequently not been proposed for structural separation.

[140] Source ECB MFI Balance Sheets (GB, DK, LV, PL, SE + Eurozone countries)

[141] The only exception is Nykredit Realkredit, whose average 2006-2011 total assets amounts to roughly 170 bn EUR.

[142] http://www.eba.europa.eu/capitalexercise/2011-EU-Capital-Exercise.aspx

[143] Switching to a 3-year moving average has shown not to considerably change results (see Annex III).

[144] All graphs and tables are based on SNL data and JRC elaborations

[145] http://ec.europa.eu/internal_market/consultations/2012/banking_sector_en.htm

[146] See also Opinion of the European Banking Authority on the recommendations of the High-level Expert Group on reforming the structure of the EU banking sector, where on page 4 one can read “the EBA suggests that available for sale components of liquidity portfolios are excluded from the first threshold calculation”. http://www.eba.europa.eu/cebs/media/Publications/Other%20Publications/Opinions/EBA-BS-2012-219--opinion-on-HLG-Liikanen-report---2-.pdf

[147] AFS composition can be volatile. For instance for Lloyds in 2011 the share of government securities (good for liquidity constraint) in AFS is roughly 70% while in 2010 it was 30% (more ABS and corporate securities. Source: Lloyds annual report). This split is not available for most of the banks in the sample.

[148] Large volumes of securities and derivatives held for trading also signal the presence of market risk due to trading activities.

[149] The Commission Services are also considering the option of separating underwriting, as this activity is highly linked to market making. However, due to lack of time and data (limited in number of operations are available) underwriting could not be investigated.

[150] Also U.K. authorities recognizes the: "significant challenge of defining proprietary trading clearly and distinguishing it from other forms of trading activity [...] The difficulty that most of us witnesses is trying to distinguish pure proprietary trading from market-making" (see 3rd Report -Proprietary trading: http://www.publications.parliament.uk/pa/jt201213/jtselect/jtpcbs/138/138.pdf).

[151] Technical details on the clustering technique adopted are presented in Annex VI see http://www.eio.uva.es/inves/grupos/representaciones/trTCLUST.pdf for further references.

[152] In Annex III we will briefly discuss the stability over the years of the thresholds, considering four moving averages over a three years span: 2006-2008, 2007-2009, 2008-2010 and 2009-2011. In general we do not see many differences except for the 2008-2010 average.

[153] We made this choice since these thresholds might become part of legislation.

[154] For instance, banks such as Allied Irish, Bankia and Monte dei Paschi are part of this group.

[155] As an example, the changes of thresholds from definition 1 to definition 2 can be obtained as follows. Definition 2 excludes AFS, whose average in the sample is 15 bn EUR. Therefore the amount threshold is shifted down from 80 bn EUR to 65 bn EUR. On the other hand, the share of AFS over total assets accounts on average for around 8%, hence the threshold for the share dimension moves from 20% down to 12%.

[156] The graph refer to the average shares 2006-2011 and is based on the subset of banks in the sample for which information on deposits is available (approximately 180 banks). For all selected banks the information on deposits is available.

[157] Covered deposit are obtained from customer deposit applying an estimated correction factor equal to 0.54, averaging over EU27 data from ECB and from EU Deposit Guarantee Schemes.

[158] Financial Stability Board list of SIFI as of 2012. http://www.financialstabilityboard.org/publications/r_121031ac.pdf

[159] Note that the number of banks may slightly varies across the years and across the definitions

[160] Fixing the threshold to 80 b€ would allow to select banks such as Lloyds and Intesa Sanpaolo respecting the clusters structure, due to the large gap existing between 50 b€ and 100b €.

[161] www.SNL.com

[162] Source ECB MFI Balance Sheets EU27

[163] For each bank we consider the z-score of its absolute and relative trading activities. The z-score is a standardized measure which allows comparing variables with different scales. For each rescaled point we compute its geometric distance from the two rescaled thresholds.

[164] We assume that these thresholds are valid for all our five time frames, and we track the 10% of the banks having the smallest distances over these five average periods.

[165] To estimate the effect of EMIR, we assume a similar netting reduction of Derivatives Assets as the ones reported on the balance sheet of US commercial banks. The Office of the Comptroller of the Currency (OCC) reported in “OCC’s Quarterly Report on Bank Trading and Derivatives Activities” legally enforceable netting agreements allowed US (commercial) banks to reduce the gross positive fair value of derivatives by around 90%. Using precaution, we estimate that 80% of the derivatives assets will netted out of the balance sheet. Of these netted derivatives, 80% will be cleared through Clearing House due to EMIR. Combining both figure, 80% * 80% = 64% will be centrally cleared. Around 35% (100%-64%) are expected to remain on the balance sheet. For the FTT initiative we assume that there will be no impact on the securities and a moderate impact on the volumes of derivatives.

Source: Financial Stability Paper http://www.bankofengland.co.uk/publications/Pages/fsr/fs_paper18.aspx

[166] Sensitivity of results to the choice of the parameters reducing the derivative size is discussed in section IV.2 of this annex.

[167] The distance is the same as the one computed in Annex IV and it is computed on the average period 2006-2011. With respect to the analysis presented in Annex IV, the banks close to the thresholds are kept into two distinct classes in order to assess if there exist differences among them in terms of structural indicators or if instead they have very similar behaviour.

[168] The number of banks in OUT Borderline is by choice the same number of banks in IN Borderline’

[169] For example, the empirical strategy of the initial papers imposed the estimation of parametric cost functions (which are restrictive) and required constructing samples of banks with similar production techniques in order to yield sound estimates. Wheelock and Wilson (2011) improve on the estimation method by using non-parametric model of bank costs which does not require such sampling assumptions. An issue that is harder to address is that the sample size of very large banks is small and the statistical techniques employed are most accurate for average companies in the industry (see De Young, 2010). Therefore inference for large banks may not be very reliable.

[170] Furthermore, from a systemic perspective, they do not address whether the benefits of large size outweigh the potential costs in terms of systemic risk that larger firms may impose.

[171] See also Davies and Tracey (forthcoming).

[172] Speech given by Andrew Haldane at the Association of Corporate Treasurers, Leeds, Credit is Trust, 14 September 2009, pp 10-11

[173] Note that this relates not only to activities but also to geographical diversification. For example, international operations might provide bank managers with more possibilities to trade against the bank's interest (see Myers and Rajan (1998)).

[174]    Smith and Stulz (1985) show that the hedging of interest rate risk can increase firm value by lowering the expected transactions cost of bankruptcy

[175] A financial institution is classified as a conglomerate if its business include at least of the following activities: i) banking, ii) insurance and iii) securities.

[176] See http://www.europarl.europa.eu/sides/getDoc.do?pubRef=-//EP//TEXT+REPORT+A7-2013-0231+0+DOC+XML+V0//EN.

[177] See http://www.europarl.europa.eu/news/en/pressroom/content/20130318IPR06668/html/Time-for-a-wholesale-restructuring-of-banking.

[178] The SYMBOL model (SYstemic Model of Banking Originated Losses) has been jointly developed by the JRC, DG MARKT, and academic experts of banking regulation. For technical details see: De Lisa R, Zedda S., Vallascas F., Campolongo F., Marchesi M. (2011), Modelling Deposit Insurance Scheme Losses in a Basel 2 Framework, Journal of Financial Services Research, 40(3), 123-141. Using bank balance sheet data as input, the SYMBOL model operates in two steps: 1) estimation of an average default probability for the assets of any individual bank, by means of the features of the Basel FIRB (Foundation Internal Ratings Based) loss distribution function. 2) Monte Carlo simulation of the distribution of aggregate losses on the basis of individual banks' asset default probabilities. See also Appendix C at the end of this report and Appendix 4 to Annex XIII of the Impact Assessment of the Bank Recovery and Resolution Directive:http://ec.europa.eu/internal_market/bank/docs/crisis-management/2012_eu_framework/impact_assessment_final_en.pdf.

[179] See e.g. section 3 for further details on simulation methodology and section 4 for additional details on the estimation of the size and capitalisation of separated entities.

[180] In their report “Trading activities and bank structural separation: possible definitions and calibration of exemption thresholds” Commission Services have proposed different definitions in order to identify different trading activities, such as market making and proprietary trading. The sample analysed in this report is selected based on definition 3, focusing on market and counterparty risk, see section 2 for additional discussion, see Appendix A for a complete list. It should be noted that not all candidate banks would necessarily be separated: the choice of simulating separation for the whole sample seems however the best possible choice in the absence of further details.

[181] This implies that banks are capitalised in line with CRD IV requirements and losses are absorbed by holders of bail-inable debt upon default or undercapitalization. A conservative choice has however been made regarding the amount of bail-inable liabilities effectively used before intervention of a public backstop, in line with the minimum levels required in the draft bank recovery and resolution framework being discussed in trilogue at the time this study was conducted. See section 3.3 for details.

[182] If the structural reform also contributes to facilitating bail-in, part of the benefits found when evaluating the impact of Bank Recovery and Resolution Framework would be attributable to structural reform.

[183] As the large and complex banks might play an important role in providing credit to the economy and market liquidity, broader and indirect costs and benefits need to be considered in an overall impact assessment as well. These costs include effects on market liquidity, interest rates, asset prices and macroeconomic competition effects. A starting situation with highly indebted and leveraged banks, households, and governments may exacerbate any of these effects as high indebtedness together with lower asset prices and higher risk premia could lead to balance sheet effects. Many of the costs and benefits are extremely hard to quantify either because of their nature (e.g. enhanced resolvability and supervision of banks through transparency), or because of non-negligible cross-linkages and interactions across costs, risks and benefits and behavioural responses (e.g. changing risk taking incentives, new market players entering).

[184] Definition 3 focuses on market and counterparty risk by not considering available for sale securities and including gross volumes of securities and derivatives held for trading (see page 10 of the cited report for a more in-depth discussion). Although the sample of selected banks varies according to the definitions used for selection purposes, definitions 2, 3 and 4 of that report would all involve the same split of trading assets and liabilities for selected banks, so that a single analysis can be considered to cover all these three cases. See also footnote 292.

[185] Due to the limited aggregate assets of these small banks (0.25% of the total sample) simulation results are not materially affected by their exclusion.

[186] The balance sheet data available in Bankscope are not detailed enough to split the banks into the TE and the DTB, for instance Bankscope does not distinguish between derivative for trading and for hedging purposes.

[187] For additional details, see page 10 of the mentioned report. The exact procedure for estimating the final balance sheet composition and capitalization of the separated entities in illustrated in section 4.

[188] See section 4 for additional details.

[189] Separation along these lines could be difficult because e.g. legal obstacles to allocating certain assets or liabilities and/or because assets and liabilities could need to be divided in ways which do not allow the accounting identity to be respected in the two separate entities. For example, the amount of deposits could exceed the assets allocated to the DTB.

[190]De Lisa, R., Zedda, S., Vallascas, F., Campolongo, F., and Marchesi, M. Modeling Deposit Insurance Scheme Losses a Basel II Framework. Journal of Financial Services Research 40, 3 (2011).

[191] See Appendix C for an explanation of the SYMBOL methodology and the calculation of the implied obligor PD.

[192] The base SYMBOL methodology is more concerned with default (i.e. credit) risk than with market risk, to which the TE is more exposed: see Appendix D for additional details on using SYMBOL to simulate losses for the TE.

[193] The choice of a 99% confidence implies an increase in the riskiness of trading entities, while the choice of ignoring the difference between the holding horizon of trading securities (10 days) and the simulation horizon (1 year) implies a decrease. A precise quantification of the impact of these two drivers was not possible.

[194] Excess losses are those losses not absorbed by regulatory capital.  In addition to these excess losses, banks recapitalisation funding needs to meet Basel 3 8% minimum capital requirements are taken into account. This assumes that all banks considered in the sample represent systemic financial stability relevance.

[195] The relevance of contagion through the interbank channel is significantly reduced if bail-in in the context of the Bank Resolution and Recovery Framework is fully effective. For simulating contagion through the bail-in channel, outcomes would be extremely dependent on assumptions about the liability and exposure split due to the sensitivity of the contagion mechanism to the shape of the network matrix and uncertainty about future holdings of bail-inable securities.

[196] The estimated shares of banks activity inside the EU is reported in Appendix D for the largest banks representing 83% of the sample’s total assets.

[197] The advantage of taking as a baseline the gross losses of the separately simulated portfolios rather than simulating the undivided bank, is that the gross losses in the scenarios before and after separation are the result of a single set of simulations and are therefore less subject to statistical artefacts and implicit correlation settings. A simulation in which the undivided bank would be simulated as having one portfolio would implicitly assume a correlation of 1 between the losses in a financial crisis of the DTB portfolio and the TE portfolio (as they are run as one), rather than 0.5 when they are run separately. This makes the tail of the undivided bank much fatter than that of the separately run bank with pooled capital and LAC and it would thus not provide an adequate baseline.

[198] Total loss absorbing capacity is assumed to be equal to 8% Total Assets. This minimum requirement is in line with the minimum amount of private bail-in necessary to trigger intervention of a Resolution Fund in the position on the BRRD of the Council of June 2013 (http://www.consilium.europa.eu/uedocs/cms_data/docs/pressdata/en/ecofin/137627.pdf). Given that the introduction of a credible bail-in regime would trigger dynamical responses of banks and markets which cannot be fully modelled here, a conservative simplifying assumption is made that all liabilities which could become eligible for bail-in in excess of this limit will be covered by collateral or other guarantees. If more funds will be available for bail-in, losses on public finances and expected losses on bail-inable first bonds might be further reduced with respect to the results presented in the following.

[199] See Appendix C for details of these definitions.

[200] See Appendix E for details of this calibration

[201] These behavioural responses are calibrated on the basis of the FIRB formula (which is also the basis for SYMBOL simulations). The percentile to which capital covers losses at the universal bank is approximated to be the average of that of the DTB (99.9) and the TE (99), weighted with the relative shares of RWA. This is about 99.72.  

[202] See section 5.2.

[203] It should be noted that not all changes in the distribution of losses necessarily produce a benefit, as losses for some stakeholders might increase following separation (generating a cost), and that not all changes in the cost of capital necessarily produce a cost, as total weighted average cost of capital for some entities might decrease following separation (generating a benefit). It should also be noted that these results are not strictly additive, as variations in losses will translate into variations in the cost of exposed liabilities if we allow risk premia to be influenced by changes in expected losses.

[204] Strictly speaking, aggregate benefits are given by the overall reduction in gross and excess losses across all banks due to the behavioural response following structural separation. Lower losses and lower bail-in reduces the risk of contagion and facilitates resolution. Aggregate costs are measured as the overall reduction in revenues of the TE due to the reduced (risk weighted) assets.

[205] See footnote 292.

[206] This variable has a coefficient of 0.17 before the introduction of Basel 2.5 regulation changes in 2011.

[207] Following the report “Analysis of possible incentives towards trading activities implied by the structure of banks’ minimum capital requirements” the liability side of derivatives is included in the analysis. DTL are Derivative Liabilities Held for Trading and DHL are Derivatives Liabilities Held for Hedging Purposes (DHA).

[208] The coefficient obtained from the regression is 0.25, but its t-statistic is so low that it has been set to zero for the purposes of calculating RWAs. See the report cited in the previous note, section 3 and appendix A, for additional details on this procedure.

[209] First, estimated RWA are calculated for the TE and the DTB based on these coefficients. These RWAs are then re-normalized to sum to the RWA of the original undivided bank.

[210] http://www.eba.europa.eu/documents/10180/87706/EBA-BS-2012-037-FINAL--Results-Basel-III-Monitoring-.pdf/778804a5-8e3e-4073-83df-afd1be0b626e

[211] See footnote 5 for complete reference.

[212] These effects are not to be confused with those linked to the stricter definition of the quality of capital introduced by Basel 3.The change in the RWA due to the change in the definition of capital measures: (i) the effects of lower RWA for exposures that are included in RWA under Basel 2 but receive a deduction treatment under Basel 3; (ii) the increase in RWA applied to securitisation exposures deducted under the Basel 2 that are risk-weighted at 1250% under Basel 3; (iii) the increase in RWA for exposures that fall below the 10% and 15% limits for CET1 deduction.

[213] The costs of the recapitalisation of banks to Minimum Capital requirements is considered as all the banks of the sample are assumed to be systemic in case of distress.

[214] The costs of the recapitalisation of banks to Minimum Capital requirements is considered as all the banks of the sample are assumed to be systemic in case of distress.

[215] For the behavioural response with increased capital in the TE, gross losses for the trading portfolio are the same as in the baseline scenario, but there is a lower recapitalisation need to the MCR due to the higher initial capital.

[216] In case of efficient capital markets there would not be a social benefit in shifting losses from bail-in bonds to equity. However, as markets are not efficient and spill-over effects and risks of contagion in case of resolution and bail-in of bondholders are likely to remain substantially higher than those of losses on equity.

[217] See footnote 21.

[218] This should also be considered part of the behavioural response, as banks may try to keep risk constant for bail-in able bonds holders, as well as for equity holders, as capital is increased towards the 8% TA LAC requirement.

[219] Under a full applicability of the Modigliani-Miller theorem WACC (for the whole banking system) would not change if the risks on the asset side would not change. The allocation of losses across the banking system should equivalently not affect the total funding costs of the banking sector (as long as there is no shift of losses from the private to the public sector, and no change in the total asset portfolio). The increase in funding costs of the TE and the reduction in costs for the DTB should cancel out at unchanged balance sheets. Modigliani, F.; Miller, M., 1958,  "The Cost of Capital, Corporation Finance and the Theory of Investment". American Economic Review 48, n. 3

[220] Effect (v) does not impact funding costs but is considered in the overall cost assessment.

[221] Costs (i) and (ii) have a direct impact on the TE and DTB WACC. Effects (iii) and (iv) imply a benefit/costs to the banks creditors in case of a financial crisis. The change in the riskiness of the banks’ capital and BiB can affect the WACC if markets anticipate it.

[222] Weighted averages are calculated based on relative share of total assets in the final situation. I.e. the weights are 69% and 31% for behavioural responses 1 and 2 and 79% and 21% for behavioural response 3, as resulting from Table 3.

[223] These estimates of funding costs are broadly consistent with those presented in the UK Government’s impact assessment (IA) for the Financial Services (Banking Reform) Bill. Based on estimates provided by the major UK banks of the likely effect on their funding costs the UK Government uses for the purposes of the IA a range of minus 10 to 0bps for the changes in cost of subordinated, long-term unsecured and short-term unsecured debt in the ring-fenced banks and a range of 0 to 75bps for non-ring-fenced banks. Note that these estimates only apply to 30-35% of overall funding costs, so to compare to WACC impact that should be adjusted accordingly. Moreover, the UK IA does not assume bail-in to be effective in the baseline. This implies that the reduction in funding costs for the DTB is smaller (as it still benefits from implicit subsidies), and the high end of the increase in funding costs for the TE is likely to reflect loss of implicit subsidy. Also the UK IA does not analyse any behavioural responses to the balance sheets. http://www.official-documents.gov.uk/document/cm85/8545/8545.pdf (p31-56).

[224] Bain (2013) reports that the 10 largest EU banks had a return on RWA in 2009-2012 of 1.3%. Ranging across years between 0.9% and 1.7% on average. Bain estimates a return of 1.6% to 1.8% is required to cover the costs of capital. Taking a range between 0.9% and 1.8%, the reduction in revenues due to the reduction in RWA in the behavioural response 2 (reduced RWA) can be estimated to be 6-13 bn per year by multiplying the reduction in RWA by the return on RWA. This lower revenue would reduce the margin in case of BR2 by 9 to 18 bps. This may overestimate the actual effect on the profit margin as the marginal return on RWA is likely to be lower than the average return on RWA. Bain (2013) “European Banking Striking the right balance between risk and return” at  http://www.bain.com/publications/articles/european-banking-bain-report.aspx

[225] The risk premiums on senior unsecured debt and on subordinated debt reflect estimates for the average for EU large banking group.  The underlying data is taken from the EBA Weekly Overview of Liquidity and Funding (WOLF), April 2013. The risk premium on equity is based on the weighted average of the ROE over the period 1999-2011 for France, Germany, UK, Spain, Netherlands, Spain, Belgium and Italy which is 7%. Regulatory requirements reducing leverage tend to also reduce ROE. Source: Federal Reserve Economic Data: http://research.stlouisfed.org/fred2, Bank's Return On Equity, Percent, Annual, Not Seasonally Adjusted and AMECO data for GDP at current prices.

[226]The minimum LAC of 8% TA is assumed to be achieved by emitting subordinate debt (in addition to equity). This allows calculations with a clear hierarchy of bail-in losses, such that the subordinate debt absorbs all bail-in losses The calculation is conducted as if new capital was replacing residual bail-in capacity, and then by bringing back residual bail-in capacity to the desired level by substituting subordinated debt for senior unsecured debt (see next section): no real substitution of capital for bail-in capacity is considered: This allows to best isolate the two effects and prevents double counting.

[227]De Lisa, R., Zedda, S., Vallascas, F., Campolongo, F., and Marchesi, M. Modeling Deposit Insurance Scheme Losses a Basel II Framework. Journal of Financial Services Research 40, 3 (2011).

[228]In the current version of SYMBOL banks’ losses are obtained by Monte Carlo sampling from a correlated multivariate normal distribution via Choleski decomposition.

[229] Vasicek O. A. “Loan portfolio value”, Risk, December 2002, http://www.risk.net/data/Pay_per_view/risk/technical/2002/1202_loan.pdf.

[230] The common factor which drives the "internal value" of firms would also be driving their "market value" in case they were traded. Under the implicit assumption that all trading risk can be represented as risk on traded credits and bonds, the base mechanism of SYMBOL can be used to simulate a loss distribution based on both default and value risk. It should be noted that, while the standard representation of the Vasicek model does not explicitly model cash flows and operates on a pure net-present value (NPV) logic, their introduction would simply result in a change of the definitions of losses and a shift of the distribution leaving results unchanged. Similar models with additional risk factors or variations of the correlation structure have been proposed several times in the literature: Grundke (2004) proposes an integration of interest rate risk introducing a Vasicek term structure model; Kupiec (2007) introduces credit migration risk associated with different term structures by credit quality, based on the Vasicek interest rate term structure model and Johnston (2008) proposes an extension to pure equity investments considering dividend cash flows and a CAPM-style correlation structure. Grundke, Peter, 2004, “Integrating Interest Rate Risk in Credit Portfolio Models.” Journal of Risk Finance, vol 5, no. 2; Kupiec, Paul, 2007, “An Integrated Structural Model for Portfolio Market and Credit Risk.” in Berlin Conference on the Interaction of Market and Credit Risk; Johnston, Mark, 2009, “Extending the Basel II Approach to Estimate Capital Requirements for Equity Investments.” Journal of Banking and Finance, vol. 33, no. 6.

[231] Stahl, Gerhard, 1997, Three cheers. Risk Magazine, Vol. 10, pages 67–69

[232] Meaning no corrections except those for the switch to Basel III. See section 4 for additional details.

[233] A high correlation of losses across banks implies fatter tails of the loss distribution. 

[234] The three SYMBOL-simulated crises can, according to the SYMBOL model, be exceeded but with a very low probability: between 0.1% (99.9% simulation) and 0.01% (99.99% simulation). Under the first simulation there is 0.1% chance that the crisis will be bigger than estimated and the resolution framework will not be able to cope with it. In the second and third case the chances are 0.05% and 0.01% respectively. However, these probabilities are very much dependent on the SYMBOL model specifications and in particular of the accuracy of the probabilities in the Basel FIRB formula. Rather than relying on these probabilities of occurrence of a systemic crisis, the aggregate outcomes of the three considered simulations can be compared to the state aid used during the recent crisis.

[235] State aid for 2008-2011 comes from DG COMP state aid official reports (439.0 billion €): http://ec.europa.eu/competition/state_aid/studies_reports/expenditure.html, Table: State aid approved (2008 – Oct 2012) and state aid used (2008 – 2011) in the context of the financial and economic crisis to the financial sector (2008 - 2011), in billion Euro.

As DG COMP figures are yet not available for 2012, state aid for 2012 is based on DG MARKT elaborations derived from DG ECFIN survey with Member States' via the Economic and Financial Committee.

Note that used state-aid measurement is subject to two kinds of biases: all recapitalization support is included as expenditure (while a part may be considered a financial transaction if it is in exchange of valuable bank shares), and losses in any given year might be unrecognized. 

[236] A “notch” is a step in the list of credit ratings, such as going from AA+ to AA in the S&P and Fitch rating scale or from Aa1 to Aa2 in the Moody’s rating scale.

[237] Global Shadow Banking Monitoring Report 2013, 14 November 2013, FSB

[238] E.g. EU banks were estimated to have had a shortfall of stable funding of EUR 2.89 trillion in 2010.

[239] Final report of the High-level Expert Group on structural bank reforms chaired by E. Liikanen, 2.10.2012.

[240]  http://en.g20russia.ru/load/782795034.

[241] Regulation (EU) 648/2012 of 4.7.2012.

[242] Journal of Financial Market Infrastructure, Autumn 2012

[243] http://www.icmagroup.org/Regulatory-Policy-and-Market-Practice/short-term-markets/Repo-Markets/repo/latest/

[244] This number, however, includes double counting of transactions between the participating institutions and may not represent the real size of the market.

[245]http://www.ecb.europa.eu/pub/pdf/other/euromoneymarketsurvey201311en.pdf?e34259b291b21d9dee4bc45bcc611b95

[246] Credit Swiss, UBS, Goldman Sachs, Morgan Stanley Deutsche Bank, Barclays, Bank of America ML, HSBC and Citi. Source: JP Morgan Cazenove, Global Equity Research, 28.5.2012. JP Morgan figures are not disclosed.

[247] JP Morgan Cazenove, Global Equity Research, 25. 5.2012, page 20.

[248] Regulation (EU) 575/2013 of 26.6.2013.

[249] Directive 2004/39/EC 21.4.2004.

[250] Regulation (EC) 1287/2006 of 10.8.2006.

[251] Directive 2009/65/EC of 13.7.2009.

[252] European Securities and Markets Authority’s Guidelines for competent authorities and UCITS management companies on ETFs and other UCITS issues of18/12/2012 ESMA/2012/832EN.

[253] Directive 2011/61/EU of 8.6.2011.

[254] Directive 2009/44/EC of 6.5.2009.

[255] COM(2011) 656 final and COM(2011) 652 final.

[256] FSB, Policy Framework for Addressing Shadow Banking Risks in Securities Lending and Repos, 29.08.2013.

[257] ESRB, Towards a monitoring framework for securities financing transactions, March 2013.

[258] D. Duffie, How Big Banks Fail and What to Do about it, Princeton University Press, 2011, p. 39.

[259]http://www.rolandberger.ch/media/pdf/Roland_Berger_Collateralized_trading_business_in_new_realities_20121102.pdf

[260] BCBS, Consultative Document on Margin requirements for non-centrally-cleared derivatives, July 2012, p. 32: “Securities or funds collected as initial margin should not be rehypothecated or reused”.

[261] IMF, Singh M., Velocity of Pledged Collateral: Analysis and Implications, 2011.

[262]  Bank for International Settlements, Committee on the Global Financial System, The role of margin requirements and haircuts in procyclicality, March 2010.

[263] "Securities lending in physical replication ETFs: a review of providers' practices”, Morningstar, August 2012.

[264]  S. L. Schwarcz, Distorting Legal Principles, Duke University School of Law, May 2010, p. 5.

[265] These numbers are for cross-border funds, i.e. funds defined as generating their assets from more than one market (threshold used is 20%). Lipper "European Fund Market Review", 2012 and 2013 editions.

[266] 49 responses were received from trade associations, intermediaries, asset managers, market infrastructures as well as public authorities.

[267] Cf. European Commission: Green Paper Shadow Banking, Brussels, 19.3.2012 and European Parliament: Report on Shadow Banking (2012/2115(INI)), A7-03654/2012, 25.10.2012.

[268]See section 10 of this annex, “detailed overview of existing and proposed fund reporting requirements”, to have a detailed picture of the exiting SFT reporting requirements in the UCITS and AIFMD frameworks

[269] 18 European stakeholders responded to the section on transparency to fund investors.

[270] http://ec.europa.eu/internal_market/finservices-retail/docs/investment_products/20120703-proposal_en.pdf

[271]  FSB, Policy Framework for Addressing Shadow Banking Risks in Securities Lending and Repos, 29.08.2013.

[272]  Cf. Annex 5, http://ec.europa.eu/internal_market/consultations/2010/securities_en.htm.

[273]  COM(2011) 656 final, 20.10.2011.

[274]  FSB consultation on Strengthening Oversight and Regulation of Shadow Banking, 18.11.2012.

[275]  FSB consultations on Strengthening Oversight and Regulation of Shadow Banking, 18.11.2012.

[276] FSB consultations on Strengthening Oversight and Regulation of Shadow Banking, 18/11/2012