Mid-term Report - ALLOCATION OF FINANCIAL INTERMEDIATION SERVICES INDIRECTLY MEASURED (FISIM) - Mid-term Report Containing a qualitative and quantitative analysis of the first results of the trial calculations for allocating and calculating FISIM as described in the Council Regulation (EC) N° 448/98 of 16 February 1998

(presented by the Commission)

1. Summary

In its article 5, Council Regulation (EC) N° 448/98 completing and amending Regulation (EC) N°2223/96 with respect to the allocation of financial intermediation services indirectly measured (FISIM) within the European system of national and regional accounts (ESA) states: "the Commission, after consulting the Statistical Programme Committee, shall submit before 31 December 2000 a mid term report and before 1 July 2002 a final report to the European Parliament and the Council containing a qualitative and quantitative analysis of the implications of the trial methods for calculating and allocating FISIM as described in Annex III".

At this stage of the trial period, and according to the regulation, the Member States had to submit to the Commission their results for calendar years 1995 to 1999.

However, this mid-term report only includes the results obtained by thirteen countries, as two Member States have not yet provided their figures and comments to the Commission.

As it is necessary to revise the figures already obtained, to complete the exercise on more years, and to know the results of all Member States before being able to assess a strong opinion, only provisional observations can be presented. At this point of the trial period they are the following :

Being calculated as margins, allocated FISIM require the use of precise data. Despite the lack in direct sources for some data, Member States were able to implement the trial methods and obtained comparable and quite stable results. Therefore, the Commission considers that allocating FISIM is feasible if improvements continue to be made in data sources.

Eurostat has set up a FISIM Task Force, grouping people in charge of doing the trial exercise; these meetings enable to assess common practise for a better comparability and to discuss about data and methodological issues. In connection with the National Accounts Working Party, where the work of the FISIM Task Force and the results of the trial methods are regularly discussed, the Task force will continue its work to prepare the final commission report which shall he submitted before 1 July 2002 to the European Parliament and the Council.

2. The question to be solved in the national accounts and its impact on Gross Domestic Product (GDP)

Financial intermediaries explicitly charge commissions and fees to their customers. The measurement of the production and consumption of these services poses no special conceptual or practical problems.

But financial intermediaries are also able to provide services for which they charge implicitly by paying or charging different rates of interest to borrowers and lenders. They pay lower rates of interest than would otherwise be the case to those that lend them money and charge higher rates of interest to those who borrow from them. The resulting net receipts of interest are used to defray their expenses and provide an operating surplus. This scheme of interest rates avoids the need to charge their customers individually for services provided and leads to the pattern of interest rates observed in practice. However, in this situation, the national accounts must use an indirect measure, called financial intermediation services indirectly measured (FISIM), of the value of the services for which the intermediaries do not charge explicitly.

In principle, the FISIM output should be allocated among the various recipients or users of the services for which no explicit charge are made, and therefore in national accounts treated as intermediate consumption by enterprises, as final consumption by households, or as exports to non residents.

In practice, however, if may be difficult to find a method of allocating FISIM among different users in a way which is conceptually satisfactory from an economic viewpoint and for which the requisite data are also available.

In the trial calculations the allocation of FISIM among users is based on the difference between the actual rates of interest payable and receivable and a "reference" rate of interest.

This reference rate represents the pure cost of borrowing funds - that is, a rate from which the risk premium has been eliminated to the greatest extent possible and which does not include any intermediation services.

To illustrate the FISIM calculations, let us take two examples :

-A loan :

-Stock 100

-Interest receivable by the financial intermediary 5%

-Reference rate 2 %

-FISIM = 100 x (5% - 2%) = 3

-A deposit :

-Stock 100

-Interest payable by the financial intermediary 1,5 %

-Reference rate 2 %

-FISIM = 100 x (2 % - 1,5 %) = 0,5

In the ESA adopted in 1995 and the data supplied by Member States, by convention, the FISIM output is not allocated among users but is treated as absorbed by the intermediate consumption of a nominal sector and, therefore, the GDP is invariant to the size of the FISIM output, contrary to commissions and fees charged directly by financial intermediaries to users. But the relative importance of the part which is explicitly charged for as commissions and fees, relative to the part implicitly financed as FISIM, is different from country to country.

Solving the problem of allocating FISIM would lead to major improvements in the methodology of the European system of national and regional accounts (1995 ESA) and in particular to a more accurate comparison of GDP levels, which would include the entire value added generated by the activity of financial intermediaries.

The purpose of the Council regulation of 16 February 1998 concerning the allocation of FISIM within the ESA is to define such a methodology to be used on a harmonised basis by all Member States.

3. The methodology of allocating FISIM

FISIM has to be calculated and allocated only on loans and deposits granted by financial intermediaries (belonging to sectors S.122 other monetary financial institutions and S.123 other financial intermediaries, except insurance corporations and pension funds) to users sectors. This is because it is only the interest rates of those loans and deposits are controlled by the financial intermediaries (FIs).

FISIM is not allocated on securities other than shares, because when financial intermediaries buy or sell securities on the open market they are price takers who cannot control the prices at which they buy or sell and hence the interest rates they receive or pay.

The calculation of FISIM on loans and deposits should be made of the basis of the difference between the actual interest payable and receivable and a "reference" rate of interest.

Annex III of the Council Regulation 448/98 presents the required statistical data for the calculation and allocation of the FISIM produced. It is stated that, "it is necessary to use the table of average stocks of loans, deposits (split by user sectors) and the securities other than shares issued by FIs for the period (average of four quarters) and the accrued interest, after reallocation of interest rate subsidies to their actual recipients as defined by the 1995 ESA."

These statistical data will allow the test calculation of six internal reference rates, and the calculation of one external reference rate for the exports and imports of FISIM:

Methods of calculating the reference interest rate

Method 1 // Interest receivable on resident interbank loans, divided by stock of resident interbank loans.

Method 2 // Interest receivable on resident interbank loans and on securities other than shares issued by FIs, divided by stock of resident interbank loans and of securities other than shares issued by FIs.

Method 3 // Method 1 applied to short-term transactions.

Published rates for securities other than shares applied to long-term transactions.

Method 4a // Average of the average rates of credit and deposit interest applied to resident user institutional sectors (other than central banks)

Method 4b // Average of Methods 4a and 1.

Method 4c // Average of Methods 4a and 2.

Method for exports and imports // Weighted average interbank rate on loans and deposits between resident FIs and non resident FIs.

Stocks and interest on loans and deposits have to be broken down by user sectors.

Household sector in national accounts includes households as consumers and as producers to the extent they have unincoporated enterprises.

In order to identify part of the FISIM to be allocated to final consumption, and thus to evaluate the impact on the GDP, it is also necessary to have a further decomposition of the stocks and interest on loans granted to the households sector by distinguishing:

-Dwelling loans (intermediate consumption, as owners occupiers are treated in the system as the owners of enterprises engaged in the production of housing services for their own final consumption so that dwellings are not consumer durables).

-Loans to households as owners of unincorporated enterprises (intermediate consumption).

-Other loans to households (final consumption)

-A similar breakdown is necessary for the stocks of deposits and interest payable to households as individuals or as owners of unincorporated enterprises.

4. Data sources

4.1. Coverage of S122 and S123 sectors

FISIM to be allocated is produced by other monetary financial institutions (sector S122, which does not include the central bank) and by other financial intermediaries, except insurance corporations and pension funds (sector S123).

Information is more reliable on the S122 than on the S123 sector, but this last sector is much smaller.

4.2. Data available on stocks of loans and deposits, and interest flows

-In most cases the required information on stock of loans and deposits is available from direct sources, with a breakdown by main domestic users sectors. But the identification of households as consumers, or as owners of dwellings, or as owners of unincorporated enterprises, fundamental for the analysis of the FISIM impact on GDP, is not always feasible. More precisely, information on loans to households in their capacity as owners of dwellings is not always available, so as the share of the deposits made by households in their "consuming" function.

-The stock of loans and deposits with non-residents is required for both the calculation of exports and imports of FISIM. The Member States have more difficulties to obtain the necessary information for the calculation of imports of FISIM than exported FISIM, which leads to less reliable estimates for imported FISIM.

-For interest flows, less detailed information is available: in a majority of countries, information on interest flows is available from direct sources, but with no breakdown between institutional sectors, and it is therefore necessary to estimate interest rates for each institutional sector, apply these to the known stocks of deposits and loans to estimate interest flows in order to calculate this breakdown.

4.3. Methods 1 and 2: Match between assets and liabilities

In principle, between resident FIs of S122 and S123 sectors, the amount of interbank loans should be the same as the amount of interbank deposits. In practice, the total of loans held by resident FIs vis-à-vis the other resident FIs (and the corresponding interest payable) do not match the total of deposits held by resident FIs vis-à-vis the other resident FIs (and the corresponding interest receivable). This point is important for the calculation of methods 1 and 2. It was agreed that Member States should calculate the reference rate on the most reliable side: the loans side or the deposits side. If there is not a more reliable side, a weighted average of the two sides should be used.

4.4. Method 3 : split into short-term and long-term transactions

Several Member states are unable to calculate the internal reference rate for method 3 because of the absence of any adequate body of information on the maturity structure of the liabilities of FISIM producers. In other words, it is either impossible or difficult to make a reliable split between short and long term loans and deposits. Some Member States having made this calculation clearly indicated that method 3 gives less reliable results than the others.

5. Main Results

5.1. Comparison between non allocated FISIM as defined now and allocated FISIM

There are slight differences in the results of the methods used to calculate the total (non-allocated to users) FISIM as defined now and the sum of allocated FISIM output. In the results there is a small difference - positive or negative in Member States - which mainly depends on interest payable less receivable by financial intermediaries on securities other than shares, because securities other than shares are included in the calculation of the total FISIM, whereas the calculation of allocated FISIM is confined to loans and deposits (the reason being that only interest rates on loans and deposits are controlled by the FIs).

5.2. Output of the Central Bank

According to the FISIM regulation, the output of the Central Bank should no longer be calculated as for other financial intermediaries, but should be calculated as the sum of costs. Member States in their reports are in favour of this method.

5.3. Occurrence of negative FISIM

Negative FISIM can occur as allocated FISIM are calculated on the basis of the difference between interest receivable by FIs and the reference interest (for loans), and of the difference between the reference interest and interest payable by FIs (for deposits).

Most Members states are reluctant to have negative FISIM, putting forward the difficulty of presenting a service with a negative value. Other Member States admit the possible existence of negative FISIM, arguing that revenues of the S122/S123 sectors are not generated only by the differential of interest receivable/payable, but also by the fees and commissions.

Negative FISIM have been observed occasionally in the allocation of FISIM by user sectors, however FISIM results are always with all methods normally positive.

Only rare and sporadic negative results were obtained with all sectors, the sector towards which negative FISIM occur more often being the insurance corporations and pension funds sector (on the deposits side).

5.4. Analysis of the impact on GDP

5.4.1. Average impact on GDP by country

The comparison of average impact on GDP by country of allocating FISIM only gives a partial view of the importance of FIs in the economy, as FIs also directly charge commissions and fees which are already allocated to user sectors and therefore already taken into account in the GDP calculation.

In some countries a decreasing trend of FISIM can be observed in the period, because FIs directly charge commissions and fees for an increasing part of their output.

For the average of all years and for all methods, the allocation of FISIM would increase GDP by about 1,4% (average of the countries).

The following table shows, by country, the impact on GDP: average of the different methods and of the five years (or four years for Member States having not provided the 1999 results yet). This impact is split into:

-the increase of final consumption of households (including the non-profit institutions when this sector cannot be presented separately) corresponding to services produced by resident FIs

-the increase of final consumption of non-market services provided by the sectors General Government and NPISHs (when this last sector is presented separately) corresponding to services produced by resident FIs

-the exports of FISIM less the imports of FISIM for intermediate consumption of market producers (as FISIM imported for final consumption is neutral on GDP). Exports and imports of FISIM are not only calculated on the basis of loans and deposits between FIs and non FIs units, but also on interbank loans and deposits between resident FIs and non resident FIs which are treated as exports or imports depending on the sign of the results.

>TABLE POSITION>

About 80% of this impact is explained by an increase of final consumption of households, corresponding to FISIM allocated on loans and deposits of households as final consumers.

5.4.2. Average impact on GDP by internal reference rate

Impact on GDP (average of all years and all Member States)

>TABLE POSITION>

The impacts on GDP are quite similar whatever the internal reference rate used.

Method 1 based on the inter-bank rate corresponds in average to an increase of 1,26% of GDP.

Method 2, which also includes the rate of bonds, has in average for the years under study a slightly higher effect (1,39% of GDP).

Method 3 has a lower impact, but can not be implemented by several Member States as they can not break down loans and deposits according to their maturity. Moreover, as it introduces two reference rates (one for long term transactions, another for short term transactions) this method has the drawback to systematically eliminate the "matching benefits", which is the use by FIs of short term deposits to finance loans with longer maturity.

Method 4a measures only the gap between loan and deposit interest rates, neither of which has any measurable relationship to the pure interest concept (both include an element of service charge, but the size of this service element within each is unknown).

Methods 4b and 4c are a mixture of method 4a and of methods 1 and 2, respectively.

From a methodological point of view, the reference rate should measure the pure cost of funds appropriate to a borrowing or lending transaction, without any element of service charge or of risk. Therefore, methods 1 and 2 are conceptually better than methods 4, but methods 4 were introduced in the regulation because it was supposed they would bring more stable results.

5.4.3. Index of "volatility"

To test the stability of the results, the "volatility" in the results of each method has been measured as follows:

Index of "volatility" of the allocation of FISIM, based on the results by country and by method

>REFERENCE TO A GRAPHIC>

Average "volatility" by country (average of methods 1, 2, 4a, 4b, 4c)

The implementation of these calculations for each Member State highlights different levels of "volatility". For the average of the five methods (method 3 is not included as it could not be implemented by several countries), the average index of "volatility" is under 10% in Sweden, Austria and the Netherlands, and between 10% and 16% in Italy, Finland, Belgium, France, Denmark, Germany, Spain and the United Kingdom.

Average "volatility" of the different methods (average of Member States)

>TABLE POSITION>

Not surprisingly, method 4a is the less volatile. However, the difference with method 2 is not significant: 1%. At this stage of the trial period, it seems that method 2 is the best method as this method is more satisfactory from a conceptual point of view than method 4a. Also because method 2 represents a qualitative advance over method 1 as a rate applicable to transactions over a wide maturity band, therefore producing more stable results.

As the impact on GDP of allocating FISIM is on average 1,4%, this "volatility" index result (12%) means that - on average - the impact on GDP of the allocation of FISIM differs by 1,4% x 12% = 0,17% of GDP, which is the average difference of the impact on GDP between one year and the following year.

The average "volatility" comes both from stock changes (expressed in percentage of GDP) and from interest rate changes.

The index of "volatility" of stocks of loans and deposits having on impact on GDP (final consumption of households, NPISHs and general government, and exports) is 5%. This means that - if interest rates remained unchanged - the volatility of FISIM would be 5% and not 12%. Conversely, this means that the "volatility" of FISIM (12%) is due, in similar parts, to changes in the levels of stocks and to changes in interest rates.

The fact that the reference rate is sometimes either closer to the interest rates on loans or closer to the interest rates on deposits appears less relevant because the "volatility" of the method 4a, which avoids such a situation, is not significantly lower than the "volatility" of other methods, and particularly method 2.

6. General conclusion

Allocating FISIM requires further improvements in statistical sources in particular to directly measure interest flows broken down by institutional sector, to break down the households' deposits according to their purpose (intermediate or final consumption) and to obtain better measures of exports and imports of FISIM.

Preferences expressed by the Member States have put forward two criteria for the identification of relevant reference interest rates. These criteria are the need of having an interest rate fairly close to the concept of "pure interest rate" and giving the most reliable results with regard to available data. The analysis of the impact on GDP, at this stage of the study, tends to favour method 2, which seems to bring best results. Method 2 is followed by methods 1, 4c, 4b, 4a and 3.

Some Member States are reluctant to do all the trial methods for allocating FISIM as defined in the regulation, for practical and sometimes theoretical reasons, and consider that none of them are really more reliable for the correct measurement of economic activity than the present zero allocation.

As indicated in the summary of this mid-term report, the Commission thinks, at this point of the trial period, that allocating FISIM is feasible if improvements continue to be made in data sources.

It is anyway necessary to revise some figures already obtained, to complete the exercise on more years, to analyse FISIM at constant prices and by industry, and to pursue the work inside the FISIM Task Force and the National Accounts Working Party before being able to make a fully sound about the reliability of the methods of allocating FISIM as defined in the regulation.