1.

In these proceedings and in Case C‑116/16, just as in four parallel sets of proceedings on the Interest and Royalties Directive, ( 2 ) the Court has been asked to rule on the conditions under which a subsidiary that has paid dividends to its parent company can be refused exemption from withholding tax pursuant to Directive 90/435/EEC ( 3 ) (‘the Parent-Subsidiary Directive’).

2.

This case concerns ‘avoidance’ of withholding tax on intra-group dividends. Within the group, dividends are paid by a Danish operational company to its shareholder in Cyprus, which passes them on (as interest on a loan) to its shareholder in Bermuda, which in turn distributes them, again as dividends, to the group parent company resident in the United States. This arrangement has been put in place in order to enjoy the tax relief granted in the US on dividends which US parent companies repatriate to the US and use for particular purposes (research).

3.

The key question that arises here is how far a multinational group can go when configuring corporate structures to reduce final liability for withholding tax on dividend distributions within the Group. Where exactly does the dividing line fall between permissible tax arrangements and likewise legal, but abusive tax arrangements? When and based on what criteria can an abusive arrangement be presumed where a company resident in a third country establishes a subsidiary in a Member State (in this case Cyprus) which does not apply a withholding tax on dividend distributions? It is solely which enables dividends from European companies in the group to be collected there without any withholding tax and then to be passed on to the third country without any withholding tax.

4.

The above questions of law ultimately concern the fundamental conflict in tax law between the taxable person’s freedom to arrange his affairs under civil law and the prevention of arrangements that are valid under civil law but are nonetheless abusive under certain circumstances.

5.

The EU law applicable to this case is the Parent-Subsidiary Directive and Articles 43, 48 and 56 EC (now Articles 49, 54 and 63 TFEU).

6.

Article 1(1) of the Parent-Subsidiary Directive states that each Member State is to apply the directive to distributions of profits received by companies of that Member State which come from their subsidiaries of other Member States.

7.

Article 1 of the Parent-Subsidiary Directive provides:

‘(1)   Each Member State shall apply this Directive:

(2)   This Directive shall not preclude the application of domestic or agreement-based provisions required for the prevention of fraud or abuse.’

8.

Article 4 of the Parent-Subsidiary Directive provides:

‘(1)   Where a parent company or its permanent establishment, by virtue of the association of the parent company with its subsidiary, receives distributed profits, the State of the parent company and the State of its permanent establishment shall, except when the subsidiary is liquidated, either:

(2)   However, each Member State shall retain the option of providing that any charges relating to the holding and any losses resulting from the distribution of the profits of the subsidiary may not be deducted from the taxable profits of the parent company. Where the management costs relating to the holding in such a case are fixed as a flat rate, the fixed amount may not exceed 5% of the profits distributed by the subsidiary. ...’

9.

Article 5(1) of the Parent-Subsidiary Directive provides:

‘Profits which a subsidiary distributes to its parent company shall be exempt from withholding tax.’

10.

The Denmark-Cyprus Double Taxation Convention (DTC) of 26 May 1981 provides as follows in Article 10(1) and (2) on the distribution of the power to tax dividends:

‘1.   Dividends paid by a company which is a resident of a Contracting State to a resident of the other Contracting State may be taxed in that other State.

2.   However, such dividends may be taxed in the Contracting State of which the company paying the dividends is a resident and according to the laws of that State, but if the recipient is the beneficial owner of the dividends the tax so charged shall not exceed:

…

11.

It follows that the source State, in this case Denmark, can tax dividends paid to a parent company resident in Cyprus only at a low rate if that person is ‘the beneficial owner’ of the dividends. The concept of ‘beneficial owner’ is not defined in the DTC.

12.

According to the referring court, the legal situation in Denmark in the years at issue was as follows:

13.

Taxation of dividends paid by Danish parent companies is regulated in Paragraph 13(1) No 2 of the Selskabsskatteloven (Law on corporation tax) which, as promulgated in Danish Official Gazette No 111 of 19 February 2004 and amended by Law No 1375 of 20 December 2004, stated as follows for the 2005 and 2006 tax years:

‘Paragraph 13. The taxable income shall not include: ...

(2)   … Dividends that the companies or associations etc. referred to in Paragraph 1(1) Nos 1 to 2a, 2d to 2g and 3a to 5b receive on shares in companies within the meaning of Paragraph 1(1) Nos 1 to 2a, 2c to 2f and 3a to 5b or companies resident outside Denmark. However, this shall apply only if the dividend-receiving company, the parent company, owns at least 10% of the share capital in the dividend-distributing company, the subsidiary, for a consecutive period of at least one year within which period the time of distribution of dividends is to be. The aforementioned share is, however, 20% if distributions were made in the 2005 and 2006 calendar years and 15% if distributions were made in the 2007 and 2008 calendar years. ...’

14.

The limited tax liability of foreign companies for dividends is regulated in Paragraph 2(1)(c) of the Law on corporation tax. The limited tax liability in 2005 and 2006 ultimately did not cover dividend distributions to a parent company on which no or reduced tax is charged pursuant to the Parent-Subsidiary Directive or a DTC. However, that only applies if it owns at least 20% of the share capital in the subsidiary for a consecutive period of one year, which must include the distribution date.

15.

A distribution from a Danish company to a Cypriot parent company (regarded as the ‘beneficial owner’ of the dividends) is thus exempt from tax under Paragraph 2(1)(c) of the Law on corporation tax, since a reduced rate applies under the DTC.

16.

If, on the other hand, there is limited tax liability on dividend distributions from Denmark under Paragraph 2(1)(c) of the Law on corporation tax, the Danish dividend payer is required under the Danish Law on withholding tax ( 4 ) to withhold tax at a rate of 28%. In the event of late payment of the tax withheld (where there is limited tax liability), interest is charged on the tax due. The default interest is payable by the person required to withhold tax.

17.

In the years 2005 and 2006 there was no general statutory rule for the prevention of abuse. However, the ‘reality doctrine’ established in case-law requires tax to be assessed on the basis of a specific analysis of the facts. This means, for example, that fictitious and artificial tax arrangements may be disregarded under certain circumstances and tax may be assessed instead based on a ‘substance-over-form’ approach. It is common ground that the reality doctrine does not provide a basis on which to disregard the legal transactions conducted in the present case.

18.

The concept of ‘rightful income recipient’ has also been established in Danish case-law. That concept is based on the fundamental provision on income taxation in Paragraph 4 of the Statsskatteloven (Danish Tax Code), which states that the tax authorities are not required to accept an artificial separation between the income-generating business/activity and the allocation of income deriving therefrom. It is therefore necessary to determine who — irrespective of the purported corporate structure — is the actual recipient of certain forms of income and therefore has a tax liability. The question is thus to whom the income is to be allocated for tax purposes. The ‘rightful income recipient’ will thus be the person who is the taxable person for the income in question.

19.

The starting point is the challenge by Y Denmark Aps (Y Denmark) to a finding that it is liable for withholding tax not retained on dividend distributions to its Cyprus-resident parent company (Y Cyprus). It assumed that the distribution was exempt from withholding tax under the Parent-Subsidiary Directive. However, the tax authorities take the view that the dividends were in fact paid to the ‘grandparent company’ resident in Bermuda, i.e. to Y Global Ltd. (Bermuda) (Y Bermuda), as Y Cyprus is simply a conduit company. Therefore withholding tax should have been retained. The facts underlying the dispute are as follows:

20.

The parent company at the top of the Y Group, Y Inc., USA (hereinafter Y USA) is a company listed in the USA. Y USA’s foreign subsidiaries are currently owned by Y Bermuda, whose sole activity — apart from acting as a holding company — is to own certain intellectual property rights attaching to the group’s products. Its day-to-day business is conducted by an (independent) management company. Y Denmark, which was established by Y USA in 2000, has since then had around 20 employees on an ongoing basis and provides sales and support services. Y Denmark also functions as a holding company for the European branch of Y Group, for example, for Y Netherlands.

21.

When US legislation was amended by the American Job Creation Act of 2004, US companies were granted the facility to repatriate dividends from foreign subsidiaries on very favourable tax terms, provided that they undertook in return to appropriate those dividends in the US for particular purposes, such as research and development.

22.

On that basis, Y USA decided to repatriate as large a dividend as possible from Y Bermuda (its wholly-owned subsidiary) for the 2005/2006 fiscal year (1 May 2005 to 28 April 2006). The dividend, to be created in part through dividend distributions by various subsidiaries to Y Bermuda, was 550 million US dollars (USD).

23.

Before the dividends were distributed, the European part of the group was restructured. Thus, Y Bermuda incorporated Y Cyprus on 9 May 2005. Y Bermuda sold to it Y Denmark’s shares. The acquisition price was paid by issuing a bond. Thus Y Cyprus was inserted between Y Bermuda and Y Denmark.

24.

Y Cyprus acts as a holding company with certain treasury activities (loans to subsidiaries). The company, which has no staff, has the same address as a management company. It follows from the annual reports in the financial statements for 2005/2006 and 2006/2007 that the primary business of Y Cyprus is to act as a holding company and that its management board members were paid remuneration of USD 571 and USD 915. According to the financial statements, USD 0 in tax was paid, because the company did not have positive taxable income.

25.

On 26 September 2005, Y Netherlands decided to distribute a dividend of EUR 76 million to Y Denmark for the 2004/2005 fiscal year. On 28 September 2005, Y Denmark’s general meeting approved a proposed dividend distribution to Y Cyprus for that fiscal year, which was also EUR 76 million. The dividend was paid to Y Denmark on 25 October 2005. On 27 October 2005, the dividend of equal amount was paid by Y Denmark to Y Cyprus, which passed it on to Y Bermuda on 28 October 2005 to repay the loan which it had contracted in connection with its acquisition of Y Denmark. On 3 April 2006, Y Bermuda distributed a dividend of USD 550 million to Y USA. Y Bermuda funded the dividend partly from equity and partly from a bank loan. On 13 October 2006, Y Denmark’s general meeting approved a further proposed dividend to Y Cyprus of 92 million Danish Crowns (DKK) for the 2005/2006 fiscal year.

26.

By notice dated 17 September 2010, the SKAT (Danish tax authority) found that Y Denmark should have retained the withholding tax on the dividends distributed to its parent company Y Cyprus in 2005 and was liable for the withholding tax.

27.

That notice was appealed to the Landsskatteret (Tax Appeals Commission), which upheld the position taken by the SKAT (Danish tax authority) in its decision of 16 December 2011, inasmuch as Y Cyprus was not the ‘beneficial owner’ of the dividends under the Denmark-Cyprus DTC, but agreed with Y Denmark that there was no basis on which to retain withholding tax, as Y Cyprus was covered by the exemption under the Parent-Subsidiary Directive.

28.

The Skatteministeriet (Ministry of Finance) lodged an appeal against the Landsskatteretten’s (Tax Appeals Commission) decision before the Østre Landsret (High Court of Eastern Denmark, Denmark). The Østre Landsret (High Court of Eastern Denmark) has now decided to make an order for reference.

29.

The Østre Landsret (Denmark) has referred the following questions for a preliminary ruling:

30.

Cases C‑116/16 and C‑117/16 were joined by order dated 13 July 2016. Written observations on the questions referred were submitted to the Court of Justice in the joined proceedings by T Danmark, Y Denmark Aps, the Kingdom of Denmark, the Federal Republic of Germany, the Kingdom of Sweden, the Italian Republic, the Kingdom of the Netherlands and the European Commission. T Danmark, Y Denmark Aps, the Kingdom of Denmark, the Federal Republic of Germany, the Grand Duchy of Luxembourg and the European Commission attended the hearing on 10 October 2017, which also included Cases C‑115/16, C‑118/16, C‑119/16 and C‑299/16.

31.

The parties to the proceedings do not dispute that, in principle, the relevant dividend payments fall within the Parent-Subsidiary Directive. It follows that Denmark, as the country in which the company making the distributions is resident, should exempt the dividends from withholding tax in accordance with Article 5 of the directive. However, Denmark regards its refusal to grant exemption from withholding tax as manifestly covered by Article 1(2) of the Parent-Subsidiary Directive. According to that provision the directive shall not preclude the application of domestic provisions required for the prevention of fraud or abuse.

32.

By its Questions 1 to 5, the referring court primarily asks if a Member State can only rely on Article 1(2) of the Parent-Subsidiary Directive to prevent fraud and abuse if it has adopted a domestic provision implementing it (B.1) and, if so, if Paragraph 2(2)(c) of the Danish Law on corporation tax or a rule in a DTC that uses the term ‘beneficial owner’ can be treated as sufficient transposition thereof (B.2). If that is the case, the referring court asks how the concept of beneficial owner should be interpreted and by whom.

33.

These questions only make sense if the requirements of Article 1(2) of the Parent-Subsidiary Directive are in fact fulfilled. That provision requires that there be, on the part of Y Cyprus, fraud or abuse of the exemption from withholding tax in the present case. Therefore Question 5 must be answered first.

34.

In that respect, I will explain the criteria for presuming abuse within the scope of the Parent-Subsidiary Directive (2). First, however, I will investigate the scope of the exemption from withholding tax under Article 5(1) of the Parent-Subsidiary Directive.

35.

As is apparent from the third recital thereof, the Parent-Subsidiary Directive seeks, by the introduction of a common tax system, to eliminate any disadvantage to cooperation between companies of different Member States as compared with cooperation between companies of the same Member State and thereby to facilitate the grouping together of companies at EU level. The directive seeks thus to ensure the neutrality, from the tax point of view, of the distribution of profits by a subsidiary established in one Member State to its parent company established in another Member State. ( 5 )

36.

To that effect, Article 4(1) of the Parent-Subsidiary Directive leaves the Member States a choice between two systems, namely between a system of exemption and one of deduction. In fact, in accordance with recitals 7 and 9 of that directive, where a parent company by virtue of its association with its subsidiary receives profits distributed otherwise than on the liquidation of that subsidiary, the Member State of the parent company must either refrain from taxing such profits in so far as they cannot be deducted by the subsidiary and tax them in so far as the subsidiary can deduct them, or tax such profits while authorising the parent company to deduct from the amount of tax due that fraction of the corporation tax paid by the subsidiary and any lower-tier subsidiary which relates to those profits. ( 6 )

37.

Thus, Article 4 of the Parent-Subsidiary Directive seeks, in respect of profits distributed to a resident parent company by a non-resident subsidiary, to avoid that subsidiary being taxed thereon in its State of establishment first and the parent company then being taxed on the same profits in its State of establishment. ( 7 )

38.

Article 4 of the Parent-Subsidiary Directive concerns economic double taxation, because the dividends are, as a rule, paid from the subsidiary's taxed income (i.e. income on which corporation tax has been paid in a Member State) and are part of the income of the parent company (and are thus subject to corporation tax again in another Member State). Thus, within large groups, the tax liability depends on the number of tiers in the group, which in most cases is based purely on organisational reasons. Thus, Article 4 of the Parent-Subsidiary Directive takes account of the fact that legal entities can be duplicated any number of times without changing the persons behind them and, by extension, their profits from business conducted via those legal entities.

39.

In order to ensure fiscal neutrality, Article 5(1) of the Parent-Subsidiary Directive goes further by exempting from withholding tax profits which a subsidiary distributes to its parent company. ( 8 ) Thus, in order to prevent double taxation, Article 5(1) of the directive lays down a general principle prohibiting withholding tax on profits distributed by a subsidiary resident in one Member State to a parent company resident in another Member State. ( 9 )

40.

By prohibiting Member States from imposing withholding tax on the profits distributed by a resident subsidiary to its non-resident parent company, Article 5(1) of the Parent-Subsidiary Directive limits the powers of the Member States to tax profits distributed by companies that are resident in their territory to companies resident in another Member State. ( 10 ) Therefore, the Member States cannot unilaterally adopt restrictive measures and make the entitlement to exemption from withholding tax provided for in Article 5(1) contingent upon various requirements. ( 11 ) Therefore, the entitlement to exemption from withholding tax does not depend on the owners of the parent company being resident or on the dividends payer disclosing how the dividends recipient will use the dividends.

41.

Article 5(1) of the Parent-Subsidiary Directive aims to prevent further (this time, legal) double taxation. As the Court has already ruled, withholding tax actually taxes the recipient of the income (in this case, of the dividends). ( 12 ) Thus, withholding tax in the dividends payer’s State of residence is simply a particular taxation technique, rather than a type of tax. If withholding tax is paid by the payer based on its place of residence and ‘normal’ tax is paid by the dividends recipient based on its place of residence, that in itself results in double taxation and, as a rule, puts them at a disadvantage compared to national companies.

42.

Precisely in the case of complex group structures spanning several countries, the cascade effect referred to above would be duplicated if there were no exemption at both levels and withholding tax were to apply each time. Obviously, this would undermine the internal market.

43.

However, it is irrelevant, for the purpose of preventing such cascading economic and legal double taxation, whether the dividends recipient is also the ‘beneficial owner’ of the dividends or suchlike. The decisive question is whether the dividends payer was charged corporation tax and the dividends recipient also has to pay corporation tax on the dividends. The same applies where withholding tax is prohibited. The key question there is whether the dividend income is subject to corporation tax in the State of residence.

44.

In that regard, it makes perfect sense that (unlike the Interest and Royalties Directive) ( 13 ) the Parent-Subsidiary Directive is ‘only’ predicated on the distribution of profits by a subsidiary to its parent company (which must have a certain minimum holding). Unlike interest payments, dividends do not, as a rule, represent operating expenditure which may be set against profit; therefore, it makes sense that, according to its wording, the Parent-Subsidiary Directive does not contain any further substantive criteria (such as drawing of dividends in one’s own name and on one’s own account or suchlike).

45.

Dividend rights ultimately follow from the company’s status as parent company under company law, which can only be enjoyed in its own name. The very possibility of acting on a third party’s account seems hardly conceivable here. In any event, it cannot be deduced merely from the fact that a ‘grandparent company’ exists. In principle, therefore, all dividend distributions by a subsidiary to its parent company in another Member State are covered if the company fulfils the requirements of Article 2 of the Parent-Subsidiary Directive, which is not contested in the present case.

46.

Limits to this are set only by Article 1(2) of the Parent-Subsidiary Directive, which provides that the directive does not preclude the application of domestic or agreement-based provisions required for the prevention of fraud or abuse.

47.

Article 1(2) of the Parent-Subsidiary Directive reflects the general principle of EU law that EU law cannot be relied on for fraudulent or abusive ends. ( 14 ) The application of this rule of EU law cannot be extended to such an extent as to cover abusive practices by economic operators, i.e. transactions that are carried out not in the context of normal commercial transactions, but solely for the purpose of wrongfully obtaining advantages provided for by EU law. ( 15 )

48.

The wording of the provision adds nothing to the understanding of abuse that is its basis. ( 16 ) However, as an exception, Article 1(2) of the Parent-Subsidiary Directive needs to be interpreted strictly. ( 17 ) With regard to measures to prevent abuse, this is demanded in particular by the principle of legal certainty. If, in terms of form, an individual meets all the conditions for claiming a right, this right may be denied on grounds of abuse only in particular cases.

49.

However, relevant pointers for assessing abuse follow from other EU directives. For example, the Mergers Directive ( 18 ) refers in the second sentence of Article 11(1)(a) to an absence of valid commercial reasons for the operation as a typical example of such motivation. Furthermore, Article 6 of the Directive laying down rules against tax avoidance practices ( 19 ) (‘Directive (EU) 2016/1164’), which was not yet in force in the years at issue, defines the concept of abuse. The criterion is whether a non-genuine arrangement has been put into place for the main purpose or one of the main purposes of obtaining a tax advantage that defeats the object or purpose of the applicable tax law. According to Article 6(2), an arrangement is regarded as non-genuine to the extent that it was not put into place for valid commercial reasons which reflect economic reality.

50.

Last but not least, the Court has held on various occasions that for a restriction of freedom of establishment to be justified on grounds of the prevention of abusive practices, the specific objective of such a restriction must be to prevent conduct involving the creation of wholly artificial arrangements which do not reflect economic reality, with a view to escaping the tax normally due on the profits generated by activities carried out on national territory. ( 20 ) As the Court has also since held on various occasions, it suffices if the arrangement is put in place not with the sole aim, ( 21 ) but with the essential aim, of obtaining a tax advantage. ( 22 )

51.

This case-law of the Court contains two mutually contingent elements. First, wholly artificial arrangements which ultimately only exist on paper are refused recognition a priori. Furthermore, decisive importance attached to circumvention of tax laws is also achievable by arrangements that exist in commercial reality. The latter group of cases may be the more frequent and is now expressly covered by the new Article 6 of Directive 2016/1164. The Court too has held in a more recent judgment that the wholly artificial nature of the arrangement was just one fact that suggested that the essential aim was to obtain a tax advantage. ( 23 )

52.

Abuse is determined from a general examination of all the circumstances of the individual case, which it is for the competent national authorities to make and which must be open to review by the courts. ( 24 ) It is for the referring court to conduct that general examination. ( 25 ) However, the Court can give the referring court some useful pointers ( 26 ) for the purpose of determining if the transactions are being carried out in the context of normal commercial transactions or solely for the purpose of wrongfully obtaining advantages provided for by EU law. ( 27 )

53.

The Court cannot judge whether a wholly artificial arrangement which does not reflect economic reality can be presumed in the present case. First, the facts disclosed by the referring court do not suffice for that. Second, it is for the referring court to appraise those facts. The Court can only give some pointers:

54.

A wholly artificial arrangement which does not reflect economic reality might be presumed in the present case. The facts disclosed by the referring court support that presumption. Thus, Y Cyprus has no staff and apparently no office premises of its own either. As a result, the company does not incur costs for either staff or premises. Also, the remuneration paid to the members of the management board suggests little activity on their part. Furthermore, asset management activities clearly generated no income of its own for the company. This all appears to be artificial. A natural person would have ceased trading long ago under such circumstances.

55.

Even though the Court found recently that the fact that the activity consists in the management of assets and the income results only from such management does not mean that a wholly artificial arrangement exists which does not reflect economic reality, ( 28 ) there is doubt here as to whether the activities of the Cypriot company may well take place solely on paper, given that even the treasury function of the company does not generate any income.

56.

In light of the fact that asset management companies in particular (may) engage per se in little activity, this criterion is also subject to very minor requirements. If a validly incorporated company does not even have tangible and human resources at its disposal on site to achieve its object (in this case treasury activities) on its own, there would certainly be cause to see it as an arrangement that does not reflect economic reality. This applies in particular if it is structurally unable to generate income of its own that would enable it to do so.

57.

In my view, a legal entity that is passive to the point that any conceivable involvement in transactions is, at most, via third parties and that develops no business of its own from which its own income and costs result is a wholly artificial arrangement. Ultimately, however, that is a question of fact on which it is for the referring court to rule.

58.

Furthermore, notwithstanding that assessment of the facts, there may be an abusive tax arrangement even outside a wholly artificial arrangement that does not reflect economic reality, as the wording of the new Article 6 of Directive 2016/1164 suggests. Thus, there may be other criteria that are key factors in the present case, especially the non-fiscal reasons to be considered.

59.

In that regard, the Court has already ruled in connection with the Parent-Subsidiary Directive that holdings structures whose sole purpose is to benefit from the tax advantages provided for under the directive are a form of abuse. ( 29 ) Hence, also in the case of the Parent-Subsidiary Directive, there must be economic reasons for the structure. Arrangements which seek a tax advantage only and have nothing to do with economic reality are not protected. ( 30 )

60.

Thus, decisive importance must be attached to other criteria in the present case, especially the non-fiscal reasons that must be taken into account.

61.

According to the case-law of the Court, the fact that either the registered office or real head office of a company was established in accordance with the legislation of a Member State for the purpose of enjoying the benefit of more favourable legislation does not, in itself, constitute abuse. ( 31 ) The mere fact that Cyprus companies were interpolated in the chain of holdings does not, therefore, automatically mean that abuse must be presumed.

62.

Furthermore, where the taxable person has a choice between two possibilities, he is not obliged to choose the one which involves paying the higher amount of tax but, on the contrary, may choose to structure his business so as to limit his tax liability. ( 32 ) Thus, again according to the Court, taxable persons are generally free to choose the organisational structures and the form of transactions which they consider to be most appropriate for their economic activities and for the purpose of limiting their tax burdens. ( 33 ) The sole fact that in the present case a business structure was chosen which did not result in the greatest tax burden (here an additional and final taxation at source) thus also cannot in itself qualify as abuse.

63.

Furthermore, other than in the case of a wholly artificial arrangement that does not reflect any economic reality, the fact that a Union citizen, whether a natural or a legal person, simply sought to profit from tax advantages available in a Member State other than his State of residence cannot in itself deprive him of the right to rely on the provisions of the Treaty. ( 34 ) Thus a transaction structure involving a Member State that waives withholding tax, as in the present case, cannot of itself be seen as abusive.

64.

To that extent, freedom of establishment includes the choice of Member State which, in the opinion of the undertaking concerned, offers the best tax situation. If that principle applies to highly harmonised VAT laws, ( 35 ) it applies a fortiori to less harmonised income tax laws, where acceptance of the differences between the tax regulations ( 36 ) of the individual Member States is intentional under EU law or a matter of conscious political acceptance.

65.

Furthermore, the Court has clarified that the tax exemption for dividends provided for under EU law is not contingent upon the origin or residence of the shareholder, as that is irrelevant for the purposes of the Parent-Subsidiary Directive. ( 37 ) Therefore, the fact that the shareholder of Y Denmark is resident in Cyprus or the shareholder of its parent company is resident in a third country (in this case Bermuda) is therefore also not abusive, when taken in isolation.

66.

However, a more significant factor is that the ultimate payments recipients are often registered in particular third countries (as a rule on small islands such as the Cayman Islands, ( 38 ) Jersey ( 39 ) or, as in this case, Bermuda), which are renowned for refusing to cooperate with other tax authorities. This may suggest an unusual overall approach, the economic reason for which is not immediately apparent.

67.

Therefore, in the present case as an abusive arrangement might be seen in the overall construction more owing to the ‘establishment’ of one of the ultimate payments recipients in particular third countries (in this case, Bermuda) than in the ‘interpolation’ of a Cypriot company. Particular importance here is attached to the purpose of the arrangement or the objective of the tax law circumvented (in this case, taxation in Denmark).

68.

First, it must be noted that Denmark has not been deprived of taxes on the profits of the operational company acquired (Y Denmark). Those profits were duly taxed in the State of residence (i.e. in Denmark). The dividends were therefore subject to Danish corporation tax.

69.

The Cypriot company has unlimited tax liability in Cyprus and is subject in Cyprus to corporation tax on its income. The fact that it had no positive income in the disputed years does not change that. Thus the requirements of Article 2 of the Parent-Subsidiary Directive are fulfilled. Exemption from tax on dividends in Cyprus is in keeping with the purpose of the directive and takes account of Danish corporation tax already paid.

70.

In that regard, the fact that Cyprus does not impose a withholding tax on dividends paid to shareholders in third countries is immaterial. That decision is a consequence of the fiscal autonomy of each individual State. If fiscal competition between Member States is admissible under EU law due to the lack of harmonisation of income taxes, a taxable person cannot be blamed for availing himself in reality (i.e. not just on paper) of the tax advantages offered by certain Member States.

71.

In the final analysis, the interpolation of the Cypriot company ultimately ‘only’ avoids tax at source on dividend payments in Denmark. As stated previously (in point 41), withholding tax actually taxes the recipient of the income (in this case, of the dividends). ( 40 ) That is achieved by having the payer withhold part of the income at source at the time of disbursement.

72.

Thus, tax at source in the dividends payer’s State of residence is simply a particular taxation technique, rather than a type of tax, intended essentially to secure (minimum) taxation of the dividends recipient. In cross-border cases in particular, proper taxation of the recipient’s income is not always ensured. As a rule, the dividends recipient’s State of residence will rarely be aware of his income from abroad, unless functioning data exchange systems exist between the tax authorities (as they do now in the Union).

73.

Therefore, two requirements must be fulfilled for an arrangement to qualify as abusive circumvention of this objective of the law (to ensure the dividends recipient is taxed). First, in the case of direct disbursement, tax must in fact be chargeable in Denmark (see point 88 et seq.). Second, there must be a risk that the income will not be declared in the actual State of receipt and thus will not be taxed.

74.

If, therefore, one reason for choosing a particular business structure is to pay dividends to investors via a third country in order to prevent their States of residence from obtaining information on their income, then that overall arrangement should, in my opinion, qualify as abuse of the law.

75.

Any such complaint of abuse might, in turn, be rebutted if the capital investment companies provide the relevant tax information to the investors’ States of residence or if the State of residence of the capital investment companies has the information in question and forwards this information to the relevant States. Any such corporate structure would not then circumvent the purpose of the tax at source avoided (see point 72 above). That too must be included by the Court in its overall examination.

76.

If the purpose of the arrangement were to bundle the dividends of the European group companies in a tax-neutral manner and then channel them to the parent company, which pays tax on them in the normal manner in its State of residence (in this case, the United States), a presumption of abuse seems to be rather hard to assume. That would apply in particular where no withholding tax would be charged on a direct distribution to the group parent company in the US, because a DTC exists to that effect.

77.

Where withholding tax is avoided on dividend payments to companies resident in third countries, the primary issue is avoidance by the actual dividends recipients (i.e. the investors) of tax on the dividends. Abuse may be assumed to exist here, in particular, if the corporate structure chosen is designed to take advantage of a lack of information exchange between the States involved to prevent the effective taxation of those shareholders. This is a matter which must ultimately be assessed by the referring court.

78.

By its third and fourth questions, the referring court asks, inter alia, whether refusal to grant the exemption from withholding tax provided for in the Parent-Subsidiary Directive under the terms of an international convention concluded between Denmark and another State (i.e. a DTC) has to be based on a basic understanding in conformity with EU law that is subject to review by the Court. It also wishes to know whether such an interpretation in conformity with EU law should take account of the commentaries on the OECD MTC and, if so, whether subsequent commentaries on an OECD MTC that postdate the directive should be taken into account in the interpretation.

79.

In the subsequent commentaries on the OECD MTC, conduit companies are not normally regarded as the beneficial owner if, though the formal beneficial owner, they have, as a practical matter, very narrow powers which render them, in relation to the income concerned, a mere fiduciary or administrator acting on account of the interested parties.

80.

Inasmuch as a Member State wishes to restrict a tax exemption pursuant to EU law to the detriment of the individual, that restriction must be interpreted in light of EU law. Therefore, the Parent-Subsidiary Directive needs to be interpreted in order to give the referring court a helpful answer. The OECD Model Tax Convention and the commentaries on the OECD Model Tax Convention might be incorporated into that interpretation.

81.

However, OECD MTCs are neither EU law nor legally binding on the Court. They are not multilateral conventions under international law; they are the unilateral acts of an international organisation in the form of recommendations to its member countries. Even the OECD does not consider these recommendations to be binding; on the contrary, according to the OECD Rules of Procedure the member countries are to consider whether their implementation is opportune. ( 41 ) This applies a fortiori to the commentaries published by the OECD, which ultimately only contain legal opinions.

82.

However, in light of settled case-law, it is not inappropriate for the Member States to derive guidance for the balanced allocation of their fiscal competence from international practice as reflected in the Model Tax Convention. ( 42 ) The same applies to guidance from any prevailing international legal opinion that may be reflected in the commentaries on the OECD MTC.

83.

However, the commentaries on the OECD MTC cannot have a direct effect on the interpretation of an EU directive (and thus on the interpretation of national law in conformity with EU law). In that respect, those commentaries simply reflect the opinion of the persons who worked on the OECD Model Tax Convention, not the views of a parliamentary legislature or indeed of the Union legislature. At most, should it transpire from the wording and history of the directive that the Union legislature took guidance from the wording of an OECD Model Tax Convention and the commentaries (available at the time) on that OECD Model Tax Convention, a similar interpretation might be appropriate.

84.

Therefore, the Court has already found that a rule in a double taxation agreement, interpreted in the light of the OECD commentaries on its applicable Model Tax Convention, cannot restrict EU law. ( 43 ) This applies in particular to changes to the OECD Model Tax Convention and the commentaries published after the adoption of the directive. Otherwise, the OECD member countries would be able to rule on the interpretation of an EU directive.

85.

Therefore, Questions 3 and 4 can be answered to the effect that the Parent-Subsidiary Directive must be interpreted under EU law autonomously and independently of Article 10 of the 1977 OECD Model Tax Convention or subsequent versions.

86.

Furthermore, the referring court ultimately asks whether a ‘dividends recipient’ within the meaning of the Parent-Subsidiary Directive should be construed in the same way as ‘beneficial owner’ within the meaning of the Interest and Royalties Directive. The answer is that it should not, since, as stated previously (at point 35), the approach of the Parent-Subsidiary Directive differs from that of the Interest and Royalties Directive and therefore deliberately avoids using the term ‘beneficial owner’.

87.

By its eighth question, the referring court asks if the Member State that does not wish to recognise that the dividends recipient is also the beneficial owner within the meaning of the Parent-Subsidiary Directive, because it is simply an artificial conduit company, is bound to state whom it deems to be the beneficial owner. The nub of this question referred is who bears the burden of proof of abuse.

88.

In order for abuse of possible legal arrangement to exist, a legal arrangement must be chosen that differs from the arrangement normally chosen and gives a more favourable result than the ‘normal’ arrangement. In the present case, the ‘normal arrangement’ would have been a direct dividend disbursement between the capital investment companies and the claimant in the main proceedings. That ‘normal arrangement’ would also have to result in a higher tax burden.

89.

In principle, it is for the tax authorities to demonstrate that the approach chosen gives a more favourable tax result than the normal arrangement, although the taxable person may have a certain duty to assist. However, the taxable person may then ‘produce, if appropriate …, evidence as to the commercial justification for the transaction in question. ( 44 ) Should it transpire from that evidence that the essential aim ( 45 ) was not to avoid the tax that would normally be assessed, the approach chosen cannot be deemed abusive, especially as it is the State that provides taxable persons with such options.

90.

It further follows from the case-law of the Court ( 46 ) that, if conduct is deemed abusive, the situation must be determined that would have existed in the absence of the circumstances that constitute the abusive practice and that redefined situation must be assessed in the light of the relevant provisions of national law and EU law. However, for that, the identity of the dividends recipient must be clear.

91.

Thus, from Denmark’s perspective, abuse within the meaning of Article 1(2) of the Parent-Subsidiary Directive can arise only where dividends distributed directly would have been taxed accordingly in Denmark. However, this would be precluded under Danish law if, disregarding the conduit company, the actual dividends recipient were also an undertaking with its seat in a different Member State or the dividends recipient were resident in a State with which Denmark has concluded a DTC. If, for example, the group parent company in the USA, rather than Y Bermuda and Y Cyprus, were to be treated as the actual dividends recipient, then that arrangement would also be exempt from withholding tax under Danish law.

92.

Therefore, the eighth question can be answered to the effect that a Member State that does not wish to recognise a company resident in a different Member State, to which the dividends were paid, as the recipient of the dividend must in principle state whom it considers to be the recipient of the dividend in order to assume that abuse exists. This is necessary in order to determine if a more favourable tax result is achieved as a result of the arrangement qualified as abusive. In cross-border cases in particular, the taxable person may have an enhanced duty to assist.

93.

By its Questions 1, 1.1 and 2, the referring court asks (1) whether Denmark can rely directly on Article 1(2) of the Parent-Subsidiary Directive to refuse the taxable person the exemption from tax provided for in Article 5(1) of that directive and, if not, to clarify (2) whether, by its current national law, Denmark has in fact adequately transposed Article 1(2) of the Parent-Subsidiary Directive.

94.

If, based on the aforementioned criteria, abuse exists within the meaning of Article 1(2) of the Parent-Subsidiary Directive, the peculiarity of the present case is that Danish law contained no specific provision transposing that provision. Nor, according to the referring court, was there any general provision to prevent abuse. Some of the parties in the main proceedings are therefore of the opinion that they cannot be denied tax relief under national law even if abuse were assumed to exist.

95.

However, it is not always necessary formally to enact the requirements of a directive (in this case Article 1(2) of the Parent-Subsidiary Directive) in specific legal provisions. On the contrary, the transposition of a directive may, depending on its content, be achieved through a general legal context, including general principles of national constitutional or administrative law, if it ensures the full application of the directive in a sufficiently clear and precise manner. ( 47 )

96.

The referring court refers in the proceedings for a preliminary ruling to the existence of two principles (the ‘reality doctrine’ and the principle of the ‘rightful income recipient’). However, it is common ground that these are irrelevant here if, in fact, the dividends are formally paid first to the Cypriot company.

97.

Article 1(2) of the Parent-Subsidiary Directive allows the Member States to apply provisions to prevent abuse. That is in keeping with practice throughout the Union. For example, all Member States have developed, to the greatest possible extent, instruments to prevent abuse of the law for the purposes of tax avoidance. ( 48 ) Thus, there is a consensus, also under national tax laws, that the application of the law cannot in any case be extended to such an extent that abusive practices by economic operators must be tolerated. This principle, which is accepted throughout the Union, ( 49 ) is now also enshrined in Article 6 of Directive 2016/1164.

98.

To that extent, all national provisions, whether adopted in transposition of the Parent-Subsidiary Directive or not, must be interpreted and applied in accordance with this general principle of law and, in particular, with the wording and purposes of the Parent-Subsidiary Directive and Article 1(2) thereof. ( 50 ) The fact that, in interpreting national law in conformity with EU law, detriment to an individual may result is not an obstacle to such an interpretation. It is lawful, by way of national law provisions, that is to say indirectly, to apply EU law to the detriment of an individual. ( 51 )

99.

Only a direct application of Article 1(2) of the Parent-Subsidiary Directive to the claimant’s detriment would be denied the Danish authorities, for reasons of legal certainty. ( 52 ) For example, a Member State cannot hold an individual to the provision of a directive that it has not transposed. ( 53 ) It is settled case-law that a directive cannot of itself impose obligations upon an individual; the directive cannot be invoked as such against him. ( 54 ) A Member State that did so would itself be guilty of ‘abusive conduct’: first, it would not have transposed a directive addressed to it (even though it could) and, second, it would be relying on a possibility of preventing abuse that was contained in a directive which it had not transposed.

100.

Nor could the competent authorities in the main proceedings rely directly against the individual on the general principle of EU law that abuse of rights is prohibited. At least in cases falling within the scope of the Parent-Subsidiary Directive, such a principle has been given specific expression in Article 1(2) of the Directive and has been expressed in a concrete manner. ( 55 ) If it were to be permitted, in addition, to have direct recourse to a general principle of law which in terms of content is much less clear and precise, there would be a danger, thus, that the harmonisation objective of the Parent-Subsidiary Directive and of all other directives containing specific provisions to prevent abuse (such as Article 6 of Directive 2016/1164) would be undermined. Moreover, such an approach would undermine the prohibition, already mentioned, on directly applying non-transposed provisions of directives to the detriment of individuals. ( 56 )

101.

This does not conflict with judgments delivered by the Court ( 57 ) in Italmoda and Cussens, in which the Court ruled that the principle of the prohibition of abusive practices must be interpreted as being capable, regardless of a national measure giving effect to it in the domestic legal order, of being applied directly in order to refuse exemption from value added tax (VAT), without conflicting with the principles of legal certainty and legitimate expectation.

102.

However, those two judgments referred exclusively to VAT, which differs from the subject matter at issue here. First, VAT is much more harmonised under EU law and, as it is coupled to the funding of the Union, has far more of an impact on interests under EU law than national income tax.

103.

Second, EU law (Article 325(1) and (2) TFEU) requires the Member States to take (effective) measures to collect VAT, ( 58 ) whereas the same does not apply under income tax law. Moreover, VAT law is particularly susceptible to fraud; therefore particularly effective enforcement of tax claims is required. In that sense, the Court itself drew a distinction in a recent judgment between VAT law and secondary EU law, which contains an express authority to prevent abuse. ( 59 ) Therefore, direct application of Article 1(2) of the Parent-Subsidiary Directive to the detriment of the taxable person is out of the question. ( 60 )

104.

It will be the task of the referring court, however, to determine whether in the present case general provisions or principles of national law (including principles established in case-law) automatically apply, as a result of which, for example, sham transactions are disregarded under tax law or reliance on particular advantages for abusive ends is prohibited.

105.

For a restriction of freedom of establishment to be justified on grounds of the prevention of abusive practices, the specific objective of such a restriction must be to prevent conduct involving the creation of wholly artificial arrangements which do not reflect economic reality, with a view to escaping the tax normally due on the profits generated by activities carried out on national territory. ( 61 )

106.

For that reason, Questions 1.1 and 2 can be answered to the effect that neither Paragraph 2(2)(c) of the Danish Law on corporation tax nor a DTC rule predicated on the beneficial owner for the purpose of taxing distributed dividends suffice to be deemed a transposition of Article 1(2) of the Parent-Subsidiary Directive.

107.

However, that would not apply to the application in conformity with EU law of the ‘reality doctrine’ and the principle of the ‘rightful income recipient’ in Denmark, both of which have been developed precisely in order to resolve the problem that civil law allows numerous arrangements, whereas tax law is applied to economic facts. Those legal principles therefore specifically target artificial arrangements or abuse of the law by the individual and therefore constitute in principle a sufficiently specific legal basis on which to restrict freedom of establishment. Inasmuch as Denmark has not expressly transposed Article 1(2) of the Parent-Subsidiary Directive, that ultimately would not matter. It is for the national court to determine that case-by-case.

108.

The ‘reality doctrine’ developed in Denmark, interpreted in accordance with EU law, might therefore suffice as a legal basis on which to ignore wholly artificial or abusive arrangements, where they exist (see point 52 et seq.) for tax purposes. In my opinion, the ‘reality doctrine’ too seems to me to be nothing other than a variation on the economic viewpoint approach underlying most provisions adopted by the individual Member States for the prevention of abuse. ( 62 ) This is also made clear at the level of EU law, for example in Article 6(2) of Directive 2016/1164, which states that an arrangement is deemed non-genuine to the extent that it was not put into place for valid commercial reasons which reflect economic reality. It is for the national court to determine whether that is the case.

109.

If the aim of the arrangement is to prevent taxation of the actual investors, then, from an economic point of view, despite a formal distribution of the dividends to the Cypriot parent company, the payment is actually made to its shareholders in the form of Y Bermuda (or even, possibly, to the group parent company Y USA). The payment to the Cypriot parent company then reflects only the (formal) reality under civil law, not the economic reality.

110.

As there is no reason in the present case why the exemption from withholding tax under Article 5 of the Parent-Subsidiary Directive should not apply, no further consideration need be given to Questions 6, 7, 9 and 10.

111.

Inasmuch as, in application of the principles enshrined in national law, interpreted in accordance with EU law, the referring court finds that the arrangement in question is an abuse, withholding tax will indeed apply under certain circumstances. However, the questions then no longer arise in the present case, as that taxation is the result of abuse and abusive reliance on EU law is not permitted. ( 63 )

112.

That notwithstanding, however, the Court has also already ruled that different treatment of national and foreign interest recipients on the grounds of different taxation arrangements relates to situations which are not comparable. ( 64 ) The same applies to national and foreign dividend recipients. Even if they were deemed to be comparable situations, restriction on the freedom of establishment would be justified under the case-law of the Court as long as the liability for Danish tax at source of the dividends recipient resident abroad is no higher than the liability for Danish corporation tax of a national dividends recipient. ( 65 )

113.

The same holds for different interest or a different accrual of the Danish corporation tax debt for the dividends recipient and of a Danish obligation to withhold Danish tax for the dividends payer. These are not comparable situations as, on the one hand, a national tax (corporation tax) is owed and, on the other, for the dividends recipient, an actually foreign tax (its corporation tax) is withheld and paid on its behalf. Differentiated accrual and interest are the result of the different technique and function of a tax at source (see point 72).

114.

In view of the foregoing, I propose that the answers to the questions from the Østre Landsret (High Court of Eastern Denmark, Denmark) should be as follows: