Jonathan Schwarz has already briefly discussed the CJEU’s judgment in the “Danish Beneficial Ownership Cases” (C-115/16, C-118/16, C-119/16, C-299/16 and C-116/16 and C-117/16), noting that the cases “represent a landmark on beneficial ownership” and comparing them to recent international tax law jurisprudence, concluding that the “voyage of discovery is certainly not at and end”. In this note, the authors would like to continue this voyage and expand on three specific points raised by the decisions. These specifics concerns are:
- the Court’s expansion of the general anti-abuse principle enshrined in EU law to areas of tax law that are subject to minimum harmonisation;
- the Court’s use of the OECD materials to define the beneficial ownership concept and its conflation with the general anti-abuse principle; and
- the Court’s reading of an effective subject-to-tax clause into the definition of a “company” laid down in the Interest-Royalties-Directive (IRD).
I. No need for a specific domestic or agreement-based provision implementing (old) Article 1(2) Parent-Subsidiary-Directive (PSD) or Article 5 (IRD)
The Court found that a specific domestic or agreement-based implementation of anti-abuse provisions is not necessary because the tax authorities may rely on the general principle that EU law cannot be relied on for abuse of fraudulent ends do deny benefits (paras 95-120 C-115/16 and paras 68-92 C-116/16). This finding is not only an unwelcome surprise, it also rests on a weak doctrinal foundation and may only be explained on account of the specificities of Danish legislation.
Let us revisit some basics first: Directives are addressed to the Member States (Article 288(3) TFEU) and require implementation into domestic laws. More specifically, a Member State may not invoke against an individual or a company a provision of a directive which has not (yet) been implemented (e.g. Case 152/84 Marshall). With respect to direct taxation, the Court held in Kofoed (C-321/05) that
“a Member State which has failed to transpose the provisions of a directive into national law cannot rely, as against Community citizens, upon limitations that might have been laid down on the basis of those provisions”.
So, if the legislator of Member State decides not to implement rules permitted by a directive’s anti-abuse reservation such as Art 1(2) in the pre-2015 PSD or Art 5 in the IRD, can the tax administration and courts nevertheless rely on an unwritten, general EU principle to counter perceived abuse? One may be inclined to answer that questions resoundingly to the negative: The Court’s precedence in Kofoed has made it (seemingly) clear that national tax authorities are precluded from relying directly, against a taxpayer, on the anti-abuse reservation of Art 15 of the Merger Directive (unless there is some way to interpret Danish law to that effect).Indeed, the Court in Kofoed noted that through the mechanism of “consistent interpretation” EU law could indirectly apply to the detriment of the taxpayer if “there is, in Danish law, a … Continue reading AG Kokott also added that recourse to “any existing general principle of [EU] law prohibiting the misuse of law” would be barred, as Art 15 is a concrete expression of such principle.See Opinion AG Kokott, 8 February 2007, Case C-321/05, Kofoed, EU:C:2007:86, para. 67, and Opinion AG Kokott, 16 July 2009, Case C-352/08, Zwijnenburg, EU:C:2009:483, para. 62. Even after Cussens (C-251/16), this was also the prevailing position in literature,See, e.g., A. Garcia Prats et al, “EU Report”, in: IFA (Hrsg), Seeking anti-avoidance measures of general nature and scope – GAAR and other rules, CDFI Vol. 103a (2018), Chapter 3.2.1; see … Continue reading and of AG Kokott in the cases at hand.
The Court took a different approach, emphasizing that the (unwritten) general principle of EU law that EU law cannot be relied on for abuse of fraudulent ends. This implies that any right or advantage can be denied based on the EU general principle of prohibition of abusive practices, regardless of any specific EU or domestic law provision, as recently confirmed for the VAT area in Italmoda (C-131/13 and others) and Cussens. Unlike AG Kokott, the Court transferred that notion also to the PSD and the IRD so that,
“in the light of the general principle of EU law that abusive practices are prohibited and of the need to ensure observance of that principle when EU law is implemented, the absence of domestic or agreement-based anti-abuse provisions does not affect the national authorities’ obligation to refuse to grant entitlement to rights provided for by [Directives 90/435 and 2003/49] where they are invoked for fraudulent or abusive ends”. (para. 111 of C-115/16 and para. 83 of C-116/16)
This obligation does not require domestic legislative implementation because, in the Court’s eyes, this is not an obligation imposed on taxpayers but rather merely part of the objective conditions required for obtaining the advantage sought. What then, one might ask, about Kofoed? The Court distinguishes: What it said in Kofoed with regard to the need for domestic anti-abuse rules or general principles and the possibility of “directive-compliant” interpretation of domestic law was just a first step and was not meant to exclude reliance on the general EU principle:
“Nevertheless, even if it were to transpire, in the main proceedings, that national law does not contain rules which may be interpreted in compliance with [Art 1(2) of Directive 90/435 or Art 5 of Directive 2003/49], this — notwithstanding what the Court held in the judgment of 5 July 2007, Kofoed (C 321/05, EU:C:2007:408) — could not be taken to mean that the national authorities and courts would be prevented from refusing to grant the advantage derived from the right of exemption provided for in [Art 4 of Directive 90/435 or Art 1(1) of Directive 2003/49] in the event of fraud or abuse of rights” (para. 117 of C-115/16 and para. 89 of C-116/16).
So far so good. It certainly means that taxpayers cannot rely on the direct effect of tax directives abusively even in the absence of a domestic anti-abuse provision or principle.
In light of the GAAR in Article 6 ATAD this issue might have little practical relevance in the future. Nevertheless, it deserves some fundamental, high-level analysis with regard to national tax sovereignty and separation of powers: Let us depart from the rather solid foundation that, e.g., the PSD only provides for minimum harmonization (and not for full harmonization as in the area of value-added taxation at issue in Cussens and Italmoda). This means that Member States may also enact more liberal rules and grant benefits that go beyond the directive, e.g., for situations where the directive’s capital ownership requirement is not fulfilled.This is not only clear from the wording of the PSD (arg “at least” in Article 3) but also the legislative history (see, e.g., Doc. 6446/84 FISC 42 of 18 April 1984, p. 2, and the Rossi-Report of … Continue reading If that assumption holds true, one might further argue that a Member State that provides for such more beneficial treatment is insofar effectively not implementing that directive but rather goes beyond it by means of plain non-harmonized domestic law (and may do so based on its sovereignty if it does not infringe on the fundamental freedoms or violate state aid rules). You see where this is going: If a Member State decides not to issue legislation to implement a directive’s anti-abuse reservation, it is effectively making a sovereign domestic tax policy decision to grant these benefits under domestic law, and that decision is not only unrelated to EU law, it also cannot logically be subject to an unwritten EU general principle that prohibits abuse of EU (and not also domestic) law.Similiarly S. Bergmann, “Missbrauch im Anwendungsbereich der Mutter-Tochter-Richtlinie”, 77 StuW 2010, 246 (254).
A different understanding would be quite a blow against the domestic separation of powers in that it undermines the decision of a national legislator not to implement an anti-abuse reservation by granting unelected tax officials and judges the power to override that decision based on an unwritten EU principle. Does this mean that the outcome in the Beneficial Ownership cases was “wrong”? Not necessarily. The Danish rules did not phrase the withholding tax exemptions in its own words, but rather explicitly referred to the PSD and the IRD by stating, e.g., that the withholding tax liability “does not apply to interest which is not taxed or is subject to reduced taxation under Directive [2003/49]” (para. 19 C-115/16 and C-116/16). This might arguably be the opening door for the Court’s analysis as the Danish rules might be read as “importing” all criteria of the directives, including – from the Court’s perspective – the general EU principle that EU law cannot be relied on for abuse of fraudulent ends and not create an independent domestic framework that goes beyond the directives and establishes domestic rights for taxpayers (even though there is indeed evidence that this was a very deliberate decision by the Danish legislator not to implement anti-abuse provisions).See for a detailed account of the development of Danish withholding tax law and the efforts in the 1990s and early 2000s to offer itself as a conduit jurisdiction H. S. Hansen, “Det store hykleri … Continue reading
II. The Court’s approach to beneficial ownership and abuse
On the condition of beneficial ownership, the Court deviated from AG Kokott’s analysis and concluded that the OECD materials are “relevant when interpreting the [IRD]” (para. 90 of C-115/16). While this may not be entirely surprising given the context of the IRD’s adoption and the use of the OECD model’s terminology, the Court did not explain how their “being relevant” is going to influence the outcome when applying the IRD to a concrete case and why the most current version of the OECD guidance should be used to interpret a directive that was proposed in 1998 and adopted in 2003. Given the more specific explanations and conditions found in the IRD that differ from the wording of the OECD model tax convention, this raises both methodological and substantive questions.
For instance, the Court’s starting point that the term ‘beneficial owner’ “cannot refer to concepts of national law which vary in scope” (para. 84) appears to be undermined by the condition in Article 1(5)(b) IRD according to which a PE is treated as the beneficial owner only if it is subject to income tax on relevant payment. In the case of a PE the concept would thus seem to vary explicitly with national tax rules. One may counter this by arguing that the situation of a PE is special: it can never actually be the beneficial owner, but is, as Article 1(5) IRD makes plain, only treated as such. However, as the Court invoked that same provision in order to explain the meaning of “company of a Member State” (para. 152 of C-115/16; see also section 3) below), it does not appear to see it as a particularity for PEs. Does this mean that taxation in the residence State of the recipient is to be considered a requirement for beneficial ownership? That would certainly appear to be the result of the Court’s judgment, but is clearly not derived from OECD guidance. While the latter makes it clear that a dividend (or interest) recipient needs to be considered the owner for tax purposes of that payment by its State of residence in order to qualify as beneficial owner (e.g. OECD Commentary para 12.2 on Article 10), actual taxation there is clearly not a condition.
As Jonathan Schwarz noted, the concepts of beneficial ownership and abuse of law are intertwined in the Court’s analysis. This may not seem surprising at first, considering the indubitable purpose of the beneficial ownership concept to avoid and abusive reliance on a tax treaty in specific circumstances. Yet the key to this is that the concept merely aims at avoiding specific types of abuses. As AG Kokott pointed out in her Opinion (para. 60 of C-115/16), the concerns addressed by that concept and the beneficial ownership concept are fundamentally different. The Court also appears to recognize the difference between both concepts at certain stages of its analysis, making it clear that denial of a benefit based on a lack of beneficial ownership does not require tax authorities to prove abuse of law (para. 138 of C-115/16; para. 111 of C-116/16).
The situation differs between the PSD and the IRD, however. Since the former does not contain an explicit beneficial ownership requirement, the CJEU appears to mangle a purposive interpretation of the PSD’s scope with its general anti-abuse provision in order to deny the directive’s benefits to companies that are not beneficial owners in cases C-116/16 and others. At the same time, it explicitly refuses to respond to the national court’s question regarding the interpretation of the beneficial ownership concept (para. 94 of C-116/16).
By contrast, in applying the IRD, the Court appears to keep the two concepts more clearly separate in cases C-115/16 and others. There, the CJEU did attempt to answer the question what precisely beneficial ownership entails, but confined itself to the statement that it is an economic concept denoting the “entity which benefits economically from the interest received and accordingly has the power freely to determine the use to which it is put” (para. 89 and 122 of C-115/16). In its subsequent analysis regarding the constituent elements of abuse of rights, the CJEU refers to the situation of a recipient company that does not “in substance” have the right to use and enjoy sum it received even “without being bound by … a contractual or legal obligation [to pass it on to a third party]” (para. 132 of C-115/16). This ostensibly goes beyond the OECD Commentary’s guidance on beneficial ownership, which confines the denial of treaty benefits to situations where such contractual or legal obligation exists (OECD Commentary para. 12.4 on Article 10). However, the context of the Court’s inquiry suggests that it did not interpret the concept of beneficial ownership in this context; it explored the concept of artificial arrangements. As a result, this may be best understood as clarifying the relationship between beneficial ownership and the abuse of law: an entity may be the beneficial owner (as interpreted in conformity, most likely, with the OECD material), yet still be denied the directive’s benefits due to the artificiality of the legal structure.
III. Does being a “company of a Member State” require its income being subject-to-tax?
Both the PSD and the IRD only apply to a “company of a Member State”. To be such qualified “company of a Member State”, a three prong-test has to be met, the third prong of which requires that the company is “subject to” on the Member States’ corporate taxes “without […] being exempt” (Art 2(a)(iii) and Art 3(a)(iii), respectively). This criterion is intensely discussed in literature,For a recent overview see, e.g., P. Arginelli, “The Subject-to-Tax Requirement in the EU Parent-Subsidiary Directive (2011/96)”, 57 ET 2017, 334 (340). and case law also provides some guidance: While the directives’ wording might suggest that we have to focus on whether the company as a taxable person is, in principle, “subject to” a domestic corporate tax (and not, e.g., a personally exempt charity or foundation), the Court seems to understand the second prong of the test (“without […] being exempt”) as referring to the treatment of the company’s income.
The Court has, for example, held a company to be “exempt” within the meaning of Art 2(a)(iii) of the PSD where (1) its income was fully exempt from corporate taxation (and only subject to a subscription tax under the local tax regime for investment funds),ECJ, 18 June 2009, Case C-303/07, Aberdeen Property Fininvest Alpha Oy, EU:C:2009:377, para. 27. or (2) where it “is entitled […] to a zero rate of taxation for all its profits, provided that all those profits are distributed to its shareholders”.ECJ, 8 March 2017, Case C-448/15, Wereldhave, EU:C:2017:180, para. 40. So while a “zero rate” seems to disqualify a company from the benefits of the directive, a reduced rate would not.See, e.g., Opinion AG Kokott, 1 March 2018, C-118/16, X Denmark, EU:C:2018:146, para. 96. The outcome is less clear in situations where a company enjoys exemption for certain items of income but not for others. Assume, for example, that a company’s dividend income and capital gains are exempt, but its interest and royalty income is taxed at normal rates.
Could, for example, the source State levy a dividend withholding tax on a distribution to such parent company based on the argument that the exemption for dividend income removes it from being a “company of a Member State”? The Italian Supreme Court recently came to that surprising result.See Italian Supreme Court, 13 December 2018, no. 32255. But that clearly goes too far, as it (1) disregards the economic double taxation in the source State that the PSD (also) aims to avoid and (2) is not in line with the Court’s “but for”-test developed in Wereldhave: The mechanisms of the PSD are “intended for situations in which, if they were not applied, the exercise by the Member States of their powers of taxation might lead to the profits distributed by the subsidiary company to the parent company being subject to double taxation” (para. 39 of C-448/15). This test seems to disregard the taxation of income not covered by the respective directive and effectively leads us to an effective “subject-to-tax”-criterion for interest and royalties under the IRD.See already the reading by D. Weber, “The proposed EC Interest and Royalty Directive”, 9 EC Tax Rev. 2000, 15 (21). And that was indeed what the Court found in the Danish Beneficial Ownership cases:
Should it turn out that “the interest received by [the Luxembourgian SICAR] is in fact exempt in that respect from corporate income tax in Luxembourg, it would then have to be stated that that company does not satisfy the third condition […] and that it cannot therefore be regarded as being a ‘company of a Member State’ within the meaning of Directive 2003/49. It is, however, for the referring court alone to make, if appropriate, the necessary checks in that regard.” (para. 151 of C-115/16)
This finding creates some tension with a broader reading of the directive in the past, which understood that “none of the provisions in Directive 2003/49 stipulates that an actual taxation of the beneficial owner (here the Luxembourg companies) in a certain amount is a requirement for the exemption”. (AG Kokott in C-118/16 para. 93) This interpretation, in turn, was supported by two – to date: not adopted – proposals be the EU Commission: It proposed both in 2003Proposal for a Council Directive on a common system of taxation applicable to interest and royalty payments made between associated companies of different Member States, COM(2003)841 final. and again in 2011 to include a more stringent “subject-to-tax” clause, indicating that the current wording might indeed only refer to subjective exemptions of the recipient company, but not to objective exemptions of its interest or royalties income.Indeed, the European Economic and Social Committee has noted with respect to the 2003 proposal that such a criterion of effective taxation would be “introducing a proviso which was not there … Continue reading
Such amendment would align the “subject-to-tax” requirement for companies in Art 3 of the IRD with the one that is already enshrined in Art 1(5) in the “beneficial ownership” test for permanent establishments: The latter has always required that the “interest or royalty payments represent income in respect of which that permanent establishment is subject in the Member State in which it is situated to one of the taxes” specifically listed in the directive. This means that a permanent establishment would not qualify as the beneficial owner if the income is either not attributable to it for tax purposes or if interest or royalties would be objectively exempt from taxation. It does, however, not require a minimum rate or effective taxation in a narrow sense; hence, beneficial ownership is not put into question just because no tax liability arises, e.g., because of loss carry-forwards, credits, or deductions. That proposed amendment now seems moot, as the Court closes the circle with a surprising systematic and teleological reasoning:
“That interpretation of the scope of the third condition […] is supported, first, by Article 1(5)(b) of Directive 2003/49, from which it is apparent that a permanent establishment can be regarded as being the beneficial owner of interest, within the meaning of the directive, only ‘if the interest … payments [which it receives] represent income in respect of which that permanent establishment is subject in the Member State in which it is situated to one of the taxes mentioned in Article 3(a)(iii) …’, and second, by the objective of Directive 2003/49, which […] is to ensure that such interest payments are subject to tax once in a single Member State.” (para. 152 of C-115/16)
In conclusion, the Court’s judgments in the Danish Beneficial Ownership cases bring a lot of new and unexpected developments in three separate areas: (1) The Court appears to extend the direct application of directives to the taxpayer’s detriment in cases of abuse to the field of direct taxation; (2) the Court wades into the interpretation of the term beneficial owner in international tax law, seemingly importing OECD guidance into EU law, while keeping the precise relationship to artificial arrangements somewhat unclear; (3) the Court adopts a surprisingly literal interpretation of the ‘subject to tax’ requirement in the PSD and IRD, thus breaking a long-standing impasse at the level of the EU legislature. All three developments are important and, considering the fact that the judgments were rendered by the CJEU’s Grand Chamber, are likely to be here to stay. Nevertheless, all three will require further elaboration. The next stages of this “voyage of discovery” are probably not far away.