Groundbreaking decision, landmark case, a milestone in EU case law, disappointing surprise: there are different ways to refer to the European Court of Justice (ECJ) judgments issued on February 26, 2019, on the withholding tax exemption provided by the Interest and Royalties Directive (joined cases C-115/16, C-118/16, C-119/16 and C-299/16, available here – the ‘IRD judgment’) and the Parent-Subsidiary Directive (joined cases C-116/16 and C-117/16 available here – ‘the PSD judgment’). However, one thing is certain: these decisions have caused upheaval in the international tax arena.
Although these cases affect EU law, it would be a mistake to confine their implications to the EU: in the majority of the cases, mainly non-EU investors will suffer indirectly the effects of these judgments, as their business structure incorporated in the EU could be denied application of the withholding exemptions.
We do not aim to analyze these judgments comprehensively. We only aim to contribute to the early stages of the debate by analyzing the ECJ’s reasoning in the cases concerning the withholding exemption for dividends in the PSD judgment, referring also to example K of the Commentaries on article 29 of the OECD Model Convention. The purpose is to review the interplay between beneficial ownership and abuse, the main concepts the ECJ addressed, as applied to dividends, and to share some disappointing aspects of the judgment.
Beneficial ownership and abuse in the Danish dividends
Unlike the IRD judgment, in the PSD judgment the ECJ does not answer the questions referred by the Danish courts on the concept of beneficial ownership of dividends. There is no guidance on whether, to be considered the beneficial owner, the recipient of the dividends should receive the payments for itself and have the power to decide how the dividends are used (this could be a reasonable interpretation that is not found in the judgment). Neither solves the question of whether the Commentaries on article 10 of the OECD Model Convention are relevant for these purposes (a difficult outcome, considering the ECJ’s reasoning on this in the IRD judgment).
This situation concerns how the Danish courts raised their questions in the preliminary rulings: deliberately or not, the ECJ found itself not obliged to answer. However, this could be consistent with the lack of any reference to a beneficial ownership requirement in the Parent-Subsidiary Directive (PSD), and with the fact that the Member States cannot unilaterally introduce restrictive measures and subject the right to exemption from withholding tax to several conditions (para. 24 in Eqiom, case C-6/16). In other words, the ECJ would not be addressing this concept to the extent that it is not, and it cannot be, a requirement for applying the withholding exemption on dividends because, once the formal requirements provided by the PSD are met, fraud or abuse are the only circumstances under which the application of a withholding exemption can be denied. Therefore, in principle, the PSD judgment only deals with the concept of fraud and abuse.
On this, it could be argued that the concept of artificial agreement described in para. 100 of the PSD judgment is closer to the concept of abuse provided by the GAAR in article 6 of the Anti-Avoidance Directive than the concept of abuse developed by the court’s case law around the ‘wholly artificial arrangement’ test (the Danish cases do not even mention it). Accordingly, it could be argued that the court has broadened the EU definition of tax avoidance with these judgments. Further, one may suggest that these judgments are endorsing the compatibility of the GAAR (secondary law) with the general principle of EU law that prohibits abusive practices (primary law) by leveling the concept of abuse for EU purposes.
Along these lines, we focus on how the concept of abuse interacts with the concept of beneficial ownership. This relationship seems to be intertwined in the judgments, as noted by Jonathan Schwarz in his post (available here), but it deserves several comments.
The indicators of abuse in the judgment may be seen as indications that overlap the criteria for analyzing beneficial ownership. Passing on all or almost all of the dividends shortly after receiving them, or having the legal or contractual obligation to pass the dividends on to a third party, are circumstances that traditionally were considered relevant in a beneficial ownership analysis. Even the fact that the recipient ‘in substance’ did not have the right to use and enjoy the dividends, referred by the Danish courts and endorsed by the ECJ as a relevant indicator, is mentioned in the Commentaries on article 10 of the OECD Model Tax Convention (para. 12.4) when discussing the concept of beneficial ownership. However, the Danish cases highlight these as evidence of an abuse of rights.
One may be deeply disappointed to learn that the same indications of abuse are used for cases of interest and dividends (just adapting the language in the IRD judgment and the PSD judgment, a mere “copy&paste”), given their different economic and legal nature. The ECJ introduces those indicators as tools to find abuse, not to argue to the contrary, by the way. However, focusing on the criteria overlap, we wonder whether it is a strange outcome. Maybe not, if we assume that a beneficial ownership test can identify certain cases of abuse. Commentaries on article 10 of the OECD Model (para. 12.5) seem to support this conclusion, acknowledging “the concept of ‘beneficial owner’ deals with some forms of tax avoidance.”
On this basis, there are two main aspects to discuss. As recent contributions to this blog have emphasized (available here), the beneficial ownership test cannot capture every case of abuse: ‘an entity may be the beneficial owner (most likely interpreted in conformity with the OECD material), yet still be denied the directive’s benefits due to the artificiality of the legal structure’.
What about the other way around? What happens when the entity is not the beneficial owner of the dividends? Theoretically, there are at least two approaches.
The first assumes that the withholding exemption can be denied if the recipient is not the beneficial owner of the dividends, regardless of the existence of fraud or abuse (para. 111 of the PSD judgment). The ECJ justifies this approach by arguing that the mechanisms of the PSD were not intended to apply when the beneficial owner of the dividends is outside the EU, because ‘exempting those dividends from withholding tax in the Member State from which they are paid could result in them not being taxed in the European Union’ (para. 113).
This approach is odd, to the extent that the beneficial ownership test seems to be construed as an autonomous requirement for applying the withholding exemption provided by the PSD. This is inconsistent with previous ECJ decisions, as noted above; apart from avoidance or abuse, there should be no other conditions for applying the exemption if the formal requirements are met. Additionally, the justification hinges on the lack of taxation of those dividends in the EU, overlooking the PSD’s aim, which is to provide relief to the double taxation inherent to dividends, from juridical (eliminating withholding taxes) and economic (forcing the parent company’s Member State to grant a credit or an exemption) perspectives.
By disregarding the PSD’s original aim of eliminating double taxation, the EJC provides incoherent solutions, particularly in case C-117/16, where the court obliges Denmark to impose a withholding tax on dividends paid off by a Danish intermediate holding without taking into account that these dividends ultimately came from business profits obtained in the Netherlands. Moreover, the ECJ overlooks that the recipient of the dividends may have to withhold taxes when those dividends are paid off to the non-EU investor, and whether this happens is within the sovereignty of the Member States. It is questionable that a Member State’s decision not to charge withholding taxes on payments outside the EU can explain the existence of an abuse that obliges another Member State to apply such withholding taxation.
The alternative interpretation is also found in the PSD judgment, where the ECJ affirms in para. 117 that the tax authorities are given the ‘the task of establishing the existence of elements constituting such an abusive practice while taking account of all the relevant factors, in particular the fact that the company to which the dividends have been paid is not their beneficial owner’. Beneficial ownership is introduced not as an autonomous requirement, but as a relevant factor to identify abuse. Then, if the absence of beneficial ownership does not imply abuse under EU law, it should not be the only relevant factor. Some authors in the past have supported this conclusion, and recently Jonathan Schwarz supported it in his post. This interpretation would be in line with the PSD’s wording, which does not referring to the beneficial ownership test.
The difference between both approaches is obvious. In the first approach, if the recipient of the dividends is not the beneficial owner, the withholding exemption can be denied based on no other reasoning. In the second approach, this condition would a relevant element for assessing the existence of fraud and abuse, which should be done based on the remaining facts and circumstances. This is important to the extent a recipient of dividends can be part of a non-artificial arrangement (e.g., for having substance and relevant non-tax reasons supporting its existence), despite not being the beneficial owner of the dividends, as example K shows.
OECD’s example K
Because of the difficulties in interpreting GAAR in relation to specific cases, tax practitioners appreciated the examples given in BEPS Action 6 on applying the PPT clause. Even assuming it is considered soft law, and the specific background used in certain examples, at least those working at the OECD (i.e., the tax administrations) shed some light. These examples, along with three others, were included in the 2017 version of the Commentaries on article 29 of the OECD Model Tax Convention.
One of these other examples is example K, in which a fund (resident in State T) incorporates a holding company in State R that operates exclusively to generate an investment return as the fund’s regional investment platform. The holding company has substance to manage its diversified portfolio, including an envisaged investment in a subsidiary resident in State S. The tax advantage is defined as a reduction from 10% (withholding tax on dividends under the double tax treaty signed between T and S) to 5% (withholding tax under the double tax treaty signed between R and S), when the general withholding rate at source is 30%.
The example confirms that, although the holding company considered the existence of a tax advantage, ‘it is necessary to consider the context in which the investment was made, including the reasons for establishing RCO in State R and the investment functions and other activities carried out in State R’. It concludes that the benefits of the double tax treaty signed between R and S should not be denied; that is, the PPT clause does not preclude its application.
We could argue that a reduction in 5 percentage points is a minor tax advantage that is not capable of sustaining the subjective element of abuse (we have pointed out the specific background of certain examples). However, one aspect should not be disregarded: the holding company operates exclusively to generate an investment return for the fund, thus, raising the question of whether it would be beneficial owner of the dividends. This circumstance seems to be irrelevant for the abuse analysis under the PPT clause, which was discarded in a case where there was no major tax advantage and the holding company had substance and different business reasons to exist.
This would likely be even more striking in a context where the fund is a regulated entity under EU law; several of the structures analyzed in these cases belong to private equity investments, which, today, would fall under the regulation established in the AIFMD.
The Danish cases are a milestone in the tax treatment of holding companies, but they are not the first milestone. The ECJ’s case law includes Eqiom and Deister-Jühler (joined cases C-504/16 and C-613/16), precedents in which the court seems to protect the use of holding companies. For example, by preventing Member States from incorporating anti-abuse rules that create a general presumption of fraud (even when non-EU investors control EU intermediate holding companies), and by admitting that the management of assets is not an abusive activity, regardless of its treatment for VAT purposes.
Are the Danish cases consistent with this case law? This is a difficult question, bearing in mind that these Danish judgments do not even mention Eqiom or Deister-Jühler. One could respond affirmatively: in the Danish cases, the tax administration still holds the burden of proof of the existence of abuse, so these intermediate holding companies are not seen as artificial arrangements per se, even if non-EU investors control them. However, the answer to the question raised here (whether beneficial ownership is an autonomous test under the PSD) could be relevant for these purposes.
If the beneficial ownership test is an autonomous test that can deny the withholding exemption, and is in any way subject to fraud or an abuse of rights, it could be an irrefutable presumption of fraud or abuse, which is precluded in Deister-Jühler. Moreover, this approach would not consider whether the holding company is carrying out a management activity, which in itself is not abusive (again, Deister-Jühler, para. 73). This is not so consistent with case law.
We cannot disregard the purpose of the PSD, which aims to provide tax relief on EU dividends to avoid double taxation, the decisive aspect being whether the dividend payer and the recipient were taxed. For these purposes, whether the recipient is the beneficial owner of the dividends is not relevant, as AG Kokott highlighted correctly (para. 36 to 47 in her Conclusions). This is particularly relevant in sectors such as private equity, in which the prompt payment of dividends to the investors as an investment return is a duty that is an essential feature of the activity.
Even if this is not the time to discuss the effects of EU law in the treaties signed by the Member States relating to avoidance and abuse, or the influence of BEPS in the Anti-Avoidance Directive, looking at example K serves as a cornerstone, as it shows that a holding company incorporated to generate an investment return can be seen as non-abusive due to other circumstances and is entitled to the benefits of a double tax treaty.
In our opinion, the beneficial ownership test, which is not mentioned in the PSD, should not be considered an autonomous test capable of denying the withholding exemption. Certain criteria relevant for assessing avoidance or abuse could overlap with the relevant criteria for the purposes of beneficial ownership, but the former analysis is more complex, has a different nature and should include reviewing the substance of the holding company and its economic rationale.
 P. Baker, Chapter 6: The Meaning of “Beneficial Ownership” as Applied to Dividends under the OECD Model Tax Convention in Taxation of Intercompany Dividends under Tax Treaties and EU Law (G. Maisto ed., IBFD 2012), Online Books IBFD (accessed August 2, 2016)
 B. Kuźniacki, “The ECJ as a Protector of Tax Optimization via Holding Companies”, Intertax, Volume 47, Issue 3.