On September 4, 2018 the European Commission published the non-confidential version of its decision in the Engie case (SA.44888), where it concluded that Luxembourg had granted illegal State aid through two tax rulings. The decision has been appealed by Luxembourg. The decision brings interesting additions and precisions to the arguments developed in the earlier cases concerning Starbucks, Fiat, Apple, Amazon, and the Belgian excess profit exemption regime. While the previously published decisions concerned primarily transfer pricing issues, this decision relates to the qualification of financing structures and the use of the general anti-avoidance rule, both under Luxembourg domestic law. The decision concerns the treatment of the profits of several companies of the Engie group which were resident of Luxembourg. Two similar structures were accepted by the Luxembourg tax authorities in two tax rulings. The structures relied on a hybrid convertible loan, whereby the instrument was qualified as debt for the borrower; accordingly, interest expenses were deducted, and reduced the profits of the borrowing company. The loan would then be converted to equity, and payments would be made in the form of shares. The shares would then be cancelled and received in cash, with no taxation under the participation exemption regime. The effect from a corporate income tax perspective is the deduction of interest expenses, on the one hand, with no corresponding taxation of income, on the other hand.
I will not deal here with a deep analysis of the facts, or with the question of whether or not the Commission is right in its assessment of the compatibility with the State aid rules of the two rulings. Instead, I will focus on a few open questions raised by this case with respect to the relation between the State aid rules and the prevention of tax avoidance. A first question concerns the determination of the reference system and of its objective. Unsurprisingly, the Commission considers that the reference system is made of the Luxembourg corporate income tax system. The Commission then argues that the objective of the system would be ‘the taxation of the profit of all companies subject to tax in Luxembourg, as determined in their accounts’ (point 179). While this objective may be correct, the justification of this assumption is mainly a reading of the basic principle according to which companies are taxed on all their profits (points 176 and 185). Most tax systems indeed include such a rule as a starting point, but then complete this rule by additions or exceptions: certain items of income may be exempt from tax (e.g. foreign business profits under the exemption method in a tax treaty, dividends under a participation exemption regime), certain categories of income normally exempt might actually be taxed (e.g. as a result of CFC rules), or the deduction of certain expenses may be denied (e.g. interest expenses) while the corresponding income is subject to tax at the level of the recipient. It is difficult to accurately determine the objective of a reference system when this system is defined in a broad manner, although there are good reasons for defining the reference system broadly. The different components of a tax system may simply pursue different objectives or have different functions, such as the fight against tax avoidance, the prevention of double taxation, or the strengthening of the fiscal competitiveness of a country, while at the same time the more general objective of the system might still be to levy tax on corporate profits. The Commission is aware of these arguments, but does not address them in details in the decision (point 191).
There is an important follow-up issue to that of the determination of the objective of the reference system. As mentioned above, the Commission argues in the Engie decision that the objective of the system would be ‘the taxation of the profit of all companies subject to tax in Luxembourg, as determined in their accounts’. Does that mean that the avoidance of tax contradicts the objective of the system, even though the lack of tax in a particular situation is in line with the letter of the law? The question is important, as a positive answer could imply an obligation under the State aid rules to actually pursue the general objective of the system, even at the cost of a breach of certain provisions of that system. Luxembourg and Engie opposed the argument of the principle of legality, according to which tax cannot be levied without explicit support in the law (see e.g. point 183). However, fully accepting the principle of legality and applying a strict reading of the law would give latitude to the Member States to grant selective advantages through the design of the tax system (what the CJEU often describes as “regulatory techniques”), eventually threatening the very purpose of the State aid rules. The effects doctrine developed by the Union courts seriously weakens the relevance of the principle of legality in this context.
A second interesting question concerns the potential existence of a selective advantage resulting from the non-application of the Luxembourg rules on the prevention of abuse of law. This argument is described as an alternative line of reasoning by the Commission (point 292). Many Member States have a general anti-avoidance rule in their domestic laws, which is now required under article 6 of the anti-tax avoidance directive (directive 2016/1164). The issue here is not to determine whether State aid law, as such, implies an obligation to include a general anti-avoidance rule, although the Commission seems to indicate that this might be the case so as to achieve the objective of the reference system (point 294). The question is rather to determine whether the non-application or the misapplication of a rule that exists in a tax system might imply a selective advantage. At a theoretical level this argument seems perfectly relevant: if a taxpayer has not been subject to the correct application of an anti-avoidance rule, there is a deviation from the reference system. This question raises more a practical issue: how to determine the correct application of the reference system? Certain types of tax rules are designed in an objective manner, and rely on a relatively straightforward application. That could be the case of interest limitation deductions based on ratios of deductible expenses, or limitation-on-benefits provisions in tax treaties, both of which are relatively objective in their application although fully objective rules hardly exist, at least because their various parameters need to be defined and applied on a case-by-case basis. But other rules, including general anti-avoidance rules, are much more subjective in their design and in their application. These rules need to be subjective, so as to give tax administrations and courts flexible tools to assess the compliance with the purpose of the law of various arrangements. For these subjective rules, it is difficult to predict in advance how they will be applied in a particular case. The mere reading of the provision, and even of the case law, might only provide an indication of how the rule may be applied in a given case. In other words, how to apply the selectivity test to subjective rules, including the general anti-avoidance rule of the Engie case, is a question that cannot receive a precise answer, which makes it virtually impossible to assess accurately whether there has been a deviation from the reference system. A similar issue is raised by the selectivity test of the application of the arm’s length principle in the transfer pricing cases, especially with respect to the possible need to choose a point in the arm’s length range as the correct application of the principle.
It may also be wondered, similarly to the decisions on transfer pricing, whether the State aid test of the application of anti-avoidance rules should only be performed based on the domestic law of the Member State, or if there might be an EU anti-avoidance rule, i.e. an interpretation of a general anti-avoidance rule that would be commanded by article 107(1) of the TFEU. That would contradict the principle stated at article 3 of the anti-tax avoidance directive, according to which the directive only sets a minimum standard, the Member States being free to adopt stricter rules. However, the directive is only a secondary source of EU law, subject to primary law. If the State aid rules indeed imply a given interpretation of the arm’s length principle that deviates from a Member State’s own interpretation (e.g. by preferring certain transfer pricing methods or a given point in the arm’s length range), then it could be wondered whether these rules should also influence the interpretation of a general anti-avoidance rule such as the one discussed in the Engie case, in particular given the already existing and still evolving case law of the CJEU on the notion of abuse. The same reasoning could also be extended to the interpretation of other tax rules, such as the notion of permanent establishment (is a Member State granting illegal State aid by not finding a place of business to be fixed, by qualifying an activity as preparatory or auxiliary, or by literaly interpreting the conclusion of contracts “in the name of” a principal?), the notion of beneficial ownership (is a Member State granting illegal State aid by finding an intermediary to be the beneficial owner of an item of income?), or the interpretation of tax treaties (is a Member State granting illegal State aid by interpreting treaties in a static or dynamic manner? This question is relevant for the McDonald’s case). The list of questions is of course endless. Consequently, the Union courts will have to make an important choice as to the determination of the content of the reference system.
Last, the type of structure used in the Engie case illustrates the more general issue of hybrid mismatches, where e.g. the same instrument is qualified as both debt and equity, or the same entity is deemed both opaque and transparent. In many cases hybrids result from the differences existing between the domestic laws of different countries. Although hybrids may be in line with the laws of each country, such arrangements produce an avoidance of tax on a global level. The Engie decision does not address the question of cross-border hybrids (point 225), which nevertheless deserves some comments given the way of reasoning developed by the Commission in other cases, especially the Apple case. How to prevent the tax consequences of cross-border hybrids is a complex matter, when hybrids simply result from differences in the legislations of different countries. Hybrid mismatch arrangements have recently been subject to substantial efforts by the OECD and the European Union, so as to eliminate their effects for corporate income tax purposes. In 2015 the OECD issued a report of 458 pages that analyses the issue at depth and contains suggestions to tackle it. The Member States are currently implementing the anti-tax avoidance directive (directive 2016/1164, as amended by directive 2017/952), which partly aims at neutralising the effects of hybrid mismatch arrangements (see article 9 of the directive). A question that can be discussed in the light of the Engie case (although it was not the issue at stake in Engie) is whether the Member States may have been infringing on the EU State aid rules by not preventing the effects of cross-border hybrid mismatch arrangements (the issue is mostly relevant for the period prior to the implementation of the anti-tax avoidance directive). In the case of cross-border hybrids the question is difficult to answer, for the simple reason that it supposes (i) to accept that the idea that hybrids are incompatible with State aid law, and if this idea is accepted (ii) to determine which country to blame. For example, in the case of an arrangement that Member State A qualifies as loan (with the consequence that related payments are deemed as interest expenses and are thus deductible for tax purposes) and that Member State B qualifies as equity (with the consequence that related receipts are deemed as tax-free dividends under a participation exemption regime), the result is the deduction of interests with no corresponding inclusion of income: if indeed cross-border hybrids are incompatible with State aid law, has Member State A been granting illegal aid by qualifying the payment as interest, or has Member State B been granting illegal aid by qualifying the receipts as dividends? Under a traditional State aid analysis, confirmed by the recent Heitkamp BauHolding case (Case C‑203/16 P), if each Member State has acted in accordance with its reference system as determined by its domestic law, there is no deviation from the reference system, and thus no potential selectivity. Even the de facto selectivity concept is hard to apply in this case, because the beneficial effect is the consequence of the legislations of two Member States, as opposed to e.g. the Gibraltar case (Joined Cases C‑106/09 P and C‑107/09 P) where the beneficial effect was the result of one set of rules. At the same time, the tax advantage of certain cross-border hybrids are normally not available in domestic situations. Multinational enterprises may thus enjoy a competitive advantage that is not available to independent enterprises or domestic groups. In this respect, the question of cross-border hybrids shares an issue of principle with the Apple decision: in both cases it can be wondered whether the effect of the lack of legislation may contravene on the State aid rules. The European Commission has argued that the Member States must enforce the arm’s length principle as a consequence of the principle of equal treatment embedded in article 107(1) of the TFEU, even when they lack such a legislation. While this position may be discussed, in my opinion the lack of legislation may, in certain cases, produce effects that contradict the objective of a tax system (assuming the objective is the taxation of net income), to the benefit of only certain taxpayers. Introducing anti-avoidance rules could thus be seen as a means to ensure the attainment of the objective of a tax system, be it by the application of different rules to different categories of taxpayers, in order to eventually tax them in a similar manner. This way of reasoning could also be supported by an analogy to the case law of the CJEU in free movement law, in which the Court has accepted the prevention of the double use of losses as a valid justification (for a recent example see Case C-28/17). Consequently, were the Union courts to confirm the way of reasoning of the European Commission in Apple, it would need to be questioned whether the State aid rules also prevent cross-border hybrid mismatch arrangements.
As the various decisions of the Commission on tax rulings share several questions of principle, certain elements of answer to the open questions discussed here will be provided with the first cases of the Union courts. It will probably be necessary to wait until the CJEU rules on these issues to have some guidance on the application of State aid law to various international tax issues. Yet we know from other areas that a decision from the Grand Chamber may not be the end, but rather the beginning of the discussion.
Associate Professor of International Tax Law, Stockholm University. Jérôme Monsenego is the author of the book “Selectivity in State Aid Law and the Methods for the Allocation of the Corporate Tax Base” (Kluwer Law International 2018), where many of the issues relevant for the selectivity test of tax rulings are analysed.
 The authentic language of the decision is French. The English version (non authentic) is available here: < http://ec.europa.eu/competition/state_aid/cases/266094/266094_2009354_269_2.pdf >.
 The report can be found at: < http://www.oecd.org/tax/neutralising-the-effects-of-hybrid-mismatch-arrangements-action-2-2015-final-report-9789264241138-en.htm >.