On 17 December 2018 the European Commission issued the public version of its decision in the McDonald’s case (SA.38945). The Commission found, contrary to its initial conclusion in the opening decision, that Luxembourg did not grant illegal State aid to McDonald’s as a consequence of the exemption of income attributed to a US branch. If the outcome seems reassuring when being compared to the opening decision of 2015, it is nevertheless puzzling in several respects and leaves various questions unanswered as to the application of State aid law to tax treaty provisions, and more generally to certain international tax rules.
The facts were relatively simple. The McDonald’s group is primarily a franchisor, with franchise rights originally developed in the US. A Luxembourg group company made a buy-in payment to enter a cost sharing arrangement with a US related company, so as to acquire the beneficial ownership of certain existing and future franchise rights. These rights were then allocated to the US branch of the Luxembourg company. The royalty fees due by franchisees would first be paid to a Swiss branch of the Luxembourg company, which provided services associated with the franchise rights. The royalty fees would then be transferred to the US branch, deduction being made of a service fee to the benefit of the Swiss branch consisting of cost coverage, plus a profit mark-up.
While the royalty fees ended back in the US, no tax was levied there. This was due to the fact that under US law, the business carried out there did not constitute a trade or business. At the same time, the income allocated to the US branch was not taxed in Luxembourg. In principle this item of income should have been taxed in Luxembourg, as this country normally taxes its residents on their worldwide income. However, the treaty seemed to prevent the residence State from taxing, by the effect of a twofold mechanism: first, the US activity would constitute a permanent establishment (PE) under the Luxembourg interpretation of the treaty. Second, the existence of such a PE would oblige Luxembourg to apply the article on the elimination of double taxation. In the treaty with the US, Luxembourg chose the exemption method. No subject-to-tax clause was added. As a result, Luxembourg considered in a tax ruling that it had to exempt the income, although it was not taxed in the US.
The question, from a State aid point of view, was whether Luxembourg granted illegal State aid by applying the exemption method, something that supposes a departure from the normal application of the set of rules in this situation. The Commission concluded that Luxembourg did not give a selective advantage to McDonald’s by exempting the income allocated to the US branch. This conclusion seems to be a correct interpretation of the State aid rules, assuming the exemption of US profits is indeed the correct consequence of the reference system. The application of the exemption method seems at first sight to be correct: on the one hand, the drafting of the treaty was clear about the fact that what mattered for the application of the exemption method was that income “may be taxed” in the US, as opposed to “effectively taxed” there; on the other hand, a Luxembourg court case (Cour administrative du Grand-Duché de Luxembourg, case number 14442c of 23 April 2002) had found this situation of double-non-taxation to be the correct application of the treaty, which was confirmed by 25 other tax rulings examined by the Commission. The commentary on article 23 of the OECD Model that existed at the time of the conclusion of the treaty made also clear that the State of residence must give exemption whether or not the right to tax is in effect exercised by the other State (para 34 of the commentary on article 23 of the 1992 OECD Model). I have argued elsewhere for the need to limit State aid control to the correct application of the reference system, both ratione materiae and ratione temporis (Selectivity in State Aid Law and the Methods for the Allocation of the Corporate Tax Base, Wolters Kluwer 2018, pp. 23-56). An opposite approach would imply that State aid law imposes a given material content to the tax systems of the Members States, although there is no objective method to determine such a material content and to apply it in a way that ensures legal certainty. A similar view is expressed by Advocate General Saugmandsgaard Øe in the A-Brauerei case (C-374/17, see mostly at paras 136-138). It is also reassuring that the Commission did not, as in the opening decision, expect Luxembourg to interpret the exemption article in the light of paragraph 32.6 of the most recent commentary on article 23 of the OECD Model (see para 90 of the opening decision). Indeed, this paragraph is more than a mere clarification and was introduced in 2000, ie after the conclusion of the US-Luxembourg treaty in 1996. The reason for setting aside paragraph 32.6 of the commentary on article 23 is nevertheless puzzling, as the Commission refers to the partnership report, and not to the lack of retroactivity of new commentaries that affect the interpretation of a treaty provision.
It results from the foregoing that a central part of the case resides in whether or not Luxembourg correctly applied the reference system, something that supposes that a PE indeed existed under the Luxembourg interpretation of the treaty. This is a prerequisite for applying the exemption method. The ruling concluded to the existence of a PE under Luxembourg law, which would legitimate the exemption method. In this respect, it is surprising that the Commission did not analyse more in depth this point in the light of the State aid rules. There seems to be arguments that could at least question this conclusion. For example, the US branch had no employees (point 22 of the decision), and from a US perspective it was considered that the “primary” business was not conducted through the branch (point 40), which means that the branch might only conduct secondary activities, leading to the lack of trade or business in the US. Is that not an indication of a potential activity that is preparatory or auxiliary? Article 5(4)(e) of the US-Luxembourg treaty included a provision stating that “the maintenance of a fixed place of business solely for the purpose of carrying on, for the enterprise, any other activity of a preparatory or auxiliary character” does not constitute a PE. This exception was potentially valid here, in which case the decision on the existence of a PE could violate the treaty because the treaty requires the application of the exemption method only if income may be taxed in the US “in accordance with the provisions” of the convention (article 25(2)(a) of the treaty). Finding a PE although it falls under the exception for auxiliary activities, and granting an exemption, could then constitute a selective advantage. Moreover, even if a PE indeed existed in the US under the Luxembourg interpretation of the treaty, income should not necessarily be attributed to that PE: although article 7 of the treaty did not apply the authorized OECD approach (requiring the presence of significant people functions for the attribution of profits), article 7(2) implemented the traditional separate entity approach. In this respect, it is only the profits that an independent enterprise acting in comparable circumstances would have earned that should be attributable to the PE, and exempted in the State of residence. The remaining profits should be exclusively taxed in the residence State, ie Luxembourg. Surprisingly, the Commission did not – at least explicitly – address this issue.
The text of the decision is also appealing as it does not contain an investigation of the transfer pricing aspects of the case. For instance, no analysis is made of the conditions for the buy-in payment. However, an interesting element in the decision is the mention that the Luxembourg company had no employees. The application of the arm’s length principle might have subjected the right of deduction of the buy-in payment to a certain substance requirement in order not to acknowledge the ownership of intangible property by solely incurring costs. Accepting the deduction of this cost without investigating the substance supporting the management of the beneficial ownership of the franchise rights may, in this case, have contradicted the arm’s length principle, assuming an independent company acting in similar conditions would not have bought franchise rights without having employees able to manage such rights and assess the risks of this investment. This could constitute illegal State aid, if one follows the interpretation of the arm’s length principle in the decisions of the Commission in the transfer pricing cases or in the OECD transfer pricing guidelines. The decision does not either contain an analysis of the conditions under which the franchise rights were transferred to the US branch.
The McDonald’s case raises other types of questions. A first question, which has to do with the major argument of the Commission in the on-going transfer pricing cases, is whether the deviation from the reference system should be limited to an assessment of the application of the legal framework in force, or if external elements may be imposed by State aid law. In cases such as Starbucks, Fiat and Apple, as well as in its 2016 notice, the Commission concluded that article 107(1) TFEU imposed taxing members of multinational enterprises on the basis of the arm’s length principle, even if such a principle is not part of the reference system. That would be a consequence of the principle of equal treatment. However, the solution in McDonald’s might lead to certain differences in treatment, if one takes into account elements external to the Luxembourg tax system. For example, differences may exist between activities in the US that constitute or do not constitute a trade or business, the latter ones not being subject to tax. Differences exist between PEs present in countries with which Luxembourg in the treaty has chosen different methods to eliminate double taxation, such as the credit method, the exemption method, or the exemption method combined to a subject-to-tax clause. Differences also exist between multinational and domestic groups, the former having higher chances of achieving double-non-taxation. In the transfer pricing cases, the Commission relied on an EU version of the arm’s length principle, which by its universal content would mitigate certain differences in treatment so as to reach a more homogeneous taxation of multinational enterprises. In the McDonald’s case, the Commission accepts more the peculiarities of the tax system of a single Member State, without imposing a universal method to tax multinational enterprises or to design tax treaties, or to interpret such treaties. While I do see the merits of this approach, it can be wondered whether the Commission has changed its view on the determination of the material content of the reference system? Or should different approaches be applied to tax treaty law and transfer pricing? Last, the point made that the assessment of 25 other tax rulings would demonstrate that Luxembourg did not give a selective treatment as it would in most cases apply tax treaties as in the McDonald’s case might, in my opinion, be at odds with the view developed by the CJEU in Gibraltar, where a tax system favouring numerous undertakings could imply illegal State aid: the high number of undertakings that benefited from double-non-taxation should not preclude the selectivity of the rule, at least if one follows the logic developed in Gibraltar.
Finally, it should not be forgotten that this case does confirm the applicability of State aid law to the interpretation of tax treaties. This raises the question of how to combine these sources of law. Many questions arise and are not answered in this case, such as whether or not State aid law should influence the reliance on the OECD commentaries, how to assess the application of subjective treaty measures such as the notion of beneficial owner or the principal purpose test, or whether the denial or the grant of treaty benefits may constitute State aid. As it is now unlikely that more guidance will come from the McDonald’s case, we will need to wait for future cases in order to better understand the relation between State aid law and tax treaty law.
Jérôme Monsenego is Associate Professor of International Tax Law at Stockholm University. He is the author of Introduction to Transfer Pricing (Wolters Kluwer 2015), and Selectivity in State Aid Law and the Methods for the Allocation of the Corporate Tax Base (Wolters Kluwer 2018).