State aid law is closely linked to the Treaty on the Functioning of the European Union (TFEU) commitment to creating, maintaining, and strengthening a common market in which competition is not distorted. Precisely what characterises companies that form part of a group of companies (multinational or not) is that transactions carried out between them are not conditioned to the free competition that governs those carried out by independent companies. This situation inherent to corporate groups may lead to distortions from a tax point of view, mainly corporate taxation. Taking into account the purpose of the State aid law and the singularities of the transfer pricing regime, it should not be surprising that the interaction of the transfer pricing regulation and its practical application, on the one hand, and the prohibition of State aid, on the other, give rise to tensions and legal problems that have arisen for decades, which are shortly analysed in this post.
The disputes concerning the transfer pricing regime and the prohibition of State aid can be classified into two broad groups or categories.
First, problems of compatibility with the aforementioned State aid prohibition regime have arisen concerning the transfer pricing regimes established by certain Member States and, in particular, concerning certain special rules laid down for specific transactions or types of taxpayers (as in the “Belgian cases” on excess profit exemption and coordination centres tax regime).
Second, compatibility problems with the prohibition of State aid have arisen concerning the application of the transfer pricing regime provided for in the regulations of the Member States by the corresponding tax administration and, more specifically, with the administrative tax rulings carried out in the framework of addressing market value. This second group includes cases such as Starbucks, Amazon and Fiat, which are even more complex than the previous ones.
This second group of cases, which the Fiat case law discussed in this post belongs to, presents some differences to the first group, which, in our opinion, are of enormous relevance for our analysis. In the first group of cases, it is a legal regime potentially considered State aid. In the second group, what is potentially considered to be State aid is not the legal regime itself (the regulation foreseen for related-party transactions). Instead, it is the application or interpretation made by the tax administration through a tax ruling.
Differentiating between the two types of situations is essential when delimiting the impact of the State aid law in transfer pricing. Moreover, it allows us to understand, among others, the reasoning contained in paragraph 122 of the CJEU’s ruling in Fiat, which is being highly questioned by the doctrine.
The conditions that must be fulfilled for a national measure to be considered State aid within Article 107(1) TFUE are well known. There is also a consensus regarding selective advantage as the most challenging aspect when addressing the requirements. Therefore, to classify a national tax measure as a selective advantage, the EU Commission has to follow the next familiar steps:
- It must begin by identifying the reference system or benchmark, i.e., the ‘normal’ tax system applicable in the Member State concerned.
- It must assess whether the tax measure at issue is a derogation from that reference system in so far as it differentiates between operators who, in the light of the objective pursued by that system, are in a comparable factual and legal situation.
- The Member State concerned cannot demonstrate that differentiation is justified because it flows from the nature or general structure of the system of which those measures form part.
The first problem concerning the application of the prohibition of State aid regarding integrated undertakings transactions is to define the benchmark.
Initially, the EU Commission – e.g., Belgian excess profit exemption (paragraph 121 et seq.) – considered the benchmark or reference system to be the standard corporate taxation system, the objective of which is the taxation of the profits of all companies, subject to corporate taxation. However, in the same decision (paragraph 144 et seq.), the EU Commission introduced a new idea for which the Belgian regime granted a selective advantage because it departed from the arm’s length principle. With this argument, the EU Commission made a very significant qualitative leap forward because the reference system was no longer the standard corporate income tax system but rather the arm’s length principle. Moreover, such a principle was not taken from the domestic tax system, which was disregarded, but from Article 107(1) TFEU itself.
This novel reasoning was possibly motivated by an attempt to overcome the Irish rules in force in the years at issue in the Apple case, which were investigated by the Commission and considered to be State aid in a Decision annulled by the General Court.
Be that as it may, this criterion was set out in the Commission Communication of 2016 and has been used in other subsequent cases related to the transfer pricing regime, and in particular, in those relating to the application of this regime through tax rulings such as the Fiat case addressed here.
The CJEU (paragraphs 72-74) resolves this first question by clarifying that the benchmark is the national law and is not a hypothetical arm’s length principle taken from Article 107 TFEU. Expressly, when affirmed according to the CJEU, “[…] It follows that only the national law applicable in the Member State concerned must be taken into account in order to identify the reference system for direct taxation, that identification being itself an essential prerequisite for assessing not only the existence of an advantage, but also whether it is selective in nature”. It is, therefore, crystal clear that, when assessing the possible aid character of a transfer pricing measure or the application (i.e., interpretation) of the transfer pricing regime, the EU Commission must inevitably look to national law and, on that basis, proceed with the analysis.
Scholars questioned that the CJEU has completely closed the door to the previous EU Commission’s doctrine as long as paragraphs 70 and 122 of the ruling leave room for its interpretation. Although, in our view, paragraphs 70 and 122 warn the Member State that when applying the arm’s length principle, it is nevertheless possible that the design of such a regime and its application by the tax administration of that Member State may give rise to a de facto selective measure. Therefore, Member States must be careful not only with their transfer pricing regulation but also – and above all – with their application, particularly in the context of tax rulings. In our view, the content of paragraphs 70 and 122 in no way contradicts the doctrine that the CJEU has clearly established in the sense of considering exclusively the domestic law adopted by the Member State as the benchmark and in no case resorting in an abstract manner to an alleged principle extracted from Article 107 of the TFEU as the EU Commission supported and the General Court admitted.
To determine whether there is an advantage stricto sensu, the second step in the State aid analysis requires comparing the benchmark with the measure under scrutiny. Again, the difference between the transfer pricing regime and the application of the transfer pricing regime in a tax ruling is essential. In the Fiat case (as in the Starbucks and Amazon cases), it is not the Luxembourg transfer pricing regime that is at first sight at issue but the application of that regime through a tax ruling. In these cases, for the tax ruling to deviate from the benchmark and constitute an advantage, it is necessary to analyse how the tax administration has applied the tax regime. In particular, whether the tax ruling respects the administrative margin of discretion inherent to the transfer pricing regime or, on the contrary, it has gone further, and the transfer pricing regime has been applied in an ‘arbitrary’ manner.
Applying the transfer pricing regime only allows for a mere value approximation. There is a significant margin of appreciation when establishing the scope and content of such a regime. Tax rulings are designed to provide legal certainty to taxpayers so that they can know with certainty, among others, how the arm’s length standard will be applied in a particular case. This administrative discretion inherent in any tax ruling is not problematic from the perspective of the prohibition of State aid. What is contrary to State aid law is for the ruling to endorse a transfer pricing outcome that does not produce a reliable approximation of market-based results following the arm’s length principle. The search for a reliable approximation of a market-based outcome’ means that any deviation from the best estimate of a market-based outcome must be limited and proportionate to the uncertainty inherent in the transfer pricing method chosen or the statistical tools employed for that approximation exercise. In this regard, it should be noted that non-compliance with the OECD’s methodological recommendations in determining the transfer price does not necessarily determine either the existence of a reduction in the tax burden or that any reduction in the tax burden – which is what must be assessed – ipso facto constitutes State aid as long as the starting point – i.e., benchmark – is the domestic legislation. It does not necessarily have to coincide with the OECD standard.
The EU Commission’s role must go beyond the mere effective control of the methodological correctness used in the tax ruling and its compliance or otherwise with the OECD Guidelines. It must demonstrate that the alleged methodological errors lead to an unreliable approximation of the arm’s length results and that it has the effect of reducing the tax burden in relation to that which would have resulted if an assessment had been made following the arm’s length standard provided for in the applicable domestic legislation, i.e., to use CJEU’s terminology, the abstract expression of that ‘principle’ taking into account how the said ‘principle’ has been incorporated into the domestic law to integrated companies in particular. It is, therefore, not surprising that the General Court had criticised the EU Commission for limiting its analysis to questioning the methodology validated in the ruling and its plausibility rather than setting out the appropriate market price range.
Unfortunately, the CJEU did not rule on Fiat on this essential issue. Instead, the CJEU merely notes the existence of an error of law in determining the ‘normal’ taxation or benchmark applicable in the Member State concerned that necessarily invalidates the entirety of the reasoning relating to the existence of a selective advantage. In this way, the CJEU refuses to enter into an assessment of the EU Commission’s subsidiary argument (paragraphs 318 et seq.) accepted by the General Court, under which it concludes that tax ruling is also considered State aid, taking as a reference system the Luxembourg domestic regime. It would have been desirable for the CJEU to have entered into a substantive analysis of such a relevant and definitive issue in the analysis of whether or not the measure is considered State aid.
Having clarified that the reference framework is domestic law, the problem now shifts to the analysis of the domestic standard itself and the application of that standard by the tax administration in the specific tax ruling. Can such a vague standard as the arm’s length be used as a benchmark? More specifically, if the domestic regulation’s implementation of the arm’s length standard leaves a wide margin of discretion to the tax administration, is it possible to infer that it constitutes state aid? This question underlies the EU Commission’s interesting subsidiary argument in Fiat (paragraphs 321-325). Even if the scheme is well designed and does not leave too much discretion to the tax administration, what characteristics must the tax ruling have for it to be regarded as State aid? On these essential aspects, there are still too many shadows.
In all pending cases, the EU Commission’s reasoning is formally the same as in Fiat. It would therefore be foreseeable that the CJEU would maintain a similar line of reasoning to that in Fiat. However, having clarified that the benchmark is domestic law, it would be more desirable for the CJEU to address other crucial questions on which it has not ruled in this case. In other words, it should go into the problematic assessment of the delimitation and application of the national regime for integrated undertaking transactions.
The EU Commission has lost a battle, but not the war. In addition to clarifying some elementary points, the CJEU has given the EU Commission a wake-up call to address these issues with greater technical precision if it wants to succeed. The short concluding paragraphs of the ruling are intended to guide the EU Commission on how to approach this analysis. In the future, the EU Commission will have to depart from the domestic regime of the Member State – i.e., domestic transfer pricing rules – and, departing from the domestic regime, carry out its own transfer pricing analysis to address whether the tax ruling under scrutiny qualifies as a reliable approximation of a market-based outcome or not.
 On the distinction between a measure’s advantage and selectivity character, see López López, H. General Thought on Selectivity and Consequences of a Broad Concept of State Aid in Tax Matters. European State Aid Law Quarterly (EStAL), 2010, 4, 807-819.
 This argument is also in the Fiat Commission Decision (vid. paragraph 219 et seq.).
 See Lang, M.; Zeiler, A.: “Discretionary Power of Tax Authorities as a State Aid Problem”, EU Tax Law and Policy in the 21st Century, Wolter Kluwer, 2017, p. 91 et. seq.
More from our authors:
Schwarz on Tax Treaties, Sixth Edition
by Jonathan Schwarz