The General Court in Amazon and Engie: A New Effect-based Approach Aimed at the Endorsement of the 'Vestager Doctrine'?

Printer-friendly version

Abstract: The rulings of the General Court in the cases of Amazon and Engie (respectively, joined cases T-816/17 and T-318/18 Luxembourg. v Commission ECLI:EU:T:2021:252 and joined cases T-516/18 and T-525/18 Luxembourg v Commission ECLI:EU:T:2021:25) are the last two episodes of the “tax ruling saga”. This Insight seeks to examine the potential innovations that these judgements could bring in the context of the application of State aid rules to unfair tax practices. Indeed, this contribution attempts to analyse the impact of the abovementioned rulings on both the notions of “State origin” and “selectivity”, as well as the new effect-based approach proposed by the Commission and endorsed by the Court. To do so, both previous relevant cases of the “saga” and the current political context are taken into consideration. The Insight concludes that an outcome-based perspective is not enough to render the “Vestager doctrine” an efficient instrument in the fight against harmful tax competition.

Keywords: State aid – tax rulings – Amazon – Engie – regulatory competition – regulatory arbitrage.

I. Introduction

On 12 May 2021 the General Court of the European Union (GC) issued two judgements in the context of the so called “tax ruling saga”. The first is the judgement in cases T-816/17 and T-318/18 (hereinafter Amazon);[1] the second is the judgement in cases T-516/18 and T-525/18 (hereinafter Engie).[2] They follow the previous decisions of the GC in the cases of Belgium Excess Profit,[3] Starbucks,[4] Fiat,[5] and Apple.[6]

In particular, in the Amazon ruling the GC annulled the decision (Amazon Decision)[7] of the European Commission (the Commission), which had found the tax ruling granted by Luxembourg authorities to the Amazon group to be incompatible State aid worth about EUR 250 million. On the other hand, in Engie the GC upheld the Commission’s decision (Engie Decision),[8] which found the tax rulings granted by the Luxembourg tax authorities to the French group Engie to be incompatible aid. Engie is a multinational corporation which operates in the field of low-carbon energy and services, and which, according to the Commission, received aid worth around EUR 120 million.

This Insight, after a short presentation of the facts of both cases (section II), will describe the most relevant topics addressed by the Court in each judgement (section III and IV). The main critical issues will then be examined and commented upon, while taking a brief look at the other episodes in the “tax ruling saga” (section V). Although the different facts, outcomes, and implications of these judgements are to be dealt with separately, some common conclusions will nevertheless be drawn (section VI). More specifically, the Insight will consider the potential impact of the judgements on the assessment of compatibility with EU law of the “Vestager doctrine”, i.e., the application of State aid law to tax rulings in order to tackle harmful tax competition.

Lastly, on 22 July 2021 both judgements were appealed before the Court of Justice (ECJ) by the losing parties: the Commission in Amazon[9] and Engie group in Engie.[10]

II. The facts of the cases

The facts of Amazon can be summarised as follows. Between 2006 and 2014, Amazon EU SARL (LuxOpCo), a company incorporated in Luxembourg under Luxembourg law, paid significant royalties to another company in the group, Amazon Europe Holding Technologies SCS (LuxSCS), in exchange for the right to use the e-commerce platform used by the Amazon group in Europe under a “cost-sharing” agreement.

Pursuant to Luxembourg law, LuxSCS was treated as a “transparent” company,[11] thus its shareholders, who were U.S. entities, bore tax liability for its income. However, LuxSCS was not considered transparent in the U.S. This ambiguity caused a phenomenon of double non-taxation, given that the shareholders could indefinitely defer their tax liability in the U.S. To put it simply, taxes were paid neither in Luxembourg nor in the U.S.[12]

Moreover, the tax ruling granted by the Luxembourg tax authorities in 2003, and extended in 2011, endorsed an application of the arm’s length principle based on the transactional net margin method (TNMM).

To fully grasp the facts of the case, it might be helpful to take a step back and briefly explain what the arm’s length principle is. Companies which belong to the same group might enter into certain transactions with the sole aim of shifting their profits from one jurisdiction to another. In order to avoid the goods and services involved being overestimated for fraudulent purposes, it is necessary to identify the price that would have been applied to a comparable transaction if it was carried out between independent companies (i.e., companies which do not belong to the same group).[13] In other words, it is necessary to apply the arm’s length principle.

There are several applicable methods to measure the arm’s length. Amongst them, the TNMM is the most common. Indeed, it allows a group to shift most of its profits to the company which plays the most important role in the transaction. Multinationals prefer this methodology since it enables them to maintain that the most important functions are carried out by the company where intellectual property rights are allocated. In essence, intangibles can be strategically allocated in order to shift corporate income to lower tax jurisdictions.[14]

A similar scheme is at the core of the Amazon case. Indeed, according to the Commission, the tax ruling at stake allowed Amazon to shift the great majority of its profits to LuxSCS in order to avoid taxation, as it produced “a result that depart[ed] from a reliable approximation of an arm's length outcome” since it was based on “inappropriate methodological choices which result[ed] in a lowering of LuxOpCo's taxable income as compared to companies whose taxable profit reflects prices negotiated at arm's length on the market”.[15]

At the end of its investigation, the Commission issued a decision finding that the tax ruling concerned constituted State aid incompatible with the internal market and thus ordering the recovery of about EUR 250 million. Amazon and Luxembourg decided therefore to bring actions for the annulment of the decision.

Turning to Engie, here the facts were as follows. Several tax rulings granted by Luxembourg authorities between 2008 and 2014 endorsed a complex scheme of financial transactions between four companies of the Engie group, all incorporated in Luxembourg under Luxembourg law. More specifically, the transactions gave rise to a financial hybrid that caused double non-taxation, as the same income was simultaneously treated as debt  – a loan whose interests were deductible – and as equity – an investment exempted from taxation under Luxembourg tax law. According to the Commission, the scheme allowed Engie to avoid taxation on 99 per cent of the profits which had been shifted.

Notably, the Commission analysed the effects of the tax treatment conferred on Engie on three different levels:[16] the individual level of the companies which directly benefited from the tax exemption on the profits; the group level; and the Luxembourg level, meaning that the Commission found selectivity in the fact that Luxembourg decided not to apply its anti-abuse tax rules. Namely, the Commission referred to art. 6 of the Luxembourg Tax Adaptation Law or Steueranpassungsgesetz, which prohibits taxes being evaded or mitigated by abuse of forms or constructions which are per se perfectly lawful under civil law.[17]

The Commission concluded its investigation issuing a recovery order for around EUR 120 million. The decision was challenged before the GC by Luxembourg and Engie, which brought actions seeking its annulment.

III. The annulment of the commission’s decision in the Amazon case

In Amazon, the reasoning of the Court has much in common with its previous judgement in Apple. This is true with reference not only to the assessment of the Commission’s competence, but also to the focus on the burden of proof in State aid cases.

The ruling shares its starting point with the other judgements of the “tax ruling saga”: that there is no doubt that the Commission has the competence to scrutinize tax rulings under art. 107 TFEU, and that this behaviour does not encroach on Member States’ fiscal autonomy.[18]

After this first clarification, the Court insists on the burden of proof borne by the Commission.[19] As in Apple and Starbucks, the GC holds that the presence of a “methodological error”[20] in the application of the arm’s length principle endorsed by the tax ruling is not enough to prove the existence of a selective advantage under art. 107 TFEU. In fact, since Member States must be granted a certain margin of appreciation in the application of transfer pricing methods, the Commission must also show that such a methodological error coincides with a reduction of the tax burden that the undertaking would have borne in the absence of the tax ruling.[21] In other words, “the Commission cannot simply assume that because the Member State acted in a seemingly arbitrary manner the outcome was wrong”.[22]

The central part of the judgement deals with the Commission’s assessment (“the primary finding of an advantage”) and is built around three fundamental questions: which party should have been subject to the test in the application of the TNMM? Did the Commission succeed in proving the existence of an advantage? Was the arm’s length principle correctly applied?

As regards the choice of the party to be tested, it must firstly be remarked that the TNMM requires that the party of the intragroup transaction to be tested to determine the correct transfer price should be the one for which the “functional analysis” is easier. The purpose of such an analysis is to assess the functions performed, the assets used, and the risks assumed by each party involved in the intragroup transaction. Once a party is chosen and tested, the residual profits (i.e., those exceeding the profits that a comparable entity would have made in a comparable transaction carried out in conditions of free competition) go automatically to the other, non-tested party (i.e., the more complex entity).

The Commission believed that the Luxembourg tax authorities erred in choosing LuxOpCo as the party to be tested, since LuxSCS only performed very limited functions within the concerned transactions and was therefore the “less complex entity”.[23] In the view of the Commission, therefore, LuxSCS should have been the tested party.

However, according to the Court, the Commission failed in demonstrating its own thesis. Indeed, the GC observes that the Commission underestimated the functions performed by LuxSCS, which provided intangible assets with a “unique and valuable” role that finds no external comparable on the market.[24] Moreover, the GC adds, the Commission has not shown that the functional analysis of the activities carried out by LuxSCS would have been easier.[25]

In the analysis of the second question the Court goes into the details of the Commission’s assessment. In a nutshell, the GC finds that the existence of an advantage is not proved, as the calculation of the remuneration of LuxSCS determined by the Commission is affected by methodological errors.[26]

With reference to the third question, the GC carefully considers the distribution of the functions linked to the intangible assets at stake, categorised as “Technology”, “Customer Data”, and “Trademarks”. In brief, according to the assessment carried out by the Court, the functions performed by LuxSCS could not be considered as mere supply of “low value adding” services.[27] Therefore, the Commission’s application of the arm’s length principle led to an erroneous result in the determination of the net margin attributable to LuxSCS.

Once those main issues are addressed, the Court goes further and analyses the three “subsidiary findings of an advantage” forwarded by the Commission in its decision.

First, the GC observes that the Commission failed to demonstrate that the profit split method,[28] and not the TNMM, should have been used to calculate the remuneration of the concerned parties. Most notably, the Court criticises the Commission’s assumption according to which in this case the contribution analysis should have been preferred over the residual analysis.[29] “Although the Commission cannot be criticised for having used a transfer pricing method that it considers appropriate,” submits the Court, “the Commission is nevertheless required to justify its choice of methodology”.[30]

Second, the Court disagrees with the Commission’s decision where it deemed the choice of operating expenses as a profit level indicator to be incorrect. Especially, the GC holds that the Commission: did not prove that such a choice “led not only to a different outcome but also to a reduction in the tax burden of the recipient of the tax ruling at issue”;[31] did not prove that total costs (and not operating expenses) would have been the appropriate profit level indicator for LuxOpCo, especially in its activities as management company;[32] and did not examine the choice of the appropriate profit level indicator for LuxOpCo as “administrator of a marketplace for third-party sellers” and “online retailer”.[33]

Lastly, the GC analyses Amazon Decision with reference to the finding that the inclusion of a ceiling of 0.55 per cent of EU annual sales for the taxable income of LuxOpCo was not appropriate. According to the Court, “as inappropriate as the ceiling mechanism may be”,[34] the functional analysis of the activities of LuxOpCo carried out by the Commission failed to demonstrate that the remuneration resulting from the regime endorsed by the concerned tax ruling was underestimated.[35]

With all of this taken into consideration, the GC decided to annul Amazon Decision.

IV. The court dismisses the appeals in the Engie case: The tax rulings granted by Luxembourg to Engie breached EU Law

As had already happened in the case of Amazon, also in Engie, at the beginning of its reasoning, the Court reiterates that the Commission does not engage in any “harmonisation in disguise” when assessing tax rulings under art. 107 TFEU.[36]

Whereas such a starting point is not surprising, the attention of readers is captured by the parts of the judgement in which the Court addresses: the pleas concerning the concept of State origin; the distinction between the concept of “advantage” and that one of “selectivity”; and the endorsement of an effect-based approach to the assessment of the aid.

As regards the first matter, art. 107(1) TFEU must be preliminarily borne in mind. The provision requires that a measure, in order to constitute State aid, must be “granted by a Member State or through State resources”. In Engie, the Court confirms that tax advantages provided by means of tax rulings should be considered as conferred through State resources, as “it is not necessary to establish in every case that there has been a transfer of State resources for the advantage granted to one or more undertakings to be capable of being regarded as a State aid within the meaning of art. 107(1) TFEU”[37] Indeed, the fundamental element that defines State origin is the fact that a public authority has been involved in the adoption of the concerned measure.[38]

 The second issue discussed is the blurred notion of “selective advantage”. Indeed, Engie and Luxembourg maintained that, given that the presence of an economic advantage and selectivity are two distinct requirements for a measure to constitute State aid, they should be separately analysed. The GC submits that, in fiscal matters, there is an overlap between the assessment of the advantage and that of selectivity since both those criteria are aimed at proving that a preferential tax treatment was accorded to a certain undertaking, when compared to the treatment conferred by the law to other undertakings in a comparable legal and factual situation.[39] More specifically, the Court underlines that, as “the finding of a derogation from the provision on abuse of law simultaneously entails the grant of an advantage”,[40] the Commission is not confusing the two criteria when observing that, in the absence of the tax ruling, the companies concerned would not have benefitted from the advantageous treatment that they received.

The last relevant feature of the judgement seems to be the general endorsement of an effect-based evaluation of the measure concerned when the Commission is assessing alleged aid. In Engie, the Court seems to acknowledge the appropriateness of such an approach on two levels: on the one hand, by focusing on the economic substance of the concerned transaction rather than on the financial scheme adopted; on the other hand, by looking at the effects of the arrangement in order to determine its abusive nature.

First, in its reasoning, the Court endorses the Commission’s approach, inspired by the teachings of the ECJ in Gibraltar[41] and based on the economic substance of the financial transaction rather than on its formal structure. Accordingly, the Commission did not err in finding a derogation from the reference system (Luxembourg tax law) where the amounts exempted at the holdings’ level were simultaneously deducted at the subsidiaries’ level.[42] In fact, the GC observes that art. 107 TFEU does not make any distinction between the techniques used to grant the aid, since the aid itself is defined by its own effects.[43] For this reason, even though there is no violation of the individual provisions concerning the participation exemption regime and interest deduction respectively, the substantial violation concerns the combination of those provisions.[44]

Second, an interesting and innovative part of Engie is the endorsement of the Commission’s assessment of the Luxembourg anti-avoidance rule. The GC observes that, once the abusive nature of the arrangement is demonstrated, the non-application of such a rule does indeed constitute a derogation from the reference tax system of Luxembourg law.[45] Moreover, the GC rejects any claim concerning a potential breach of freedom of establishment (art. 49 TFEU) by recalling the recent ruling of the ECJ in the N Luxemburg 1 (hereinafter Danish Beneficial Ownership Cases), according to which “any finding that there is an abusive or fraudulent arrangement […] would also result in the fundamental freedoms guaranteed by the FEU Treaty being inapplicable”.[46]

Therefore, in light of the above, the GC rejected the appeals and confirmed the Commission’s decision.

V. Applying an effect-based approach to State aid control is not enough to effectively tackle harmful tax competition

The two rulings provide readers with food for thought. Indeed, EU lawyers are getting used to the depth of the GC’s investigations concerning the Commission’s practice within the field of State aid assessment. Nonetheless, it might prove hard to understand the import of these judgements without contextualization from the other episodes of the “tax ruling saga”.[47] Therefore, also the latter will be taken into account in this section. First, the issue of retroactivity will be briefly mentioned. Second, this section will try to assess the potential impact of Amazon and Engie on the concepts of State origin and selectivity. Last, the focus will shift to what seems to be the Court’s new effect-based approach towards the relationship between State aid and tax rulings.

In Amazon the GC observes that the Commission cannot retroactively apply the non-binding principles of the OECD Transfer Pricing Guidelines, described as a mere non-binding tool.[48] Such a finding becomes particularly interesting when considering that the retroactivity of the Commission’s interpretation of the arm’s length principle was part of the pleas in law of Luxembourg in Fiat and Ireland in Apple. This matter acquires huge procedural significance given that the protection of legitimate expectations and the principle of legal certainty are two grounds on which a recovery order can be countered.[49] Readers might see in Amazon the confirmation of the non-retroactivity of the Commission’s interpretation of the arm’s length principle. This might have further implications when considering that a detailed reference to the assessment of tax rulings in the Commission’s official documents appeared for the first time only in 2016 “Notice on the notion of State aid”.[50]

Another relevant issue when taking into account the whole “tax ruling saga” are the considerations of the GC in Engie concerning the concept of “State origin”. Although that part of the judgement might seem obvious, the nature of “State resources” of tax rulings has been debated. Certain authors have argued that “tax ruling cases use theoretical benchmarks to identify State resources”,[51] meaning that the Commission assumes that the selective advantage conferred to multinationals corresponds to a loss of tax revenue for the State. According to those scholars, that presumption fails to be accurate to the extent to which the Commission seems to ignore the phenomena of regulatory competition and regulatory arbitrage.[52] Indeed, it is unlikely that without the ruling the State would have attracted the multinational to which the ruling itself has been granted.[53] Thus, from this practical point of view, there is no loss but only a gain for the State which grants the ruling. Hence, considering tax rulings as conferred through State resources would result in a misconception. This is the very reason behind Member States’ interest in challenging decisions which entitle them to recover hundreds of billions of euros.

Such an argument was indirectly put forward before the GC. In Fiat, Luxembourg complained about the potential “serious economic repercussions” of recovery orders by reference to recent critiques addressed to the Commission in a whitepaper issued by the U.S. government.[54] The fear concerned a loss of confidence amongst U.S. multinationals which could potentially undermine the attractivity, and thus the income, of Member States such as Luxembourg or Ireland.

The Court did not show sensitivity towards that logic. Instead, not only did it observe that such a reason could not fall within the scope of art. 16(1) of the Procedural Regulation[55] (which prohibits recovery when it would be contrary to general principles of Union law), but it also argued that “it is clear that the recovery of the aid at issue cannot, as such, have negative economic effects for the Grand Duchy of Luxembourg, since the sums recovered are allocated to its public finances”.[56] Leaving aside recovery orders, and looking at the statement through the different lens of State origin assessment, it seems that the Court confirmed that a broader phenomenon, such as regulatory competition, cannot be taken into account when evaluating State aid.

In Engie, the Court sticks to its point. Accordingly, in the “tax ruling saga” State origin is not in question. Further clarifications would have been desirable, since the GC’s reasoning only refers to previous case law and does not directly address the abovementioned criticism.[57]

State origin has not been the only element of art. 107 TFEU in the spotlight during the “tax ruling saga”. In fact, authors have widely discussed the notion of selectivity.[58] It is unfortunate that in Amazon the finding that there was no advantage has prevented the Court from analysing such a matter, as had already happened in Starbucks and Apple. Nevertheless, some clarifications on selectivity came from Engie. Most prominently, the Court focused on the distinction between the existence of an advantage and the selectivity of the measure conferring such an advantage. The merger between those two concepts into the blurred notion of “selective advantage” derives from the presumption set by the ECJ in Commission v MOL.[59] Accordingly, in the case of an individual aid, “the identification of the economic advantage is, in principle, sufficient to support the presumption that it is selective”.[60] Such an interpretation has been criticised in the literature,[61] and, to a limited extent, by the Court itself.[62] With Engie, the Court may have in effect approved a joint assessment of advantage and selectivity. At least, as far as tax measures are concerned.

In Engie the Court has provided the State aid law landscape with interesting novelties. Most prominently, the endorsement of an effect-based approach[63] to the concept of abuse of law might have a significant impact on the tensions between free movement and regulatory arbitrage.[64] Can companies exploit their freedom of establishment in order to gain fiscal advantage? What are the limits?

In its ruling, the Court censures the tax strategies of the French multinational on two levels. First, in order to identify double non-taxation, the GC focuses on the overall economic substance of intragroup financial transactions, ignoring the individually lawful nature of each part of the arrangement. Second, it highlights the global effects of such an arrangement in order to qualify Engie’s behaviour as abusive.

Those two lines of reasoning follow a common pattern, which is the shift from a form-based towards a more effect-based approach. It is worth noticing that such a development is not new for EU competition law. The main examples are provided by the evolution of the ECJ’s jurisprudence in the application of art. 102 TFEU.[65] In a certain sense, it might be argued that State aid law is working as the bridge that allows Luxembourg judges to carry over this evolution from competition law – or, more precisely, antitrust law – to free movement rules. Once this step is taken, a new weapon against regulatory arbitrage and unfair regulatory competition could be deployed.

 Indeed, the Court seems to endorse the overruling of the “too formalistic”[66] criterion constituted by the “wholly artificial arrangements” set in the jurisprudence of the ECJ following Cadbury Schweppes and Cadbury Schweppes Overseas,[67] in favour of the principles laid down within the framework of the OECD BEPS[68] Project and already adopted in the Danish Beneficial Ownership Cases.[69] Accordingly, “the right of taxpayers to take advantage of competition engaged in by the Member States on account of the lack of harmonisation of taxation of income cannot be raised against the application of the general principle that abusive practices are prohibited”.[70]

Additionally, also in Amazon the Court highlights that “[a]rticle 107(1) TFEU defines a measure which reduces the burdens normally imposed on an undertaking in relation to its effects”.[71]

Such a shift in case law is coupled with EU hard law. Directive 2019/2121[72] has introduced pre-conversion, pre-merger or pre-division certificates as mandatory requirements for EU cross-border operations.[73] These “pre-operation certificates”[74] cannot be issued by Member States authorities “where it is determined in compliance with national law that a cross-border conversion is set up for abusive or fraudulent purposes leading to or aimed at the evasion or circumvention of Union or national law, or for criminal purposes”.[75] In other words, the effect-based a posteriori assessment of State aid can be told to find its counterpart in this a priori mechanism of control.

Although pre-operation certificates seem an interesting initiative, it cannot be ignored that the exclusive reference to national law risks somehow undermining their effectiveness. Such a reference, indeed, sets 27 different parameters against which abusive purposes are to be assessed. The directive shall be implemented by Member States by 31 January 2023.[76] By that time, it is desirable that EU courts set and consolidate common criteria.[77] Otherwise, the concrete danger is a fragmented and inconsistent enforcement of anti-avoidance measures.

As a matter of fact, coming back to the judgements, the new effect-based approach is not enough to regard as convincing the Commission’s strategy of tackling harmful tax practices through the backdoor[78] of State aid law. The case of Engie provides a good example.

First, it has been argued that the application by tax authorities of anti-abuse rules to tax rulings “makes no sense”,[79] as tax authorities could simply not issue the rulings in the first place. A possible rebuttal for this argument might be that, in theory, anti-abuse rules could represent grounds for refusing to grant the ruling. However, whether or not negotiating such rulings falls within the scope of national tax authorities’ discretion. Therefore, no grounds for refusing are in principle needed.

Second, it has been said that in Amazon and Apple the main focus of the GC is on the high burden of proof borne by the Commission to demonstrate selectivity. It is hard not to notice that this is in sharp contrast with Engie. Indeed, in the latter ruling the analysis of the reference system seems quite shallow, in the absence of any comparison with potential or concrete benchmarks. Again, prioritising substantive assessments cannot justify legal uncertainty nor the elimination of procedural guarantees.

Last, the assessment of the abusive nature of the arrangement is still carried out by the GC in accordance with (Luxembourg) national law. Even though anti-avoidance rules are now in principle set at the EU level,[80] this does not per se ensure a uniform transposition across the 27 national legal orders. This is especially true – similarly to what has been said for pre-operation certificates – in the absence of clear and consistent case law.

These concerns, in addition to the several unsolved questions regarding, amongst others, State origin and selectivity, suggest once again that State aid law is not the appropriate tool to engage in a solid and consistent fight against harmful tax practices.[81] As rightly observed by Advocate General (AG) Jääskinen’s opinion in Gibraltar, “harmful institutional or tax competition between Member States clearly does not fall within the mechanism for controlling State aid established by the Treaty” and “the legitimate objective of combating harmful tax competition cannot justify distortion of the European Union’s legal framework established in the area of competition law applicable to State aid, or even the adoption of ad hoc solutions conflicting with the rule of law as enshrined in art. 2 TEU”.[82]

Furthermore, it is interesting to notice a coincidence that might have significant political weight. Indeed, in the “tax ruling saga” the strategy of the Commission has so far been successful only against European companies (Fiat and Engie), whereas it has failed in the cases of U.S. multinationals (Starbucks, Apple, Amazon, and, to a certain extent, McDonald’s).[83] It has been observed that this trend might prove helpful in the ongoing negotiations for the development of multilateral anti-avoidance tools.[84] Interestingly enough, when commenting those judgement, Executive Vice-President M. Vestager took the opportunity to declare that the EU is “close to achieving a historic global agreement on the reform of the international corporate tax framework”.[85]

Six days after the publication of Amazon and Engie, on 18 May 2021, the Commission published its communication “Business Taxation for the 21st Century”.[86] This document represents a step forward when compared to the communications “Tax Good Governance in the EU and beyond”[87] and “An Action Plan for Fair and Simple Taxation Supporting the Recovery Strategy”.[88] Indeed, considering also that the European Council recently reiterated its “strong preference for and commitment to a global solution”,[89] the EU welcomes the considerable boost coming from the new U.S. administration. More specifically, the discussion concerning the digital levy is now coupled with the debate about a minimum global corporate tax.[90]

At the beginning of June, in the context of the G7, an agreement for a 15 per cent global minimum rate was reached.[91] On 10 July 2021 such an agreement was ratified during the G20 meeting.[92] However, a few EU Member States involved in negotiations within the G20/OECD Inclusive Framework refused to sign up. This is the case of Ireland, Estonia, and Hungary, champions of tax competition.[93]

The outcome might be a political dichotomy with unforeseeable consequences affecting the other pillar of negotiations, i.e., the digital levy that the EU is going to apply to U.S. multinationals.

In this fragmented context, the bargaining position of the Commission might be strengthened if the ECJ endorses the “Vestager doctrine”.[94] However, the analysis carried out in this Insight suggests that this might not be the case. In fact, the use of State aid law for such purposes does not seem appropriate. Nor do the (albeit interesting) novelties brought by Amazon and Engie render it less problematic. We have no option but to await the outcome of the appeals proposed before the ECJ.

Lastly, and for the sake of completeness, it should be remembered that the Commission has four ongoing in-depth investigations at the time of writing. Three investigations concern other tax rulings granted by the Netherlands in favour of Nike[95] and Inter IKEA[96] and by Luxembourg in favour of Huhtamäki.[97] The fourth ongoing investigation concerns the reopening[98] of the Belgium Excess Profit case after the annulment of Commission’s decision by the General Court.[99]

VI. Conclusions

The judgements of the GC in Amazon and Engie cannot be regarded as revolutionary, since the Court avoids taking a firm and ultimate position on the “Vestager doctrine”. They do, however, introduce significant novelties in the field of EU State aid law. This is particularly true with reference to both specific matters such as the notions of “State origin” and “selectivity” and more general considerations such as the endorsement of a less taxpayer-friendly approach to the assessment of abusive practices.

It is foreseeable that media attention and, consequently, public opinion will be focused on the Amazon ruling. In fact, this is not the first time that Amazon’s EU tax practices find themselves in the public eye. In 2017, a much quoted report showed that over the period 2013-2015 Amazon registered almost no profits within the EU, with an estimated profitability between 0.3 and five per cent, and even a net loss in 2014.[100] Notably, one might have expected a follow-up on the obiter dicta of Apple, where the Court took the opportunity to express regrets about the “the incomplete and occasionally inconsistent nature of the contested tax rulings”.[101] Instead, none of those comments has been reiterated in Amazon, save for the ill-concealed critique concerning the “inappropriate”[102] ceiling mechanism endorsed by the tax ruling.

On the one hand, in Amazon the GC confirms the teachings of Apple, in as much as it is crucial that the Commission meets the high threshold set for proof of State aid. On the other hand, in Engie the GC focuses on the nature of the substantive effects of a measure rather than on its formal scheme. However, some legal questions remain unsolved.

More specifically, as regards State origin, the Court has not proven open to discuss in much detail those reasonable objections which take into consideration regulatory competition. As a matter of fact, this phenomenon gives rise to a tangible paradox. Indeed, to fall within the scope of art. 107 TFEU, the aid should be granted through State resources. However, due to regulatory arbitrage, State resources themselves are ultimately increased by the granting of the aid. Arguably, this is confirmed by Member States’ great interest in challenging recovery orders. In this regard, it remains to be seen whether the ECJ will assign substantive weight to regulatory competition in the assessments under art. 107 TFEU.[103]

Moreover, further clarifications on the concept of “selectivity” appear desirable, given that the matter, perhaps currently the most discussed in State aid literature, has mostly been avoided in the GC’s rulings, save for the (limited) arguments forwarded in Engie.

Interesting innovations have been brought by the new effect-based approach proposed in Engie. Accordingly, the economically abusive substance of arrangements prevails over free movement considerations. Unfortunately, the absence of clear common criteria for the assessment of tax avoidance practices hinders the effectiveness of such outcome-based solutions and their potential reach. A similar criticism can be advanced for the anti-abuse mechanism introduced by Directive 2019/2121.

All in all, these two new chapters of the “tax ruling saga” bring new questions and help foster the debate about the “Vestager doctrine”. This is particularly true when the current hot political context is taken into consideration. From a strictly EU law point of view, the many doubts concerning the applicability of art. 107 TFEU to tax rulings suggest that looking for an alternative approach to tackle harmful tax competition between Member States might be more productive.

In one way or another, the whole “tax ruling saga” will certainly play an important role in the development of a fair and sustainable EU tax governance and, considering the relevance of such a matter, in the future of European integration.[104] Commentators are now awaiting the outcomes of the Commission’s ongoing investigations and the appeals brought before the ECJ in Belgium Excess Profit, Fiat, Apple, Amazon and Engie.

European Papers, Vol. 6, 2021, No 2, European Forum, Insight of 29 October 2021, pp. 1097-1116
ISSN 2499-8249 - doi: 10.15166/2499-8249/515

* LL.M. candidate, Maastricht University,

[1] Joined cases T-816/17 and T-318/18 Luxembourg v Commission ECLI:EU:T:2021:252 (hereinafter Amazon).

[2] Joined cases T-516/18 and T-525/18 Luxembourg v Commission ECLI:EU:T:2021:251(hereinafter Engie).

[3] Joined cases T-131/16 and T-263/16 Belgium v Commission (hereinafter Belgium Excess Profit) ECLI:EU:T:2019:91.

[4] Joined cases T-760/15 and T-636/16 Netherlands v Commission ECLI:EU:T:2019:669 (hereinafter Starbucks).

[5] Case T-755/15 Luxembourg v Commission ECLI:EU:T:2019:670 (hereinafter Fiat).

[6] Joined cases T-778/16 and T-892/16 Ireland v Commission ECLI:EU:T:2020:338 (hereinafter Apple).

[7] Decision 2018/859/EU of the Commission of 4 October 2017 on State aid SA.38944 implemented by Luxembourg to Amazon.

[8] Decision 2019/421/EU of the Commission of 20 June 2018 on State aid SA.44888 by Luxembourg in favour of ENGIE.

[9] Case C-454/21 Engie Global LNG Holding and Others v Commission pending.

[10] Case C-457/21 P Commission v and Others pending.

[11] A “transparent” company is a company which is not subject to corporate income taxation. Instead, its shareholders bear tax liability for the company’s income. See PJ Wattel, ‘Forum: Interaction of State Aid, Free Movement, Policy Competition and Abuse Control in Direct Tax Matters’ (2013) World Tax Journal 128, 132 ff. According to Luxembourg tax law, a limited partnership is not subject to corporate income taxation.

[12] Such a scheme provides an example of “hybrid mismatch”: a situation where differences in the tax treatment of an entity under two or more national tax laws are exploited to achieve tax advantages. See the explanation provided by the OECD at OECD, Action 2 Neutralising the Effects of Hybrid Mismatch Arrangements The mismatch in the case of Amazon is well described by the Commission in Decision 2018/859 cit., note 119: “due to LuxSCS's treatment as a fiscally transparent entity in Luxembourg, royalty payments from LuxOpCo to LuxSCS are not considered taxable income of LuxSCS in Luxembourg, but of its partners in the US. […] By contrast, since the US does not consider LuxSCS as fiscally transparent, but rather as a separate corporate entity resident in Luxembourg, the taxation of the LuxSCS's partners in the US may be deferred indefinitely, so long as none of LuxSCS's profits are repatriated to the US”.

[13] From an economic point of view, intragroup transactions are shielded from market forces, while independent companies operate under normal market conditions.

[14] An analysis of the potential application of economic, financial and statistical techniques to assess the arm’s length in the case of corporate tax arrangements is provided by J Kavanagh and N Robins, ‘Corporate Tax Arrangements Under EU State Aid Scrutiny: The Application of the Market Economy Operator Principle’ (2015) European State Aid Law Quarterly 358, 366.

[15] Decision 2018/859 cit. paras 562-564. The application of the arm’s length principle to tax ruling cases is justified by the Commission by reference to joined cases C-182/03 and C-217/03 Belgium v Commission ECLI:EU:C:2006:416. Such a reference, however, is strongly criticised in the literature. See, amongst others, T Joris and W De Cock, ‘Is Belgium and Forum 187 v. Commission a Suitable Legal Source for an EU “At Arm's Length Principle”?’ (2017) European State Aid Law Quarterly 607.

[16] See the analysis carried out by L Lovdahl Gormsen, European State Aid and Tax Rulings (Edward Elgar 2019) 36 ff.

[17] See Decision 2019/421 cit. para. 289 ff. Art. 6 Steueranpassungsgesetz was amended in 2019 in the context of the implementation of the general anti-avoidance rule (GAAR) provided by art. 6 of the ATAD Directive (Directive 2016/1164/EU of the Council of 12 July 2016 laying down rules against tax avoidance practices that directly affect the functioning of the internal market), which reproduces the GAAR introduced by Directive 2015/121/EU of the Council of 27 January 2015 amending Directive 2011/96/EU on the common system of taxation applicable in the case of parent companies and subsidiaries of different Member States.

[18] See Amazon cit. para. 113. See also: Belgium Excess Profit cit. para. 62; Fiat cit. para. 113; and Apple cit. para. 105. The principle according to which Member States must comply with EU law also in fields that fall outside the scope of EU competence is set by the ECJ in, amongst others, case C-438/05 The International Transport Workers’ Federation and The Finnish Seamen’s Union ECLI:EU:C:2007:772 para. 40: “even if, in the areas which fall outside the scope of the Community’s competence, the Member States are still free, in principle, to lay down the conditions governing the existence and exercise of the rights in question, the fact remains that, when exercising that competence, the Member States must nevertheless comply with Community law”.

[19] Amazon cit. para. 125. See also Apple cit. para. 319.

[20] Amazon cit. para. 123.

[21] Ibid.: “the mere finding of a methodological error does not in itself suffice, in principle, to demonstrate that a tax ruling conferred an advantage on a specific company and, thus, to establish that there is State aid within the meaning of Article 107 TFEU”; see also Apple cit. para. 319; and Starbucks cit. para. 201.

[22] These are the words of A Lamadrid, ‘The Fiat and Starbucks Judgements’ (25 September 2019) Chillin’Competition

[23] Decision 2018/859 cit. para. 549.

[24] Amazon cit. para. 244.

[25] Ibid. para. 250.

[26] Ibid. para. 296.

[27] Ibid. paras 294-295.

[28] For a definition of the profit split method, see the document of the EU Joint Transfer Pricing Forum, The Application of the Profit Split Method Within the EU of March 2019 JTPF/002/2019/EN,, section 2: “the PSM seeks to establish or test, in line with the arm’s length principle, an approximation of the results that independent enterprises would be expected to have achieved by engaging in transactions in comparable circumstances”.

[29] Under the profit split method, profit can be split following two methods: i) the contribution analysis and ii) the residual analysis. Following the former i), the relevant profits are allocated between the parties on the basis of a functional analysis similar to that one of the TNMM. Under the latter ii), profits are divided between the initial remuneration, for which external comparables exist on the market, and the residual unique and valuable contributions.

[30] Amazon cit. para. 501. In particular, the GC finds that “the contribution analysis consists of directly allocating the combined profits between the various parties to the transaction and does not take account of the remuneration initially calculated for LuxOpCo. Accordingly, it cannot be assumed that applying the profit split method would necessarily have led to higher remuneration for LuxOpCo” (ibid. para. 536).

[31] Ibid. para. 546.

[32] Ibid. para. 555.

[33] Ibid. para. 556.

[34] Ibid. para. 585.

[35] Ibid. para. 587.

[36] Engie cit. paras 149-153. On the same line of reasoning, see above.

[37] Joined cases C-399/10 P and C-401/10 P Bouygues and Bouygues Télécom v Commission and Others ECLI:EU:C:2013:175 para. 100, quoted in Engie cit. para. 220.

[38] Engie cit. para. 215: “In order to assess whether a measure is imputable to the State, it is necessary to examine whether the public authorities were involved in the adoption of that measure”. The Court refers to case C-405/16 P Germany v Commission ECLI:EU:C:2019:268 para. 49.

[39] Engie cit. para. 241: “it is apparent from the Court’s case-law that those two criteria may be examined together, as the ‘third condition’ laid down in Article 107(1) TFEU, relating to the existence of a ‘selective advantage’”.

[40] Ibid. para. 250.

[41] See joined cases C-106/09 P and C-107/09 P Commission and Spain v Government of Gibraltar and United Kingdom ECLI:EU:C:2011:732 para. 134: “the criteria forming the basis of assessment which are adopted by a tax system must also, in order to be capable of being recognised as conferring selective advantages, be such as to characterise the recipient undertakings, by virtue of the properties which are specific to them, as a privileged category, thus permitting such a regime to be described as favouring ‘certain’ undertakings”. In that case, the Court held that the fact that a measure is in principle applicable to all corporations cannot itself exclude (de facto) selectivity.

[42] Engie cit. paras 327 and 335.

[43] Ibid. para. 351.

[44] Art. 164 juncto art. 166 of the loi modifiée, du 4 décembre 1967, concernant l’impôt sur le revenue.

[45] Engie cit. para. 472.

[46] Joined cases C-115/16, C-118/16, C-119/16 and C-299/16 N Luxemburg 1 ECLI:EU:C:2019:134 (hereinafter Danish Beneficial Ownership Cases) para. 177. In any case, in Engie the facts of the case concerned a wholly internal situation. Hence, the lack of a cross-border element prevented art. 49 TFEU from applying (Engie cit. para. 474).

[47] On the main legal issues involved in the “tax ruling saga”, see A Arena, ‘The Commission's ATR Decisions Before the CJEU: Untangling the Legal Conundrums of a Recent Trend in Fiscal Aids’ in R Mastroianni and A Arena (eds), 60 Years of EU Competitions Law: Stocktaking and Future Prospects (Editoriale Scientifica 2017) 227 ff.

[48] Amazon cit. paras 154-155. Therefore, the only guidelines applicable to the ruling concerned were those issued by the OECD in 1995.

[49] On the application of those principles see case C-5/89 Commission v Germany ECLI:EU:C:1990:320 paras 14-16; and joined cases C-72/10 and C-77/10 Costa and Cifone ECLI:EU:C:2012:80 para. 74.

[50] European Commission, Notice on the notion of State aid as referred to in art. 107(1) of the Treaty on the Functioning of the European Union of 19 July 2016, C/2016/2946, para. 5.4.4. Significantly, tax rulings were not even mentioned in the 2014 Draft Notice. An analysis of the applicability of the principle of legitimate expectations to State aid cases is provided by JS Pastoriza, ‘The Recovery Obligation and the Protection of Legitimate Expectations: The Spanish Experience’ in I Richelle, W Schön and E Traversa (eds), State Aid Law and Business Taxation (Springer 2016) 247.

[51] A Giraud and S Petit, ‘Tax Rulings and State Aid Qualification: Should Reality Matter?’ (2017) European State Aid Law Quarterly 233, 235. The issue of State origin is also mentioned by J Kavanagh and N Robins, ‘Corporate Tax Arrangements Under EU State Aid Scrutiny’ cit. 359.

[52] “Regulatory competition” is described as “a process whereby legal rules are selected (and de-selected) through competition between decentralised, rule-making entities” by C Barnard and S Deakin, ‘Market Access and Regulatory Competition’ (Jean Monnet Working Paper 9-2001) 4. The notion was first developed by CM Tiebout, ‘A Pure Theory of Local Public Expenditures’ (1956) Journal of Political Economy 416 ff. See also K Gödker and L Hornuf, ‘Regulatory Competition’ in A Marciano, GB Ramello (eds), Encyclopedia of Law and Economics (Springer 2019) 1787-1788: according to the authors, “regulatory competition describes the activity of private or public lawmakers who intend to produce novel or alter current legislation in response to competitive pressure from other private or public lawmakers”, whereas regulatory arbitrage is “a legal planning technique that is carried out to avoid taxes and other legal rules to circumvent regulatory costs”. A general overview of the phenomenon of regulatory arbitrage is provided by K Pistor, The Code of Capital. How the Law Creates Wealth and Inequality (Princeton University Press 2019) 67 ff. For an analysis of the various potential effects of regulatory competition, see also D Murphy, ‘Interjurisdictional Competition and Regulatory Advantage’ (2005) JIEL 891 ff.

[53] See A Giraud and S Petit, ‘Tax Rulings and State Aid Qualification’ cit., where the authors submit that “Ireland’s tax authorities negotiated Apple’s taxable basis precisely because Apple had the possibility of shifting the corresponding revenues elsewhere if a satisfying compromise was not found”. P Nicolaides, ‘State Aid and Tax Rulings’ (2016) European State Aid Law Quarterly 416, 426, seems to agree: “[…] if the comparison is between the amount of tax that is paid without the tax ruling and the amount of tax that is paid with the tax ruling, then group companies do not receive a selective advantage, but a selective disadvantage. They pay more than what they would have paid without the tax rulings”. See also M Orlandi, ‘Interpelli (tax ruling), accordi preventivi sui prezzi di trasferimento, principio di libera concorrenza ed aiuti di Stato: la nuova frontiera della disciplina della concorrenza’ in P Boria (ed.), La Concorrenza Fiscale tra Stati (CEDAM 2019) 123, 146, where the author observes that in this sense the logic behind tax rulings is similar to the one behind settlement agreements negotiated between private citizens and tax authorities.

[54] Fiat cit. para. 415. See U.S. Department of the Treasury, ‘The European Commission’s Recent State Aid Investigations of Transfer Pricing Rulings’ (2016) This whitepaper was issued in response to the Commission’s decision in Apple case (Decision 2017/1283/EU of 30 August 2016 on State aid SA.38373 implemented by Ireland to Apple).

[55] Regulation (EU) 2015/1589 of the Council of 13 July 2015 laying down detailed rules for the application of art. 108 of the Treaty on the Functioning of the European Union.

[56] Fiat cit. para. 415.

[57] Engie cit. paras 212-223.

[58] See, amongst the others, L Lovdahl Gormsen, European State Aid and Tax Rulings cit., A Giraud and S Petit, ‘Tax Rulings and State Aid Qualification’ cit.; P Nicolaides, ‘State aid and Tax Rulings’ cit.; M Orlandi, ‘Interpelli (tax ruling), accordi preventivi sui prezzi di trasferimento, principio di libera concorrenza ed aiuti di Stato’ cit.; N Bayón Fernández, ‘The Selective Advantage Criterion in Tax Rulings: The Path Towards a More Coherent and Thorough Analysis of Selectivity’ (2021) Journal of European Competition Law and Practice 200.

[59] Case C-15/14 P Commission v MOL ECLI:EU:C:2015:362.

[60] Ibid. para. 60.

[61] In particular, see L Lovdahl Gormsen, European State Aid and Tax Rulings cit. 53, where the author observes that: first, in Commission v MOL the Court distinguishes between individual aid and general schemes, observing that the presumption is only applicable to the former. However, both in Commission v MOL and in the sole subsequent judgement in which that statement has been so far reiterated, the facts concerned general schemes. Hence, that presumption has not yet been applied by the Court to an individual aid case. Second, there is no express ground in the judgement for exempting the Commission from the duty of applying the three-steps test (see ‘Notice on the notion of State aid’ cit. para. 128). Third, the Commission v MOL case was related to a mining fee and not to fiscal measures.

[62] See Fiat cit. paras 121 and 332; and Starbucks cit. para. 129: “it must be noted that the Commission’s approach of examining the criteria of advantage and selectivity concurrently is not itself incorrect, given that both the advantage and the selective nature of that advantage are examined. Nevertheless, the Court considers it appropriate to consider, first of all, whether the Commission was entitled to conclude that there was an advantage, before going on, if necessary, to examine whether that advantage had to be considered to be selective”.

[63] Such an approach is also described as “outcome-based” by P Nicolaides, ‘Can Selectivity Result from the Application of Non-Selective Rules? The Case of Engie’ (2019) European State aid Law Quarterly 15 ff. For the sake of completeness, it must be observed that also in Amazon the GC highlights that “Article 107(1) TFEU defines a measure which reduces the burdens normally imposed on an undertaking in relation to its effects” (Amazon cit. para. 124).

[64] That relationship is described in F Costamagna, ‘At the Roots of Regulatory Competition in the EU: Cross-border Movement of Companies as a Way to Exercise a Genuine Economic Activity or just Law Shopping?’ (2019) European Papers 185. Furthermore, a complete overview of the evolution of the relationship between the abuse of law and the Court’s case law is provided by A Ballancin and F Cannas, ‘The Development of the Doctrine of Abuse of Law and the Danish Cases: Time to Shift the Focus from Non-Genuine Arrangements to Single (Abusive) Transactions?’ (2021) EC Tax Review 70 ff. See also P Van Cleynenbreugel, ‘Regulating Tax Competition in the Internal Market: Is the European Commission Finally Changing Course?’ (2019) European Papers 225. Against the need for a specific doctrine of abuse in the case of cross-border mobility see M Gelter and L Vicente, ‘Abuse of Companies Through Choice of Incorporation?’ (ECGI – Law Working Paper 473-2019)

[65] See case C-209/10 Post Danmark ECLI:EU:C:2012:172 paras 41-42; case C-23/14 Post Danmark ECLI:EU:C:2015:651 paras 48-49; and case C-413/14 P Intel v Commission ECLI:EU:C:2017:632. For a complete overview on the matter, see A Jones, B Sufrin, and N Dunne, EU Competition Law. Text, Cases, and Materials (Oxford University Press 2019) 375 ff.

[66] The adjective is borrowed from F. Vanistendael, ‘From Abuse to Base Erosion, How Did it Come to This?’ in W Haslenher, K Pantazatou, G Kofler, and A Rust (eds), A Guide to Anti-Tax Avoidance Directive (Edward Elgar 2020) 15.

[67] Case C-196/04 Cadbury Schweppes and Cadbury Schweppes Overseas ECLI:EU:C:2006:544. In this landmark case, the ECJ referred to “premises, staff and equipment” (para. 67) as the objective factors thanks to which the economic activity of an undertaking within a certain Member State can be regarded as “genuine”, entailing that no abuse could be found.

[68] BEPS is the acronym for Base Erosion and Profit Shifting.

[69] The shift appears sufficiently clear in Danish Beneficial Ownership Cases cit. paras 90-91, 93 and 103.

[70] Ibid. para. 108.

[71] Amazon cit. para. 124.

[72] Directive 2019/2121/EU of the European Parliament and of the Council of 27 November 2019 amending Directive (EU) 2017/1132 as regards cross-border conversions, mergers and divisions.

[73] A criticism to the new mechanism of pre-operation certificates is advanced by S Binard and L Schummer, ‘The Case for Further Flexibility in Matters of Cross-Border Corporate Mobility’ (2019) European Company Law 31 ff. Further critical comments are made by M Winner, ‘Protection of Creditors and Minority Shareholders in Cross-border Transactions’ (2019) European Company and Financial Law Review 44 ff.

[74] Directive 2019/2121 cit. para. 33. Art. 1 introduces new arts 86m, 127, and 160m which deal with pre-conversion, pre-merger and pre-division certificates respectively.

[75] Directive 2017/1132 of the European Parliament and of the Council of June 2017 relating to certain aspects of company law, art. 86m(8) as amended by Directive 2019/2121.

[76] Ibid. art. 3.

[77] Even after the adoption of Directive 2019/2121, the need for further harmonisation of company law, especially in the context of cross-border mobility, is expressed by C Gerner-Beuerle, FM Mucciarelli, E Schuster and M Siems, ‘Making the Case for a Rome V Regulation on the Law Applicable to Companies’ (ECGI – Law Working Paper 549-2020)

[78] The term is borrowed from S Garben, ‘Confronting the Competence Conundrum: Democratising the European Union through an Expansion of its Legislative Powers’ (2015) OJLS 55, 63.

[79] P Nicolaides, ‘Can Selectivity Result from the Application of Non-Selective Rules?’ cit. 27.

[80] See the general anti-avoidance rules introduced by Directive 2015/121/EU of the Council of 27 January 2015 amending Directive 2011/96/EU on the common system of taxation applicable in the case of parent companies and subsidiaries of different Member States, art. 1 and Directive 2016/1164/EU of the Council of 12 July 2016 laying down rules against tax avoidance practices that directly affect the functioning of the internal market, art. 6.

[81] Several reasons are listed in E Traversa and A Flamini, ‘Fighting Harmful Tax Competition through EU State Aid Law: Will the Hardening of Soft Law Suffice?’ (2015) European State Aid Law Quarterly 323, 326, where the authors report the opinion of those according to whom “State aid provisions are not a suitable instrument for tackling harmful tax competition, not only because of the wording of article 107 TFEU, but also because of the nature of State aid control: namely, it is based on a case-by-case approach, it is limited to the territory of one Member State and it does not take into consideration other Member States’ practices”. A similar opinion is expressed by PJ Wattel, ‘Comparing Criteria: State Aid, Free Movement, Harmful Tax Competition and Market Distorting Disparities’ in I Richelle, W Schön and E Traversa (eds), State Aid Law and Business Taxation cit. 59, 69. See also R Mason and S Daly, ‘State Aid: The General Court Decision in Apple’ (2020) Tax Notes International 1329, where the authors hold that “[t]he commission cannot solve base erosion and profit shifting by issuing state aid recovery orders against specific U.S. companies. Systemic problems call for comprehensive solutions”; P Nicolaides, ‘Can Selectivity Result from the Application of Non-Selective Rules? The Case of Engie’ cit. 28; E Forrester, ‘Is the State Aid Regime a Suitable Instrument to Be Used in the Fight Against Harmful Tax Competition?’ (2018) EC Tax Review 19 ff; L Lovdahl Gormsen, ‘EU State Aid Law and Transfer Pricing: A Critical Introduction to a New Saga’ (2016) Journal of European Competition Law & Practice 369 ff; and G Perotto, ‘How to Cope with Harmful Tax Competition in the EU Legal Order: Going Beyond the Elusive Quest for a Definition and the Misplaced Reliance on State Aid Law’ (2021) European Journal of Legal Studies 309 ff.

[82] Commission and Spain v Government of Gibraltar and United Kingdom ECLI:EU:C:2011:215, opinion of AG Jääskinen para. 134.

[83] On 19 September 2018, the Commission issued a decision in case SA.38945, finding that the non-taxation of certain McDonald's profits in Luxembourg did not lead to illegal State aid.

[84] On the American criticism to the Commission’s approach and its consequences, also mentioned by Luxembourg’s pleas in Fiat, see the abovementioned U.S. Department of the Treasury’s whitepaper. For an overview on the main American criticisms to the “Vestager doctrine” see also K Richard, ‘Are All Tax Rulings State Aid: Examining the European Commission's Recent State Aid Decisions’ (2018) Houston Business and Tax Law Journal 1. The fortuitous trend of the “tax ruling saga” was already noticed by A Lamadrid, ‘The Fiat and Starbucks Judgements’ cit.: “politically, the credibility of the Court and of the EU system may be reinforced by the fact that – fortuitously – it was only the US company (and the Netherlands) getting out of the hook, not the European one (and Luxembourg)”; and PJ Wattel, ‘The Cat and the Pigeons: Some General Comments on (TP) Tax Rulings and State Aid After the Starbucks and Fiat Decisions’ in I Richelle, W Schön and E Traversa (eds), State Aid Law and Business Taxation cit. 185, 193: “the Commission should show […] that it does not disproportionally target US groups […]”.

[85] Statement by Executive Vice-President Margrethe Vestager following today's Court judgments on the Amazon and Engie tax State aid cases in Luxembourg, in European Commission Statement 21/2468 of 12 May 2021

[86] Communication COM(2021) 251 final from the European Commission to the European Parliament and the Council of 18 May 2021, Business Taxation for the 21st Century.

[87] Communication COM(2020) 313 final from the Commission to the European Parliament and the Council of 15 July 2020 on Tax Good Governance in the EU and beyond.

[88] Communication COM(2020) 312 final from the Commission to the European Parliament and the Council of 15 July 2020, An Action Plan for Fair and Simple Taxation Supporting the Recovery Strategy.

[89] Statement of the Members of the European Council SN 18/21 of 25 March 2021,

[90] On 5 April 2021 United States Secretary of the Treasury Janet Yellen addressed a speech to the Chicago Council on Global Affairs in which she called for a global minimum tax on companies, in order to end a “30-year race to the bottom”. See Editorial, ‘Janet Yellen calls for a global minimum tax on companies. Could it happen?’ (8 April 2021) The Economist See also Commission Staff Working Document, Annual Report on Taxation 2021 (Part 3), 20 May 2021, SWD(2021) 109 final 115: “the Commission actively participates in international discussions taking place in the Organisation for Economic Co-operation and Development concerning the reform of the corporate taxation framework, including a revision of the right to tax notably for the digital economy and the implementation of a minimum effective taxation of multinational enterprise’s profits”.

[91] See A Williams, J Politi and M Khan, ‘US Proposes Global Corporate Tax Rate of at Least 15% in International Talks’ (21 May 2021) Financial Times; M Valsania, ‘Yellen propone una corporate minimum tax globale del 15%. Sì da Italia, Francia e Germania’ (21 May 2021) Il Sole 24 Ore, and Editorial, ‘G7, accordo “storico” su tassa minima globale al 15% per le grandi imprese. Franco: ora proposta su tavolo G20. Draghi: è passo storico’ (5 June 2021) Il Sole 24 Ore

[92] See G Dangor, ‘G20 Signs Off On 15% Global Minimum Corporate Tax – Here’s How It Will Work’ (11 July 2021) Forbes

[93] See C Giles, ‘World’s Leading Economies Agree Global Minimum Corporate Tax Rate’ (1 July 2021) Financial Times

[94] On how the “tax ruling saga”, which “sits between competition law and tax law”, shows the broadness and global dominance of EU competition law see A Bradford, The Brussels Effect. How the European Union Rules the World (Oxford University Press 2020) 107.

[95] State aid: Commission opens in-depth investigation into tax treatment of Nike in the Netherlands, Press Release IP/19/322 of the European Commission of 10 January 2019. On 14 July 2021, the GC issued its judgement in case T-648/19 Nike European Operations Netherlands and Converse Netherlands v Commission ECLI:EU:T:2021:428, dismissing the action brought by Nike in order to obtain the annulment of the opening decision of the Commission (Decision C(2019) 6 final of the Commission of 10 January 2019 on State aid SA.51284 (2018/NN)).

[96] State aid: Commission opens in-depth investigation into the Netherlands' tax treatment of Inter IKEA, Press Release IP/17/5343 of the European Commission of 18 December 2018.

[97] State aid: Commission opens in-depth investigation into tax treatment of Huhtamäki in Luxembourg, Press Release IP/19/1591 of the European Commission of 7 March 2019.

[98] State aid: Commission opens in-depth investigations into individual “excess profit” tax rulings granted by Belgium to 39 multinational companies, Press Release IP/19/5578 of the European Commission of 16 September 2019.

[99] Belgium Excess Profit cit.

[100] See P Tang and H Bussink, ‘EU Tax Revenue Loss from Google and Facebook’ (PvdA and S&D 2017) 4.

[101] Apple cit. para. 479; see also the press release 90/20 of the General Court of the European Union of 15 July 2020 ‘The General Court of the European Union annuls the decision taken by the Commission regarding the Irish tax rulings in favour of Apple’: “although the General Court regrets the incomplete and occasionally inconsistent nature of the contested tax rulings, the defects identified by the Commission are not, in themselves, sufficient to prove the existence of an advantage for the purposes of Article 107(1) TFEU”. In particular, the Court seems to criticise the “regrettable methodological defect” constituted by the “lack of evidence submitted to the Irish tax authorities concerning the functions actually performed by the Irish branches and the assessment of those functions for the purpose of determining the profit to be allocated to those branches” (Apple cit. paras 347-348).

[102] Amazon cit. para. 585.

[103] See A Giraud and S Petit, ‘Tax Rulings and State Aid Qualification’ cit.

[104] On the relevance of EU tax governance for the process of integration, see in particular PJ Wattel, ‘Taxation in the Internal Market’ in P Koutrakos and J Snell (eds), Research Handbook on the EU’s Internal Market (Edward Elgar 2017) 319 ff.