State Aid Enforcement After Amazon
Ruth Mason (ruth.mason@law. virginia.edu) is the Edwin S. Cohen Distinguished Professor of Law and Taxation at the University of Virginia School of Law. She thanks Philip Baker for comments on an earlier draft.
In this article, the author summarizes the General Court of the European Union’s judgment in Amazon and analyzes what it means for the future of state aid.
- The Dispute
- State Aid Basics
- The ‘Normal’ Base and OECD Guidelines
- Disputation of the Commission’s Pricing Methods
- What’s Important About Amazon
- Taking Stock of Tax State Aid
On May 12 Amazon and Luxembourg scored a major victory in their long-running dispute with the European Commission. The General Court of the European Union (GCEU) vacated the commission’s state aid decision (C(2017) 6740 final), holding that the commission had not met its burden to prove that Luxembourg conveyed a tax advantage to Amazon via its transfer pricing ruling for the company.1 Appealable to the EU’s highest court, Amazon has dealt yet another blow to the commission’s state aid strategy in tax cases.
Briefly, in Amazon, the GCEU confirmed the approach it took in Fiat2 and Apple3:
the commission may use the contemporaneous OECD arm’s-length standard to “check” a member state’s application of the arm’s-length standard;
minor or even major methodological errors (including not using the best transfer pricing method) by a member state do not constitute state aid, unless the commission can show they lead to a reduction in tax; and
the commission bears the heavy burden of proving that the state reduced the company’s tax.
The Amazon structure was the same one at issue in the recent U.S. Tax Court decision affirmed by the U.S. Court of Appeals for the Ninth Circuit,4 but rather than focusing on the income split between the U.S. and Luxembourg parts of the Amazon group, the state aid case focused on the split within the Luxembourg entities.
LuxSCS, Amazon’s intangibles holding company in Europe, was regarded as a passthrough in Luxembourg (but not the United States). That company, which had no employees, was the same entity that entered the cost-sharing agreement (CSA) that was the subject of the U.S. case. LuxSCS received royalties from its subsidiary, LuxOpCo, Amazon’s EU headquarters, and LuxOpCo held Amazon’s EU operating companies.
At issue was the income of LuxOpCo. The commission said it should be higher; Luxembourg and Amazon said LuxOpCo received an arm’s-length allocation,5 which was all that was required under Luxembourg and EU law. Throughout the period under dispute, LuxOpCo had declared its profits consistently with a transfer pricing ruling that Luxembourg issued in 2003.
State Aid Basics
The prohibition of state aid prevents EU states from subsidizing particular companies or types of companies (including multinationals) in a way that distorts private competition in the EU.6 The standard is not completely clear,7 but the basic idea is that if a state permits a company to pay less tax than would have been due under the state’s “normal” tax system, and that tax reduction isn’t justified, then it is illegal state aid, and the commission can order the state to recover the tax reduction from the company, with interest.
The ‘Normal’ Base and OECD Guidelines
The commission’s approach to state aid requires it to identify the state’s “normal” tax base so that it can measure the special relief as a deviation from that base. Several of the state aid transfer pricing cases have involved major disputes over the normal base, but in Amazon, the parties agreed that the appropriate baseline to judge whether Luxembourg illegally subsidized Amazon consisted of the arm’s-length income-allocation standard for multinationals because that was the standard under Luxembourg law.8
As in Fiat and Apple, the GCEU in Amazon described the commission as using the arm’s-length standard to check that Luxembourg properly approved the proposed transfer pricing method, as well as the annual income declarations that Amazon made under that method. Citing the Forum 187 decision from the Court of Justice of the European Union,9 the GCEU once again stated that if member states do not “make a distinction between integrated undertakings and stand-alone undertakings for the purposes of their liability to corporate income tax,” then use of the arm’s-length standard as a check on the accuracy of the state’s income-allocation determinations is permissible under the state aid rules. That reasoning relies on the conclusion that the absence of that kind of distinction means the “law is intended to tax the profit arising from the economic activity of such an integrated undertaking as though it had arisen from transactions carried out at market prices.”10
To be clear, all the parties agreed that the arm’s-length standard was relevant in Amazon. Thus, Amazon differs from Apple because in Apple Ireland did not agree that its domestic law required the arm’s-length standard. It is unclear from the cases so far what happens when a state does draw a distinction between multinationals and stand-alone companies.11
Despite agreeing that arm’s length was the right standard, the parties still disagreed over specifics. Amazon and Luxembourg objected to the commission’s use of OECD guidelines (from 2010 and 2017) that post-dated the issuance of the challenged ruling. They complained that the commission applied the development, enhancement, maintenance, protection, and exploitation (DEMPE) functions to a ruling dating back to 1995 and to income declarations that predated the adoption of that approach. As it had in Fiat and Apple, the GCEU confirmed that it was appropriate for the commission to use the OECD guidelines to check a member state’s transfer pricing rulings but held that the commission could not retroactively apply that guidance, at least when the guidance propounded new standards or approaches, rather than clarifying preexisting approaches. Because the DEMPE standards were not merely clarifying, the commission could not use them.
Disputation of the Commission’s Pricing Methods
Most of the GCEU’s 80-page decision focuses on the transfer pricing methods the commission used to conclude that the tax ruling (and the income declared by LuxOpCo consistently with that ruling) conveyed an illegal advantage because it allowed LuxOpCo to pay too little tax. In the commission’s view, LuxSCS should have had more income, and LuxOpCo less. Luxembourg had approved a ruling for Amazon based on the transactional net margin method (TNMM), with LuxOpCo as the less complex tested party. LuxOpCo received a cost markup based on what Amazon and Luxembourg argued were routine functions. LuxSCS ended up with a large residual (amounting to about 80 percent of the combined profits of the two companies) under that one-sided transfer pricing method, and because LuxSCS was not a separately taxable entity in Luxembourg, it did not pay tax on the residual there (because it was a hybrid, the income also did not flow through to its U.S. parent).
The case recites much detail about the commission’s various transfer pricing arguments. But the major lines of factual division were clear: Amazon and Luxembourg argued that the profits resulted from technology and other intellectual property invented in the United States and owned by LuxSCS as a result of its CSA with its U.S. parent. In contrast, the commission argued that LuxOpCo’s efforts contributed more to the company’s overall profits than its intragroup remuneration reflected. Thus, argued the commission, in approving both the initial ruling and Amazon’s annual declarations of income consistent with the ruling, Luxembourg “departed from a reliable approximation of a market-based outcome” in violation of the state aid rules.
TNMM and a New External Standard
The commission made several alternative arguments: first, that Luxembourg should have used the TNMM with LuxSCS (not LuxOpCo) as the tested party and that the remuneration for LuxSCS should have been based on a markup on costs, which would have put the residual in LuxOpCo, where it would have been taxed by Luxembourg. The commission claimed that it could rely on a 5 percent markup, a figure it got from a 2010 report of the Joint Transfer Pricing Forum (JTPF) claiming that 5 percent was the most common markup for “low-value-adding intra-group services.” Because, in the commission’s view, LuxSCS performed only low-value-adding functions, the 5 percent markup was appropriate. The commission further said the markup should apply only to costs other than those LuxSCS incurred under the buy-in and CSA with the United States. The buy-in and CSA costs should have received no markup at all, in the commission’s view.
Relying heavily on OECD guidance, the GCEU held that the commission was wrong to conclude that between LuxSCS and LuxOpCo, LuxSCS ought to be the tested party under the TNMM. The commission’s reasoning was based on the notion that all of LuxSCS’s activities were passive because it merely held intangibles. Drawing on contemporaneous OECD transfer pricing guidelines, however, the GCEU held that the choice of the tested party was not governed primarily by complexity but rather by the availability of reliable data for comparables, and that although the party whose reliable data is most readily available is typically the less complex party, it need not be. Because LuxOpCo performed more routine functions, whereas LuxSCS held unique intangibles, the court concluded that it was not wrong for Luxembourg to conclude that more reliable comparables would be available for LuxOpCo than for LuxSCS, so LuxOpCo could properly be treated as the tested party. Under that view, the residual would properly go to LuxSCS, not LuxOpCo.
More broadly, the GCEU rejected the commission’s claim that LuxSCS did not perform active functions; it disagreed with the commission’s conclusion that owning, improving (through the CSA with the U.S. parent), and making available for use by LuxOpCo the Amazon intangibles was a passive and noncritical activity. The GCEU also cited the 1995 OECD transfer pricing guidelines that recommended that the party owning unique and valuable assets should not be used as the tested party because reliable comparables would not be available to calculate its profit.
For good measure, the GCEU also rejected other arguments made by the commission in relation to its argument that LuxSCS should have been the tested party, including the notion that LuxSCS did not and could not bear the relevant intangibles risks. The court also rejected the commission’s argument that LuxSCS should earn no markup on its cost-sharing and buy-in payments to the United States. In addition to concluding that a markup of zero would not be negotiated at arm’s length, the GCEU admonished the commission for raising new arguments that it had not made in its original administrative decision — namely, that the arrangement was abusive because LuxSCS was a hybrid or a fictitious company. Finally, for the other costs, on which the commission was willing to permit a markup to LuxSCS, the GCEU rejected the commission’s use of the 5 percent markup derived from the JTPF report.
Because the GCEU rejected the commission’s transfer pricing analysis, it rejected the commission’s assertion that any downward deviation from that analysis constituted a tax advantage amounting to state aid.
The commission had a second argument ready once the court rejected the idea that the LuxOpCo should have received the residual under the TNMM. It argued that because both LuxSCS and LuxOpCo performed important functions, the right transfer pricing method was a profit split based on their relative contributions. The commission did not try to estimate those relative contributions, instead asserting that such a method would have inevitably led to higher profits in LuxOpCo. The court rejected that assertion on the grounds that the commission has the burden to prove the advantage, and although it need not do so with extreme specificity, it cannot do so by assumption.
Although it was superfluous, given the commission’s failure to carry its burden of proof, the GCEU reviewed the commission’s argument that LuxOpCo performed unique and valuable functions. It concluded that the commission overvalued the human contributions of the 60 employees in LuxOpCo and undervalued the human contributions in the United States and the importance of technology and IP obtained by LuxSCS through its CSA with its U.S. parent. But the GCEU conceded that LuxOpCo did make both unique and valuable, as well as valuable but not unique, contributions to Amazon’s profits, including adaptation of technologies for Europe, translation, pricing, inventory, marketing, and accumulation of customer data. The fatal problem with the commission’s analysis, however, was that it failed to prove that a profit split based on the (properly construed) relative contributions of LuxOpCo and LuxSCS would have resulted in more income in the operating company and less in the intangibles holding company.12
The commission reasoned that because the challenged transfer pricing ruling assumed that LuxOpCo performed only routine management functions when it actually did more than that, the allocation of income to LuxOpCo under the ruling must necessarily be too low, even though the commission did not try to calculate the difference between the two methods.
Although the GCEU agreed that “there is a difference between proving the advantage and quantifying it,” it still said the commission could not simply assume that there would be more income for one party under an alternative analytical method. The commission had to show its work, although the court allowed that that did not mean that it “must necessarily carry out a new analysis with the same level of detail as that carried out by the Member State.” The court then announced a vague standard:
[The commission] must identify a number of specific factors from which it can be concluded with certainty that the arm’s length remuneration for the company’s functions, as identified by the Commission, was necessarily greater than the remuneration received pursuant to the tax measure in question.
The GCEU concluded that the commission failed to meet that standard, demonstrating at most “the probability of the existence of an advantage.”
The Ruling Used the Wrong PLI
In its third argument, which came into play because the GCEU rejected the commission’s first two theories of the case, the commission said the ruling used the wrong profit-level indicator (PLI) to determine LuxOpCo’s income — it used a markup on operating expenses but should have used all expenses.13 The GCEU rejected that argument because the commission failed to connect the use of the supposedly wrong PLI with any tax savings. For its part, Amazon claimed that use of total costs would not have changed anything because operating costs were the largest component of all costs.14
The Ruling Should Not Have Had a Ceiling
The original transfer pricing ruling set a ceiling on the royalties that LuxOpCo would pay LuxSCS, which the commission argued was unsupported by the transfer pricing guidelines or arm’s-length reasoning. The GCEU agreed, but once again held that the commission failed to show that the ceiling produced a tax advantage for LuxOpCo. The inclusion of a ceiling was, in the court’s terms, “a methodological error,” but the commission never proved its materiality: The profits declared consistently with the ruling (and ceiling) could still be within the arm’s-length standard, and if they were not, it was up to the commission to prove it.
What’s Important About Amazon
Amazon breaks little new ground. It includes an important, but not novel, holding regarding the burden of proof in state aid cases. The GCEU held that the commission could not retroactively apply OECD guidance unless subsequent guidance was merely clarifying. In dictum, the court confirmed the permissibility of the JTPF as a new standard-setter in tax state aid cases.
Retroactive Use of OECD Guidance
As in Fiat, the GCEU in Amazon confirmed that the commission could use OECD guidance to check the member state’s application of its own domestic-law arm’s-length standard.15 But in Amazon, the GCEU rejected the notion that the commission could apply OECD guidance retroactively, at least when the new guidance went beyond merely clarifying old rules or guidance. Thus, the commission could not use the 2017 DEMPE analysis in a case dating back to the early 2000s. However, the GCEU’s analysis did not expressly rule out using DEMPE (or other modern guidance that goes beyond a mere clarification) to review the application in any given year of an old ruling to current facts.
The Burden of Proof
As with Fiat and Apple, the GCEU made clear that the commission’s real failure in Amazon was on burden of proof. It reiterated that proving that a state applied the arm’s-length standard incorrectly would not by itself be sufficient to show that the state thereby conveyed a tax advantage. Thus, according to the court, “it cannot be concluded that mere non-compliance with methodological requirements necessarily leads to a reduction in the tax burden.”
The court also spoke of a margin of appreciation: The state does not have to be perfect in applying the arm’s-length standard; it just has to ensure “a reliable approximation of a market-based outcome” that “does not go beyond the inaccuracies inherent in the methodology.”16
Relatedly, the commission must do its legwork — it cannot simply assume the conclusion of various alternative transfer pricing methods that it claims the member state should have used. While the court noted that the commission is not required to conduct a “new analysis with the same level of detail as that carried out by the Member State,” it still must “identify a number of specific factors” that show that the enterprise should have paid more tax. Unfortunately, the court gave no indication about what those factors might be.
Philip Baker has asked whether the commission will begin to solicit expert transfer pricing reports and, if so, whether state aid cases will start to resemble domestic transfer pricing cases in that they are characterized by dueling experts.17
Joint Transfer Pricing Forum
The commission’s primary argument was that Luxembourg should have required Amazon to use the TNMM with LuxSCS, the hybrid company that entered the CSA with the U.S. parent and owned the IP, as the tested party. That would have placed the residual with LuxOpCo, a company taxable in Luxembourg.
With LuxSCS as the tested party, the commission needed to calculate its remuneration. As mentioned, it suggested a markup on a subset of costs based on a 2010 JTPF report that low-value-adding intragroup services were typically remunerated in the range of 3 to 10 percent, with the most common markup being 5 percent. The commission therefore applied a 5 percent markup to the subset of LuxSCS that it decided deserved to be remunerated.
Luxembourg objected to reliance on the JTPF report, arguing that at most it could be used as a safe harbor. The GCEU concluded that although the commission had used the report inappropriately in determining a markup due to LuxSCS (because LuxSCS was not providing low-value-adding services), in principle, the commission could rely on the report rather than carry out “its own comparability analysis and its own examination of the comparable net margins on the market,” if its underlying data were relevant and reliable. Although the court was not explicit, its decision seemed to depend on the fact that the report was published.
Taking Stock of Tax State Aid
Although Amazon does not break much new ground, it helps us take stock of the tax state aid saga. Several years ago, in the heat of the financial crisis, the U.S. Congress and U.K. Parliament called on executives of some of the world’s largest multinationals to testify about their tax avoidance strategies.18 That testimony — especially about negotiated rates and confidential tax rulings with EU states — drew the scrutiny of regulators in the European Commission. They opened investigations against several states for alleged state aid to companies, nearly all of which were — in those early days — U.S.-headquartered.
The commission was aggressive; it sought to apply what I described in these pages as a sui generis arm’s-length standard in transfer pricing cases.19 The commission claimed that its sui generis standard applied regardless of domestic law, and although the sui generis standard was largely consonant with the OECD arm’s-length standard, compliance with the OECD standard was not a complete shield from state aid scrutiny.20 That application of novel theories, combined with the apparent discriminatory targeting of U.S. companies, prompted respected academic Michael Graetz to question the EU’s commitment to the rule of law.21
But critics of the commission’s state aid cases should draw some relief from the judicial resolution of those cases. The GCEU has disciplined the commission’s more extreme and unprincipled claims. Here are the outcomes so far of the cases involving U.S. (or U.S.-intensive) companies:
Amazon — the GCEU annulled the commission’s finding of aid, which the commission may appeal to the CJEU;
Apple — the GCEU annulled the commission’s finding of aid, which the commission appealed to the CJEU;
Fiat — the GCEU upheld the commission’s finding of aid, which Fiat and Ireland appealed to the CJEU;22
McDonald’s — the commission found no aid;
Nike — the commission found aid in its opening decision, which Nike appealed to the GCEU; and
Starbucks — the GCEU annulled the commission’s finding of aid, which the commission did not appeal.
In annulling several of the commission’s decisions — and even in Fiat, in which it upheld the commission’s decision — the GCEU did not sugarcoat its criticism of the commission.23
Amazon is no different. The court criticized the commission for raising novel arguments at bar and for retroactively applying modern OECD guidance. It observed that the commission’s primary theory of the Amazon case (involving TNMM) was “vitiated by numerous errors” so that its calculation “cannot be regarded as sufficiently reliable or capable of achieving an arm’s length outcome.” The court’s criticism of the commission’s alternative profit-split theory was even harsher: The commission made “mere unverified assumptions” regarding the outcome of a profit-split analysis without providing verifiable evidence, an approach the court found theoretical and insufficient to determine even the order of magnitude of the remuneration that LuxOpCo should have received under the commission’s proposed method.
The GCEU thus appears to be holding the commission to EU law, but we will have to see what happens with the cases when their appeals reach the CJEU.24
Several important questions remain after Amazon. First, unless the commission abandons it, we will have to wait for the Apple appeal in the CJEU to learn the full fate of the commission’s misguided claim that it can apply a sui generis arm’s-length standard in state aid cases.25
Second, for those, like me, who have spent the last several years wondering if the goal of state aid enforcement is only to promote level competition among private enterprises, or to both level competition and reduce tax competition, the Amazon decision gives no answer. Although it noted that the commission made both claims, the GCEU never reached the distortion-of-competition element of state aid analysis because it disposed of the case at an earlier step by concluding that the commission had not shown that Luxembourg conveyed a tax advantage to Amazon. Thus, one of the most important questions raised by the prohibition of tax state aid — what, precisely, is it for? — has never received a clear answer from the commission or EU courts.
My comments should not be taken as an argument that tax rulings cannot convey state aid or even that the commission should abandon state aid review of tax rulings. On the contrary, because of their obscurity, the complexity of the laws they apply, their confidentiality, and their application to only a single taxpayer, tax rulings represent an ideal mechanism for governments to deliver benefits to a favored taxpayer while denying similar treatment to the taxpayer’s competitors. When those rulings are secret and unilateral — and all the recent state aid cases involved secret, unilateral rulings — states can use them to impose different, and more favorable, tax rules than those available under domestic law.
But it is crucial for legal certainty and the rule of law for the commission and EU courts to propound clear state aid standards. One that seems to be emerging from the GCEU cases is that when states adopt the arm’s-length principle into their domestic laws, adherence to contemporaneous OECD transfer pricing guidelines in granting rulings and accepting income declarations will serve as a safe harbor.
Moreover, member states have been given some room for error. Although I have espoused more ambitious academic claims about how to interpret the prohibition of state aid in the tax area,26 the emerging standard is a reasonable one.
1 Luxembourg v. Commission and Amazon EU Sàrl and Amazon.com Inc. v. Commission, joined cases T-816/17 and T-318/18 (GCEU 2021).
2 Luxembourg v. Commission, joined cases T-755/15 and T-759/15 (GCEU 2019).
3 Apple Sales International and Apple Operations Europe v. Commission, joined cases T-778/16 and T-892/16 (GCEU 2020).
4 Amazon.com v. Commissioner, 934 F.3d 976 (9th Cir. 2019), aff’g 148 T.C. 108 (2017) (holding that under the 1995 regulations, cost-sharing agreements did not require a buy-in payment for residual business assets, such as workforce in place and going concern, because they did not fall under the then-definition of intangibles in the regs and code). The regulations have since been amended to make it clear that buy-in payments must include payments for residual business assets. See U.S. Treas. reg. section 1.482-7A.
5 Luxembourg and Amazon separately appealed the commission’s decisions, and the GCEU joined their cases.
6 For the basics of state aid and its doctrine, see Ruth Mason, “Tax Rulings as State Aid FAQ,” Tax Notes, Jan. 23, 2017, p. 451.
7 For a critical look at the commission’s method for identifying state aid, see Mason, “Identifying Illegal Subsidies,” 69 Am. U.L. Rev. 479 (2019).
8 That distinguishes Amazon from Apple, and Ireland intervened in Amazon to complain that the commission applied a “sui-generis arm’s-length standard.” The commission set that question aside as inadmissible for mootness in Amazon, but I have written in these pages that a fundamental legal infirmity in the commission’s Apple decision is that it relies on a sui generis state aid standard. See Mason, “Tax Rulings as State Aid — Part 4: Whose Arm’s-Length Standard?” Tax Notes, May 15, 2017, p. 947.
9 Belgium and Forum 187 ASBL v. Commission, joined cases C-182/03 and C-217/03 (CJEU 2006).
10 For criticism of that reasoning and its derivation, see Mason, “State Aid Special Report — Part 6: Arm’s Length on Appeal,” Tax Notes, Feb. 5, 2018, p. 771. See also Philip Baker, “A Week of State Aid, Transfer Pricing and Tax Rulings,” Intertax (May 2021).
11 In Belgian Excess Profits (Commission Decision 2016/2326), the commission found that distinction to itself constitute state aid. The decision was annulled on other grounds.
12 The court observed that under the 1995 OECD transfer pricing guidelines:
The choice of profit split method, including on the basis of the contribution analysis, depends decisively on having identified external data from independent undertakings in order to determine the value of the contribution made by each. . . . However, the Commission did not seek to identify whether any such reliable data were available.
13 The commission’s argument was actually more indirect — without affirmatively arguing for all costs, it argued that operating costs was the wrong indicator. In any case, the commission never showed any advantage based on a comparison between the method actually used and what it thought should have been used.
14 Amazon also noted at the hearing that the method suggested by the commission would result in more operating profit for LuxOpCo than Amazon actually had in Europe in the relevant period.
15 As in Fiat, the GCEU in Amazon based that holding on the renowned expertise of the drafters of the OECD transfer pricing guidelines and the consensus nature of that guidance.
16 See also Stephen Daly, “The Power to Get It Wrong,” L.Q. Rev. (2021) (arguing that the commission should seek to ensure that tax authorities carry out the function of collecting tax objectively and dispassionately).
17 Baker, supra note 10.
18 For more on that history, see Mason, supra note 6.
19 For the problems with that standard, as well as speculation of why the commission erroneously thought it was justified, see Mason, supra note 8.
21 Michael J. Graetz, “Behind the European Raid on McDonald’s,” The Wall Street Journal, Dec. 3, 2015 (“To a U.S. lawyer steeped in the requirements of due process and the rule of law, the process by which these fines were decided doesn’t pass the smell test.”).
22 For commentary on the Fiat hearing before the CJEU Grand Chamber, see Baker, supra note 10.
23 See Mason, “Implications of the Rulings in Starbucks and Fiat for the Apple State Aid Case,” Tax Notes Int’l, Oct. 7, 2019, p. 15; and Mason and Daly, “State Aid: The General Court Decision in Apple,” Tax Notes Int’l, Sept. 7, 2020, p. 1317.
24 In Gibraltar, for instance, the GCEU annulled a commission decision that introduced a novel theory of state aid (the commission departed from the domestic-law reference base and found state aid by disproportionate impact), saying the new theory was not supported by precedent. Gibraltar v. Commission, joined cases T-211/04 and T-215/04 (CJEU 2008). On appeal, the CJEU approved the commission’s novel technique for identifying aid and reinstated the commission’s decision. Commission and Spain v. Gibraltar and United Kingdom, joined cases C-106/09 P and C-107/09 P (CJEU 2011). The CJEU later took the unusual step of limiting the precedential value of Gibraltar. Commission v. World Duty Free Group, joined cases C-20/15 P and C-21/15 P (CJEU 2016).
25 Based on its intervention in Amazon (and Fiat) to argue that “the commission misinterpreted the concept of advantage” under article 107(1) of the Treaty on the Functioning of the European Union “by using an inappropriate criterion, namely a ‘sui generis’ arm’s length principle,” Ireland seeks to secure a ruling that forecloses the imposition of that standard before the CJEU takes up the Apple appeal.
26 See Mason, supra note 7 (arguing that the commission could use the U.S. Supreme Court’s internal consistency test to evaluate domestic allocation rules for state aid).