OPINION OF ADVOCATE GENERAL
KOKOTT
delivered on 8 June 2023 (1)
Case C‑457/21 P
European Commission
v
Grand Duchy of Luxembourg,
Amazon.com, Inc.,
Amazon EU Sàrl
(Appeal – State aid – Grant of advantageous transfer pricing – Selective advantage under tax law – Determination of the reference system – OECD Transfer Pricing Guidelines as a reference system – Binding the Court of Justice to the reference system chosen by the Commission and the General Court – Assessment of the correct transfer price – Distinction between the assessment of facts and the consideration of points of law – Reduced standard of review for the assessment of a possible misapplication of tax law by national tax authorities – Tax ruling)
I. Introduction
1. The present appeal once again (2) concerns the review of a tax ruling in the light of the rules on State aid. While such tax rulings, on the one hand, contribute to legal certainty, there is at times a latent suspicion, on the other hand, that in some Member States they are based on agreements between tax authorities and taxpayers which are harmful to competition.
2. The present appeal also (3) concerns transfer pricing which may deviate from the arm’s length principle. The Luxembourg tax authorities had communications with Amazon.com in 2003 regarding the appropriate amount of a royalty between two subsidiaries. The amount of that royalty has an effect on the corporate income tax liability of Amazon EU Sàrl, established in Luxembourg. The higher the royalty is set, the less corporate income tax is payable in Luxembourg. In order to determine the appropriate royalty, Luxembourg and Amazon.com relied on a particular mutually agreed method. The Commission regarded that transfer pricing agreement as State aid since, in the Commission’s view, it was not consistent with the arm’s length principles of the OECD. The Commission made its own calculation to determine the appropriate amount of the royalty in accordance with a different method and arrived at a lower royalty. Since that would have resulted in a higher corporate income tax burden, the tax ruling was considered to have granted a selective advantage to the subsidiary paying the royalty.
3. In the proceedings before the General Court against that Commission decision on State aid, the parties primarily disputed the details of the method which had been or ought to have been applied. There was no dispute concerning whether the arm’s length principles of the OECD could be the correct reference system for a review of State aid in the present case. The General Court could not find that the determination of transfer pricing was erroneous, and annulled the decision on the grounds that no selective advantage had been demonstrated.
4. Subsequent to that General Court judgment, the Court of Justice established in Fiat Chrysler Finance Europe v Commission (4) that the OECD principles can only be used as a reference system for determining a selective advantage if they have been incorporated into national law. That is not the case in Luxembourg law. The Commission nevertheless takes the view that the General Court’s annulment of the Commission’s decision is vitiated by an error of law. Since the reference system was not in dispute between any of the parties, the Commission considers that the judgment in Fiat Chrysler Finance Europe v Commission has no bearing on the present appeal. The Commission avails itself of the ultra petita principle in that respect.
5. If that were so, the question would then arise as to the extent to which the Court of Justice has jurisdiction to review the ‘correct’ calculation of a transfer price in the context of the appeal. The Commission alleges in particular that the General Court assessed certain points (legal ownership of the intangible assets, contractual allocation of rights and obligations, functional analysis of the interposed subsidiaries, etc.) differently from the Commission. However, the General Court is, in principle, responsible for the assessment of facts, and at most a distortion of facts can be alleged in an appeal before the Court of Justice. Points of law, by contrast, lie within the inherent responsibility of the Court of Justice. (5) The present case demonstrates, however, that distinguishing between them is difficult when reviewing a selective advantage. It may be necessary to limit the intensity of the judicial review (that is to say, apply a modified standard of review) in the event that the Court of Justice interprets national tax law in view of the existence of a selective advantage.
II. Legal framework
A. European Union law
6. The legal framework within EU legislation is Article 107 et seq. TFEU.
B. Luxembourg law
7. Article 164(3) of the amended loi concernant l’impôt sur le revenu (Law on income tax; ‘the LIR’) of 4 December 1967 stipulates as follows:
‘The taxable income shall include hidden profit distributions. A hidden profit distribution arises in particular where a shareholder, stockholder or interested party receives either directly or indirectly benefits from a company or an association which he or she normally would not have received if he or she had not been a shareholder, a stockholder or an interested party.’
C. OECD Model Tax Convention and Guidelines
8. The Organisation for Economic Co-operation and Development (‘the OECD’) has published a number of non-binding guidance documents on international taxation. In particular, the OECD Model Tax Convention forms the basis of many bilateral tax conventions in the area of taxes on income and capital between OECD member countries and an increasing number of non-member countries. It also contains provisions on the appropriate profit attribution between companies within a multinational corporate group. In that respect, in the version in use from 2003 until 2014, Article 9(1) of the OECD Model Tax Convention reads as follows:
‘[Where] conditions are made or imposed between the two enterprises in their commercial or financial relations which differ from those which would be made between independent enterprises, then any profits which would, but for those conditions, have accrued to one of the enterprises, but, by reason of those conditions, have not so accrued, may be included in the profits of that enterprise and taxed accordingly.’
9. The OECD additionally provides guidance to tax administrations and multinational enterprises on the application of the arm’s length principle in its Transfer Pricing Guidelines, which are under continuous development. For the years concerned in the present case, primarily the Transfer Pricing Guidelines adopted on 13 July 1995 (‘the 1995 OECD Transfer Pricing Guidelines’) are relevant.
10. Paragraph 1.13 of the 1995 OECD Transfer Pricing Guidelines reads as follows:
‘… The arm’s length principle is sound in theory since it provides the closest approximation of the workings of the open market in cases where goods and services are transferred between associated enterprises. While it may not always be straightforward to apply in practice, it does generally produce appropriate levels of income between members of MNE [multinational enterprise] groups, acceptable to tax administrations. This reflects the economic realities of the controlled taxpayer’s particular facts and circumstances and adopts as a benchmark the normal operation of the market.’
11. The OECD Transfer Pricing Guidelines describe five methods to determine an arm’s length price of intra-group transactions: (i) the comparable uncontrolled price method (CUP method), (ii) the cost plus method, (iii) the resale price method, (iv) the transactional net margin method (TNMM) and (v) the transactional profit split method. In general, the most appropriate transfer pricing method must be applied with reference to the circumstances of the case. However, for difficult cases, where no one method is conclusive, a flexible approach would allow the evidence of more than one method to be used in conjunction. Multinational corporate groups retain the freedom to apply methods not described in those guidelines to establish transfer prices, provided those prices satisfy the arm’s length principle.
III. Background to the dispute
12. The dispute arose from tax rulings issued in 2003 by the Grand Duchy of Luxembourg at the request of Amazon.com on the corporate income tax treatment of two companies newly established in Luxembourg for the purposes of Luxembourg corporate income tax for the years 2006 to 2014.
A. The tax ruling
13. The context was a planned restructuring of the activities of Amazon.com, Inc., established in the United States of America, and the companies which it controls (together: ‘the Amazon group’). The companies of the Amazon group in Europe engage in online activities, in particular online retail and the provision of various online services. To that end, the Amazon group manages a number of internet sites in various languages of the European Union, including amazon.de, amazon.fr, amazon.it and amazon.es.
14. Prior to May 2006, the Amazon group’s European business was managed from the United States. In particular, the retail activities and service provision on the European internet sites were managed by two companies established in the United States, namely Amazon.com International Sales, Inc. (‘AIS’) and Amazon International Marketplace (‘AIM’), and other companies established in France, Germany and the United Kingdom.
15. The restructuring, which took place in 2006, involved the creation of two companies established in Luxembourg. Specifically, these were Amazon Europe Holding Technologies SCS (‘LuxSCS’), an ordinary limited partnership (société en commandite simple) under Luxembourg law, the partners in which were American companies, and Amazon EU Sàrl (‘LuxOpCo’).
16. As a first step, LuxSCS concluded a number of agreements with certain companies of the Amazon group established in the United States, as follows:
– the licence and assignment agreements for pre-existing intellectual property concluded with Amazon Technologies, Inc. (‘ATI’), a member of the Amazon group established in the United States (together: ‘the buy-in agreement’);
– a cost-sharing agreement concluded in 2005 with ATI and A9.com, Inc., a member of the Amazon group established in the United States. Under the buy-in agreement and the cost-sharing agreement, LuxSCS obtained the right to exploit certain intellectual property rights and ‘by-products’ of those intangible assets, which were owned and being further developed by A9.com and ATI. The intangible assets covered by the cost-sharing agreement consisted essentially of three categories of intellectual property, namely technology, customer data and trade marks. Under the cost-sharing agreement and the buy-in agreement, LuxSCS could also sub-license the intangible assets, in particular with a view to operating the EU websites. In return for those rights, LuxSCS was required to make buy-in payments and to pay its annual share of the costs related to the cost-sharing agreement development programme.
17. As a second step, LuxSCS, for its part, concluded a licence agreement with LuxOpCo with effect from 30 April 2006 in respect of the abovementioned intangible assets (‘the licence agreement’). Under that agreement, LuxOpCo, as the licence holder, acquired the right to exploit the intangible assets in return for payment of a royalty to LuxSCS (‘the royalty’).
18. Lastly, LuxSCS concluded an agreement for the licensing and assignment of intellectual property rights with Amazon.co.uk Ltd, Amazon.fr Sàrl and Amazon.de GmbH, under which LuxSCS obtained the right to exploit certain trade marks and the intellectual property rights in respect of the EU internet sites.
19. In 2003, Amazon.com applied for a tax ruling on those matters. It concerned the calculation of the amount of the royalty to be paid by LuxOpCo to LuxSCS from 30 April 2006 onwards and was based on a transfer pricing report. The authors of that report, in essence, proposed a transfer pricing method for determining the transfer prices on the basis of which LuxOpCo’s corporate income tax liability in Luxembourg could be determined. In particular, Amazon.com requested confirmation that the transfer pricing arrangement determining the rate of the annual royalty would result in an ‘appropriate and acceptable profit’ for LuxOpCo with respect to the transfer pricing policy and Article 56 and Article 164(3) of the LIR. The method of calculating the royalty payable by LuxOpCo to LuxSCS was described as follows in the application:
‘1. Compute and allocate to LuxOpCo the ‘LuxOpCo Return’, which is equal to the lesser of (a) [confidential]% of LuxOpCo’s total EU Operating Expenses for the year and (b) total EU Operating Profit attributable to the European Web Sites for such year;
2. The License Fee shall be equal to EU Operating Profit minus the LuxOpCo Return, provided that the License Fee shall not be less than zero;
3. The Royalty Rate for the year shall be equal to the License Fee divided by total EU Revenue for the year;
4. Notwithstanding the foregoing, the amount of the LuxOpCo Return for any year shall not be less than 0.45% of EU Revenue, nor greater than 0.55% of EU Revenue;
5.(a) In the event that the LuxOpCo Return determined under step (1) would be less than 0.45% of EU Revenues, the LuxOpCo Return shall be adjusted to equal the lesser of (i) 0.45% of Revenue or EU Operating Profit or (ii) EU Operating Profit;
(b) In the event that the LuxOpCo Return determined under step (1) would be greater than 0.55% of EU Revenues, the LuxOpCo Return shall be adjusted to equal the lesser of (i) 0.55% of EU Revenues or (ii) EU Operating Profit.’
20. Amazon also sought confirmation from the Luxembourg tax authorities of the tax treatment of LuxSCS, of its partners established in the United States and of the dividends received under that structure. The letter explained that LuxSCS, as a an ordinary limited partnership (société en commandite simple), did not have a separate tax personality from that of its partners and that, as a result, it was not subject to corporate income tax or net wealth tax in Luxembourg.
21. The Luxembourg tax authorities approved that assessment and calculation by means of a tax ruling in November 2003.
B. The procedure and decision of the Commission
22. On 24 June 2014, the European Commission requested the Grand Duchy of Luxembourg to provide information on the tax rulings granted to the Amazon group. On 7 October 2014, it published the decision to initiate a formal investigation procedure under Article 108(2) TFEU.
23. During the investigation thereby initiated, the Commission requested a variety of information from the Grand Duchy of Luxembourg and Amazon.com. Among the responses to those requests for information, Amazon.com submitted a copy of an Opinion of the United States Tax Court of 23 March 2017 which had been issued in relation to an action brought by the Internal Revenue Service (the United States federal government tax collection agency; IRS) concerning the amount of payments related to the agreements referred to in point 16 above.
24. In addition, Amazon.com submitted to the Commission a new transfer pricing report drawn up in 2017 by a tax adviser, the purpose of which was to verify ex post whether the royalty paid by LuxOpCo to LuxSCS in accordance with the tax ruling at issue was consistent with the arm’s length principle.
25. On 4 October 2017, the Commission adopted Decision (EU) 2018/859 on State aid SA.38944 (2014/C) (ex 2014/NN) implemented by Luxembourg to Amazon (‘the decision at issue’). (6)
26. Article 1 of that decision reads as follows:
‘The tax ruling [at issue], by virtue of which Luxembourg endorsed a transfer pricing arrangement … that allowed [LuxOpCo] to assess its corporate income tax liability in Luxembourg from 2006 to 2014 and the subsequent acceptance of the yearly corporate income tax declaration based thereon constitutes aid ….’
27. The Commission explained that, where a tax ruling endorses a result that does not reflect in a reliable manner what would result from a normal application of the ordinary tax system, that ruling will confer a selective advantage on the beneficiary. That selective treatment, it held, results in a lowering of that taxpayer’s tax liability and a selective advantage as compared to companies in a comparable factual and legal situation. The Commission also considered that, in the present case, the tax ruling at issue had conferred a selective advantage on LuxOpCo by lowering its corporate income tax liability in Luxembourg. That conclusion is based on a primary finding and three subsidiary findings.
28. In Section 9.2.1 of the decision at issue, the Commission considered that the tax ruling at issue had produced an outcome which departed from a reliable approximation of a market-based outcome. In the Commission’s view, endorsing a transfer pricing arrangement that attributed a remuneration to LuxOpCo solely for the ‘routine’ functions performed by LuxOpCo and which attributed the entire profit generated by LuxOpCo in excess of that remuneration to LuxSCS in the form of a royalty payment was incompatible with the arm’s length principle.
29. According to the Commission, the TNMM rather than the CUP method would have been the most appropriate transfer pricing method to determine the royalty to be paid by LuxOpCo under the licence agreement. It took the view, however, that the party performing unique and valuable functions was LuxOpCo and not LuxSCS. Accordingly, the tested party for the purpose of applying the TNMM should have been LuxSCS rather than LuxOpCo. That would have resulted in a lower royalty and consequently a higher corporate income tax burden for LuxOpCo.
30. In Section 9.2.2 of the decision at issue, entitled ‘Subsidiary finding of an economic advantage’, the Commission sets out its subsidiary finding of an advantage, under which, even if the Luxembourg tax administration were correct in having accepted the analysis of LuxSCS’s functions set out in the 2003 transfer pricing report, the transfer pricing arrangement endorsed by the tax ruling at issue was, in any event, based on inappropriate choices that produced an outcome which departed from a reliable approximation of a market-based outcome. In that context, the Commission made three separate subsidiary findings.
31. In its first subsidiary finding, the Commission stated that LuxOpCo had been inaccurately considered to perform only ‘routine’ management functions and that the profit split method, together with the contribution analysis, ought to have been applied. In its second subsidiary finding, the Commission found that the choice of operating expenses as a profit level indicator was incorrect. In its third subsidiary finding concerning the advantage, the Commission considered that the inclusion of a ceiling of 0.55% of EU turnover was inappropriate.
C. The procedure before the General Court and the judgment under appeal
32. By application lodged at the Registry of the General Court on 14 December 2017, the Grand Duchy of Luxembourg brought the action in Case T‑816/17.
33. By application lodged at the Registry of the General Court on 22 May 2018, Amazon EU and Amazon.com (together, ‘Amazon’) brought the action in Case T‑318/18.
34. By document lodged at the Registry of the General Court on 16 April 2018, Ireland applied for leave to intervene in Case T‑816/17 in support of the form of order sought by the Grand Duchy of Luxembourg. By order of 29 May 2018, the President of the Seventh Chamber, Extended Composition, granted Ireland leave to intervene.
35. Cases T‑816/17 and T‑318/18 were joined for the purposes of the oral part of the procedure and, later, for the purposes of the judgment.
36. The Grand Duchy of Luxembourg and Amazon raised a number of pleas in support of their actions, most of which overlapped. The Grand Duchy of Luxembourg and Amazon disputed, in essence, the Commission’s finding that there was an advantage in favour of LuxOpCo as provided for in Article 107(1) TFEU. They also challenged the Commission’s subsidiary findings on the existence of a tax advantage in favour of LuxOpCo as provided for in Article 107(1) TFEU. In particular, they disputed the Commission’s primary and subsidiary findings in respect of the selectivity of the tax ruling at issue. The Grand Duchy of Luxembourg moreover submitted that the Commission had infringed the Member States’ exclusive competence in the area of direct taxation. They disputed, in particular, whether the 2017 version of the OECD Guidelines, as used by the Commission in adopting the decision at issue, was relevant in the present case.
37. In its statement in intervention, Ireland alleges, inter alia, infringement of Article 107(1) TFEU in that the Commission has not established the existence of an advantage in favour of LuxOpCo and infringement of Articles 4 and 5 TEU in that the Commission engaged in disguised tax harmonisation.
38. By the judgment under appeal, the General Court upheld the action and annulled the Commission decision at issue. In paragraph 133 et seq. of the judgment under appeal, however, the General Court rejected as inadmissible the argument put forward by Ireland that the Commission had applied an arm’s length principle in accordance with the OECD Guidelines which was not enshrined in Luxembourg law, because the argument put forward by Ireland was not connected with the considerations which the Grand Duchy of Luxembourg included in its pleas.
39. The General Court held that Ireland’s arguments concerned, in essence, the legal basis relied on by the Commission as regards the obligation imposed on the Grand Duchy of Luxembourg to apply the arm’s length principle. It found therefore, that Ireland called into question the sources of law of that principle, as applied by the Commission in the decision at issue. In addition, the General Court held, Ireland’s arguments related to the interpretation of the content of that principle and not to its application by means of a transfer pricing method. The first plea raised by the Grand Duchy of Luxembourg, however, did not, in the view of the General Court, relate to the question of the source of law of that principle or to questions of interpretation of that principle.
IV. The procedure before the Court of Justice
40. On 27 July 2021, the Commission lodged the present appeal against the judgment of the General Court. The Commission contends that the Court of Justice should:
– set aside the judgment of the General Court of the European Union (Seventh Chamber, Extended Composition) of 12 May 2021 in Joined Cases T‑816/17 and T‑318/18, Luxembourg and Others v Commission (EU:T:2021:252);
– reject the first plea in Case T‑816/17 and the second, fourth, fifth and eighth pleas in Case T‑318/18;
– refer the case back to the General Court for consideration of the pleas not already assessed;
– alternatively, give final judgment in the matter in accordance with Article 61 of the Statute of the Court of the Justice of the European Union;
– reserve the costs of the present proceedings, if it refers the case back to the General Court, or order Luxembourg, Amazon EU, and Amazon.com to pay the costs of the proceedings, if it gives final judgment in the matter.
41. Luxembourg, Amazon EU, and Amazon.com contend that the Court of Justice should dismiss the appeal and order the Commission to bear the costs. In the alternative, Luxembourg contends that the proceedings should be referred back to the General Court of the European Union.
42. Before the Court of Justice, all the parties, not including Ireland, submitted written observations on the questions communicated by the Court and, with Ireland participating, presented oral arguments at the hearing of 16 March 2023.
V. Legal assessment
43. The Commission puts forward two grounds of appeal against the judgment. The first ground of appeal challenges the complaint, set out in paragraphs 162 to 297 of the judgment under appeal, of an advantage found in the decision at issue. First, it alleges that the General Court misapplied the arm’s length principle when it found that the Commission’s functional analysis was flawed. Second, it argues that the General Court erred in criticising the calculation of the appropriate royalty in the decision at issue.
44. By its second ground of appeal, the Commission challenges the rejection, in paragraphs 314 to 538 of the judgment under appeal, of the subsidiary finding of advantage in the decision. It contends first that the General Court erred in the standard of proof for a finding of advantage. Second, it submits that the General Court misapplied the arm’s length principle and, moreover, invoked arguments on its own motion in order to reject the finding of advantage which the applicants had not put forward.
45. In the light of those grounds of appeal, the question first arises whether the Court of Justice must concern itself with the details of calculating the ‘correct’ transfer price for the royalty under the OECD arm’s length principles. That would imply that those principles constitute the relevant reference system, which has been questionable since the judgment of the Court of Justice in Fiat Chrysler Finance Europe v Commission (7) (see A.1.b.3 below). However if the question of the correct reference system no longer arose because no plea to that effect had been submitted, it would have to concern itself with the details of calculating the ‘correct’ transfer price (see A.1.b.2 below).
46. In the event that the 1995 OECD Guidelines on the arm’s length principle are to be relied on, clarification is required concerning the extent to which the Court of Justice can review the ‘correct’ calculation of a transfer price in the context of the appeal. In that respect, a distinction must also be made between the assessment of facts and the consideration of questions of law (see A.2.b.1 below). An additional element of importance is the relevant standard of review for assessing the application of national tax law (see A.2.b.2 below). These questions also apply to the second ground of appeal (see B below).
A. The first ground of appeal: erroneous criticism of the finding in the decision at issue of advantage as provided for in Article 107(1) TFEU
1. The existence of State aid (selective advantage)
47. By its first ground of appeal, the Commission submits that the General Court erred, in paragraphs 162 to 297 of the judgment under appeal, in criticising the Commission’s premiss that LuxOpCo performed unique and valuable functions and that the royalty payments to LuxSCS had therefore been excessive. In that respect, the Commission submits that the General Court misapplied the arm’s length principle and therefore erred in annulling the decision at issue.
48. Accordingly, the Commission complains that the General Court, in agreement with the view of Luxembourg and Amazon, saw no State aid as provided for in Article 107(1) TFEU in the tax rulings.
(a) Findings of the General Court
49. In paragraph 296 of the judgment under appeal, the General Court substantiated the finding of no State aid by holding that the Commission had not demonstrated the existence of a selective advantage. The elements set out in the primary finding of an advantage, it argued, did not establish that LuxOpCo’s tax burden had been artificially reduced as a result of over-pricing the royalty. In the view of the General Court, the calculation applied by Luxembourg to determine the appropriate royalty for LuxOpCo to pay to LuxSCS is not vitiated by any error of laws. Luxembourg, it found, had correctly applied the relevant OECD Guidelines.
50. In that respect, the General Court held in paragraph 137 that the arm’s length principle, as applicable, in its view, in the instant case, may be inferred from Article 164(3) of the LIR. That was apparent, it found, from recital 241 of the decision at issue, and that conclusion had not been called into question by the parties.
51. In paragraph 154, the General Court held that the Commission was entitled to base its findings concerning the existence of an advantage on the 1995 OECD Guidelines. In contrast, the 2017 version of the OECD Guidelines, not yet published at the time in question, ought not to have been used as a reference as had been done in the decision. In paragraph 162 et seq., it found, moreover, that the TNMM chosen by the Commission in the decision at issue and its application to LuxSCS were not correct.
52. As a result, in relation to its examination of Luxembourg’s and Amazon’s pleas seeking to dispute the existence of a selective advantage, the General Court considered Article 164(3) of the LIR, read in conjunction with the 1995 OECD Guidelines, to be the relevant reference system on which the Commission had relied. Applying that reference system, it came to the conclusion that the Commission, by selecting the TNMM, applied to LuxSCS, had erred in its decision by inferring there had been a selective advantage.
(b) Assessment
(1) Existence of a selective advantage
53. In accordance with the settled case-law of the Court of Justice, classification as ‘State aid’ as provided for in Article 107(1) TFEU requires, first, that there must be intervention by the State or through State resources; second, that the intervention must be liable to affect trade between Member States; third, that it must confer a selective advantage on the beneficiary; and fourth, that it must distort, or threaten to distort, competition. (8)
54. In the present case, the only problematic issue is whether there was a selective advantage. In accordance with the settled case-law of the Court of Justice, for tax measures, selectivity must be determined in a number of stages. The first step is to identify the ordinary or ‘normal’ tax regime applicable in the Member State concerned (the ‘reference system’). (9) The second step is to judge, on the basis of that ordinary or ‘normal’ tax regime, whether the tax measure at issue is a derogation from that ordinary system, in so far as it differentiates between operators which, in the light of the objective pursued by the ordinary regime, are in a comparable factual and legal situation. (10) If a derogation from ‘normal taxation’ has been established, the last step is to assess whether the derogation is justified.
55. In determining the reference system, the starting point can only ever be the national legislature’s decision as to what it deems to be ‘normal’ taxation. Fundamental taxation decisions, particularly the decisions about taxation techniques but also the objectives and principles of taxation, are therefore for the Member State to make. (11) It follows that neither the Commission nor the Court of Justice can judge national tax law by some ideal or fictitious tax system. (12) After all, outside the spheres in which EU tax law has been harmonised, it is for the Member State concerned, by exercising its own competence by virtue of its fiscal autonomy, to determine the characteristics constituting the tax which define the reference system or the ‘normal’ tax regime for the purposes of analysing the condition relating to selectivity. (13)
56. The determination of the reference system constitutes the starting point for the comparative examination to be carried out in the context of the assessment of selectivity. Therefore, an error made in that determination necessarily vitiates the whole of the analysis of the condition relating to selectivity and thus the existence of State aid as provided for in Article 107 TFEU. (14) Consequently, the Commission’s appeal can only be successful if the General Court erred in concluding that there was no aid even though the reference system relied on by the Commission was the correct one (see A.1.b.3 below). The same might apply if the Court of Justice is bound by the reference system which the General Court relied on and found to be appropriate (see A.1.b.2 below).
(2) Review of the reference system in the context of the appeal
57. First, it is necessary to clarify whether a possible error in determining the reference system in the decision at issue could affect the judgment of the General Court. The Commission submits that it could not, as neither Luxembourg nor Amazon challenged the determination of the reference system in their actions. The Commission submits that the General Court also based its considerations on the reference system relied on by the Commission. Consequently, it contends, the Court of Justice can no longer object to that choice of reference system. However, such a line of argument cannot be accepted.
58. Both Luxembourg, with the support of Ireland, and Amazon disputed the existence of a selective advantage and thereby called into question the lawfulness of the decision at issue in the light of the objective concept of State aid as provided for in Article 107(1) TFEU. The General Court subsequently annulled the decision on the grounds that no selective advantage had been demonstrated. The Commission is therefore challenging that annulment on the grounds that the concept of State aid was misconstrued.
59. However, the annulment is only erroneous if the decision at issue was correct in concluding that there was a selective advantage. In that situation, the General Court’s judgment to the contrary would have to be set aside. However – as explained in point 56 above – the lawfulness of that finding and of the finding that there was prohibited State aid as provided for in Article 107(1) TFEU necessarily depends on the reference system having been determined correctly. Without that, it is not possible to conduct the comparative analysis which is a prerequisite for assessing the selectivity of the advantage. The question whether the reference system was determined correctly in the decision at issue is inextricably linked with that of the existence of a selective advantage. It therefore constitutes, as a conditio sine qua non, a question of law to be reviewed in the context of the appeal.
60. Luxembourg and Amazon called into question the lawfulness of the decision at issue on the grounds that there was no selective advantage. The determination of the reference system made in that decision cannot be considered in isolation. Nor can it have become final in isolation and thereby exempt from further review by the Courts of the European Union. That is not altered by the fact that neither Luxembourg nor Amazon explicitly challenged the reference system used by the Commission. On the contrary, in their pleas disputing the existence of a selective advantage, they simultaneously challenged all the conditions necessary for it to be possible to affirm a selective advantage. Those conditions include the correctly determined reference system, since, as the Court of Justice has already ruled on a number of occasions (15), the review of a selective advantage would otherwise be erroneous per se.
61. For that reason, moreover, it is irrelevant that, in paragraph 136 et seq. of the judgment under appeal, the General Court rejected Ireland’s intervention as being inadmissible on purely formal grounds. Ireland had concentrated on the reference system. Admittedly, Ireland has not challenged that rejection. However, the issue of the correct reference system was part, as a conditio sine qua non, of the question of whether there was a selective advantage, which was raised by the other parties (Luxembourg and Amazon).
62. Contrary to the Commission’s assertions in its reply to the written questions put by the Court of Justice, (16) that also does not breach the ultra petita principle. Annulling the decision at issue on the grounds of an incorrect choice of reference system does not go beyond the forms of order sought. Those were, after all, aimed at exactly such an annulment on the grounds that there was no selective advantage. The same applies to the appeal, since the Commission is of the opinion that the General Court erred in concluding that there was no selective advantage.
63. In that respect, the Court of Justice can (and must) first examine whether the reference system relied on by the General Court and the Commission was determined correctly and whether, as the General Court held, no selective advantage resulted from it. If that is not so, the Court of Justice cannot, as a result, claim that the General Court erred in law.
64. It may be that the reasoning of the General Court was erroneous but that the outcome would nonetheless still be correct, the error having no relevance to the operative part of the General Court’s ruling. (17) If the grounds of a judgment of the General Court disclose an infringement of EU law but its operative part is shown to be well founded on other legal grounds, such an infringement is not capable of bringing about the annulment of that judgment. Rather, only a substitution of grounds must be made. (18)
(3) The choice of the correct reference system
65. It is therefore necessary to clarify whether the reference system used by the Commission in the decision at issue was chosen correctly.
66. In recital 241 of the decision at issue, the Commission identifies Article 164(3) of the LIR as the relevant national legal framework, stating that it was considered to enshrine the arm’s length principle until the end of 2016. The Commission contends that, as that provision makes no distinction between international and national transactions, it can be affirmed that the Luxembourg transfer pricing rules reflect the OECD Transfer Pricing Guidelines even though Article 164(3) of the LIR makes no express reference to those guidelines. In the footnote present there, however, the Commission refers to recital 294 of the decision at issue, which states Luxembourg’s view that no reference was made in Luxembourg law to the 1995 OECD Transfer Pricing Guidelines. Luxembourg contends that, on the contrary, national transfer pricing rules were applied at the time.
67. In that respect, the reference system on the basis of which the Commission reviewed and affirmed the existence of a selective advantage remains unclear. In its remarks on the assessment of the contested measure in recital 392 et seq. of the decision at issue, the Commission speaks of transfer prices which do not resemble prices negotiated at arm’s length (recital 402) or a departure from a reliable approximation of a market-based outcome (recital 406, likewise recital 584) without, however, characterising them as subject to specific provisions of Luxembourg law.
68. What is clear from the footnotes, however, is that the Commission refers exclusively to the OECD Transfer Pricing Guidelines from different time periods. In connection with the review of a 2003 tax ruling, it refers to the 2010 OECD Transfer Pricing Guidelines (see footnotes 409, 419, 429, 646, 677) or even to the 2017 OECD Transfer Pricing Guidelines (see footnotes 410, 417, 447, 679), or to the 1995 OECD Transfer Pricing Guidelines (see footnotes 411, 418, 430, 647), or to all three of them together (see footnotes 426, 427, 428, 635, 649).
69. Luxembourg law, on the other hand, does not appear to have played any role in that review. That becomes clear in recitals 410 and 411 of the decision at issue. The Commission sees the selective advantage in an erroneous functional analysis, as would follow from the 1995, 2010 and 2017 OECD Transfer Pricing Guidelines.
70. As the Court of Justice has recently pointed out, (19) without harmonisation in that regard, any fixing of the methods and criteria for determining an ‘arm’s length’ outcome falls within the discretion of the Member States. Although the member States of the OECD recognise the merits of using the arm’s length principle in order to establish the correct allocation of company profits between different countries, there are significant differences between those States in the detailed application of transfer pricing methods. As the Commission itself states in recital 255 et seq. of the decision at issue, the OECD Guidelines provide for a number of methods of approximating arm’s length pricing in relation to transactions and profit allocation between companies in one group.
71. Parameters and rules external to the national tax system at issue cannot be taken into account in the examination of the existence of a selective tax advantage as provided for in Article 107(1) TFEU and for the purposes of establishing the tax burden that should normally be borne by an undertaking, unless that national tax system makes explicit reference to them. (20) As the Court of Justice goes on to point out, (21) that is also an expression of the principle of legality of taxation, which forms part of the legal order of the European Union as a general principle of law, requiring that any obligation to pay a tax and all the essential elements defining the substantive features thereof must be provided for by law. That is because the taxable person has to be in a position to foresee and calculate the amount of tax due and determine the point at which it becomes payable. (22)
72. If the principle of legality of taxation is taken seriously, then applying the OECD Transfer Pricing Guidelines in the manner of applicable law and as a reference system is only possible if applicable law makes explicit reference to those guidelines. In my view, consistent administrative practice which serves to flesh out some feature of a legal definition (that of hidden profit distributions in the present case) could be sufficient to expand the reference system to encompass all or part of the OECD Transfer Pricing Guidelines. However, for consistent administrative practice (23) to be taken into account as a reference system, the Commission would have to have investigated that administrative practice in the Member State concerned and to have described that in its decision. That has not been done in the present case. Moreover, it would appear to be precluded that administrative practice in Luxembourg in 2003 did actually take into account the 2010 and 2017 Transfer Pricing Guidelines, which had not yet been adopted.
73. By basing its review of the appropriate amount of the royalty exclusively on the OECD Transfer Pricing Guidelines from 1995, 2010 and even 2017, the Commission as a result applied a different arm’s length principle from that established in Luxembourg law – as it had done before, in Fiat Chrysler Finance Europe v Commission. (24) As is apparent from recitals 241 and 294 of the decision at issue, Luxembourg law does not refer to the OECD Transfer Pricing Guidelines.
74. Contrary to the Commission’s oral arguments, the fact that, in 2003, Luxembourg’s administrative practice took account of the 1995 OECD Transfer Pricing Guidelines in determining transfer pricing does not alter that conclusion. That fact was consequently confirmed by Luxembourg and Amazon. Nonetheless, both also stressed the need to differentiate between taking those OECD Transfer Pricing Guidelines into account and applying them in the manner of binding law. They contended that these were merely non-binding guidelines which had been included in the considerations associated with applying Article 164(3) of the LIR. At the hearing, Luxembourg said in response to a written question from the Court of Justice that the OECD Transfer Pricing Guidelines had been important as principles of interpretation. They had, however, no legal force at the time and had been regarded as non-exhaustive. Other transfer pricing methods could therefore have been applied.
75. The actual extent to which the 1995 OECD Transfer Pricing Guidelines were taken into account in Luxembourg is not apparent from the decision at issue. Instead, the fact that the decision at issue repeatedly relies on the as yet unpublished 2010 and 2017 versions of the OECD Transfer Pricing Guidelines demonstrates that the Commission did not base its arguments on the legal situation and administrative practice in Luxembourg in 2003, when the tax ruling was issued.
76. In addition, nothing to contradict that can be inferred from the Commission’s submission at the hearing, where it expressed the view that the OECD, in publishing the 2010 and 2017 Transfer Pricing Guidelines, merely sought to describe once more (in a declaratory manner) the 1995 Transfer Pricing Guidelines which already existed without making any changes to the content. The OECD certainly did attempt to develop the content of the Transfer Pricing Guidelines by publishing the 2010 and 2017 versions. For example, in the context of Actions 8 to 10 of the so-called BEPS project, the so-called ‘DEMPE’ functions (being the value-creation aspects of development: Development, Enhancement, Maintenance, Protection and Exploitation) were developed at the OECD level. (25)
77. The aim of that development was to bring the transfer pricing rules into line with value creation between connected enterprises to ensure that transfer prices reflect the economic circumstances of a transaction. In 2017, the so-called DEMPE concept was incorporated into the latest version of the OECD Transfer Pricing Guidelines. In contrast, that concept is not present in the 1995 Transfer Pricing Guidelines. In 2003 – as Luxembourg pointed out at the hearing – it was as yet unknown. Nevertheless, it was the very concept upon which the Commission relied in the decision at issue (see just recital 262 et seq.).
78. The Commission therefore confined itself to identifying, in the objective pursued by the general corporate income tax system in Luxembourg, the abstract expression of the arm’s length principle and to examining the tax ruling at issue without taking into account the manner in which that principle has actually been incorporated into Luxembourg law, or applied, with regard to integrated companies in particular.
79. It follows from all of the foregoing that the Commission did not take the Luxembourg national law in force when the tax ruling was issued as the relevant reference system for its review of a selective advantage. Consequently, all the subsequent considerations in the decision at issue are also vitiated by an error of law. The Commission therefore erred in law in applying Article 107(1) TFEU. The General Court was therefore correct in annulling the Commission decision at issue – albeit on different grounds – in the absence of a demonstrated selective advantage. The Court of Justice is not required to rule on whether those other grounds – which the Commission expressly challenges in its appeal – are tenable.
(4) Conclusion
80. Since the Commission did not take Luxembourg law (where appropriate in conjunction with Luxembourg administrative practice at the time) as the relevant reference system for its review of a selective advantage, the findings in the decision at issue are vitiated by error. The General Court was therefore fully entitled – albeit on grounds other than those it relied on – to annul the Commission decision at issue. The first ground of appeal put forward by the Commission is therefore unfounded.
2. In the alternative: flawed functional analysis by the General Court – a question of fact or of law?
81. On the other hand, if the Court of Justice were to follow the Commission’s argument that there is no longer any need to examine the choice of correct reference system because the parties did not expressly challenge that choice by means of their pleas, it would then be necessary to examine the Commission’s first ground of appeal in more detail.
82. By that ground of appeal, the Commission alleges that the General Court erred in law in its application of the arm’s length principle when it found that the Commission’s functional analysis was flawed. Second, it claims that the General Court erred in criticising the calculation of the appropriate royalty in the decision at issue.
(a) Findings of the General Court
83. The General Court founded the absence of State aid in paragraph 156 et seq. of the judgment under appeal on the absence of a selective advantage. In the view of the General Court, the calculation applied by Luxembourg to determine the appropriate royalty for LuxOpCo to pay to LuxSCS is not vitiated by any error of law. It held that the OECD Guidelines had been correctly applied, as explained in point 49 et seq. above.
(b) Assessment
(1) Is the correct application of the arm’s length principle a question of fact or a question of law?
84. The Commission submits that the General Court erred in law in its application of the arm’s length principle. In that regard, it argues, the General Court erred in criticising the Commission’s functional analysis in the decision at issue. The Commission had been correct in its assessment that LuxSCS had been the least complex party and accordingly the correct subject for the arm’s length method chosen by the Commission (the TNMM in the present case). In the Commission’s view, the General Court’s assessment that LuxSCS performed active functions (and that LuxOpCo was therefore the company to be tested) was, however, erroneous.
85. The question arises whether those alleged errors can be reviewed in the context of the appeal. After all, the assessment of facts – including, in principle, the General Court’s findings on national law – (26) and evidence does not constitute a question of law which the Court of Justice can review in the context of an appeal. (27) The Court of Justice has jurisdiction only to determine whether that law was distorted. (28) However, the Commission does not allege a distortion of facts in its appeal, and such a distortion appears to be precluded in the present case. The General Court ‘only’ assessed the functions of the companies concerned (LuxOpCO and LucxSCS) differently, and therefore found a different arm’s length method to be correct and reviewed the tax ruling at issue with reference to that method. In so doing, it was unable – unlike the Commission – to discern a selective advantage in the amount of the royalty paid by LuxOpCo to LuxSCS.
86. It is now clear from the case-law of the Court of Justice that the correct determination of the correct reference system by the Commission or the General Court is a question of law which is subject to review on appeal. (29) That is because the Court of Justice can assess, as a question of law, the legal classification which has been given to national law by the General Court in the light of a provision of EU law. (30) That is pivotal within the scope of Article 107(1) TFEU, as the selective advantage necessary to identify State aid arises only from a derogation from the ‘normal’ national tax system, the reference system. An argument seeking to challenge the choice of appropriate reference system made in the first step of reviewing a selective advantage is therefore admissible in the context of appeal proceedings.
87. A matter which has not yet been expressly decided is whether specific classification under that reference system (and thus the application of national law in the present case) is to be considered a question of law, which the Court of Justice can review in the context of the appeal, or whether that application of national law is to be treated as an assessment of facts.
88. I agree with the Commission, which ultimately wishes to have the correct determination (interpretation) of the reference system and the correct classification (application) in relation to that reference system treated identically. It is very difficult to separate the two, as the present case demonstrates.
89. Is the transfer pricing method (CUP method or TNMM) which results from correct application of the OECD Guidelines on the arm’s length principle still part of the reference system and therefore a question of law? Or is the choice of the transfer pricing method a classification in relation to the reference system (the OECD Guidelines on the arm’s length principle) and therefore a question of fact? The difference would be a matter of gradations at most. Since the determination of the right transfer pricing method, like a relevant fact, establishes the framework for the calculation of the transfer price, that determination of the right transfer pricing method is also, in my view, part of the legal reference system.
90. That, however, leads to the problem that the Court of Justice is called upon in the appeal to review in detail the factually and legally complex issue of the existence of a selective advantage. If, in the context of tax-related State aid, that depends on a departure from the national reference system, then the Court will have to interpret and apply national law (meaning, in the present case, the Luxembourg arm’s length principle or, in the Commission’s view, the OECD arm’s length principles) in that regard. As I have recently explained in detail elsewhere, (31) it is not actually for the court to do so. The problems associated therewith should be accounted for by means of a modified standard of review (see point 91 et seq. below).
(2) A better alternative: modified standard of review
91. Not every departure from the national reference system (the law on tax, in the present case) which benefits the taxpayer, and consequently not every incorrect tax ruling in favour of the taxpayer, can constitute State aid as provided for in the Treaties.
92. Even if such an interpretation were covered by the wording of Article 107(1) TFEU, that outcome would contradict the standard of review developed by the Court of Justice for State aid schemes in the form of general tax laws. The Court has emphasised that, given the current state of harmonisation of EU tax law, the Member States are free to establish the system of taxation which they deem most appropriate. (32) That explicitly applies to the area of State aid. (33) That discretion enjoyed by the Member States extends to determining the characteristics constituting each tax and applies in particular to the implementation and configuration of the arm’s length principle for transactions between affiliated companies. (34)
93. That discretion enjoyed by the Member States only exceeds its limits if the Member States abuse their tax law in order to grant advantages to individual undertakings in circumvention of the rules on State aid. (35) Such an abuse of fiscal autonomy can be assumed when tax law is configured in an obviously inconsistent manner. (36) In its more recent case-law, therefore, the Court of Justice only subjects general taxation decisions to scrutiny for compliance with State aid law if they have been designed in a manifestly discriminatory manner, with the aim of circumventing the requirements of EU law on State aid. (37)
94. There is no apparent reason not to transpose that case-law to situations where the law is misapplied in favour of the taxpayer. It follows, then, that not every incorrect tax ruling but only those which are manifestly erroneous in favour of the taxpayer constitute a selective advantage. Derogations from the applicable national reference system are manifestly erroneous if they cannot be plausibly explained to a third party, such as the Commission or the Courts of the European Union, and are therefore equally evident to the taxpayer concerned. Such cases constitute circumventions of State aid law by means of a manifestly discriminatory application of the law.
95. To this effect, Advocate General Pikamäe has explained that what governs the setting of ‘normal’ taxation are the rules of positive law. (38) In applying that normative framework, Advocate General Pikamäe was of the opinion that the national tax authorities need to be ascribed a margin of appreciation in, for example, transfer pricing. (39) The Court of Justice also emphasises that any fixing of the methods and criteria for determining an ‘arm’s length’ outcome falls within the discretion of the Member States. (40)
96. Individual tax assessments (whether normal tax assessments or advance tax rulings), especially if they contain findings on appropriate transfer pricing, should therefore only be reviewed on the basis of a restricted standard of review. If that review is restricted to a review of plausibility, not every incorrect application of national tax law is evidence of a selective advantage. Only the manifest derogation in favour of the taxpayer of a tax ruling (or tax assessment) from the reference system can constitute a selective advantage. In the absence of such a manifest derogation, the tax assessment may be unlawful, but a possible derogation from the reference system does not of itself mean that it constitutes State aid as provided for in Article 107(1) TFEU.
97. It follows that the Court of Justice must not examine whether the CUP method or, more correctly, the TNMM should actually have been applied. It only needs to decide – if the OECD Guidelines were the relevant reference system – whether the method applied by the Luxembourg tax authorities (the CUP method in the present case) was manifestly incorrect.
98. That cannot be said to be so in the present case. Both methods are mentioned in the OECD Transfer Pricing Guidelines. No method has a compelling priority over the other. Rather, in general, the method which is most appropriate must be chosen. (41) However, for difficult cases, where no one method would be expedient, a flexible approach can allow more than one method to be used together. Multinational corporate groups even retain the freedom to apply methods not described in those guidelines to establish transfer prices, provided those prices satisfy the arm’s length principle.
99. In recital 252 of the decision at issue, the Commission itself acknowledges that the CUP method chosen by Luxembourg is relevant in the present case. In recitals 253 and 254 of the decision at issue, the Commission describes the CUP method as a direct transfer pricing method and the TNMM as an indirect method. The Commission then uses the extensive remarks in recitals 392 to 561 of the decision at issue and the 2010 and 2017 OECD Transfer Pricing Guidelines, which at the time were not in effect, in order to argue that the relevant method was the TNMM rather than the CUP method and that the company to which it should have been applied was LuxSCS rather than LuxOpCo. Although that may be correct (since 2017), it nonetheless also demonstrates that the choice of the CUP method and the testing of LuxOpCo would not (at the time) have appeared prima facie to be incorrect.
100. The interpretation upon which Luxembourg relies in the tax rulings may therefore be incorrect, but it is by no means a manifestly erroneous interpretation of national law (or, in the present case, of the OECD Transfer Pricing Guidelines relied on by the Commission and the General Court). Ultimately, therefore, the General Court’s decision that the tax ruling conferred no selective advantage is not vitiated by error of law.
101. Consequently, the first ground of appeal put forward by the Commission is unfounded. That would also be so if, in the alternative, the reference system erroneously chosen by the Commission and erroneously considered by the General Court to be correct were to be accepted.
3. Summary regarding the first ground of appeal
102. It remains the case, therefore, that the Commission’s first ground of appeal cannot succeed from any perspective. It has emerged that the General Court judgment under appeal is correct in its outcome because the Commission relied on an erroneous reference system in the decision at issue and was therefore unable to demonstrate a selective advantage.
103. Even if the reference system erroneously relied on (the OECD Transfer Pricing Guidelines of 1995, 2010 and possibly even 2017) were taken as the starting point, the CUP method selected in the Luxembourg tax ruling would not have been manifestly the incorrect method, nor was it manifestly misapplied. For that reason, the Commission was also unable to demonstrate via the primary findings (recitals 409 to 561) of the decision at issue that the tax ruling had conferred a selective advantage in favour of Amazon.
B. The second ground of appeal: incorrect standard of proof in relation to the existence of an advantage and misapplication of the arm’s length principle
104. By its second ground of appeal, the Commission challenges the rejection, in paragraphs 314 to 538 of the judgment under appeal, of the advantage which it sought to demonstrate in the alternative and via three different routes in the decision. It contends, first, that the General Court erred in the standard of proof in relation to a finding of advantage. Second, it submits that the General Court misapplied the arm’s length principle and failed to state sufficient reasons for its ruling.
1. Findings of the General Court
105. In paragraph 308 of the judgment under appeal, the General Court held that the Commission must demonstrate that an advance tax ruling used to calculate an undertaking’s remuneration confers an economic advantage. In order to do so, it must prove that that remuneration deviates from an arm’s length outcome to such an extent that it cannot be regarded as remuneration which would have been received on the market under competitive conditions.
106. In that regard, it found (see paragraph 310 of the judgment under appeal) that although it was for the Commission to demonstrate specifically that the methodological error resulted in a reduction in the tax burden of the recipient of the advance tax ruling, the General Court did not rule out that, in some cases, a methodological error may be such as to make it impossible to arrive at an approximation of an arm’s length outcome and may necessarily lead to an undervaluation of the remuneration that would have been received under market conditions.
107. In paragraph 312 of the judgment under appeal, however, the General Court held that the Commission had confined itself to identifying errors in the transfer pricing analysis. It found that that was, in principle, insufficient to establish that there was in fact a reduction in LuxOpCo’s tax burden. That was, it held, due to the absence of a comparison in the decision at issue between the result that would have been obtained using the transfer pricing method advocated by the Commission and the result obtained pursuant to the tax ruling at issue.
108. As regards the first subsidiary finding of an economic advantage, the General Court held in paragraph 520 of the judgment under appeal that the Commission had not demonstrated the existence of an advantage, but had, at most, demonstrated the probability of the existence of an advantage.
109. The Commission – as the General Court stated in paragraph 530 of the judgment under appeal – had not succeeded in establishing that, had the profit split method on the basis of the contribution analysis been applied, LuxOpCo’s remuneration would have been greater. Accordingly, it held that the first subsidiary finding did not support the conclusion that the tax ruling at issue conferred an economic advantage on LuxOpCo. In the General Court’s view, apart from the fact that the Commission had not sought to determine what LuxOpCo’s arm’s length remuneration would have been in the light of the functions identified by the Commission in its own functional analysis, the first subsidiary finding contained no specific evidence to establish to the requisite legal standard that the errors in the functional analysis and the methodological error identified by the Commission, relating to the choice of method itself, had actually led to a reduction in LuxOpCo’s tax burden.
110. In addition, as the General Court noted with regard to the second subsidiary finding of an economic advantage (paragraph 547 of the judgment under appeal), it should be held that the Commission had not sought to ascertain that it was arm’s length remuneration or, a fortiori, whether LuxOpCo’s remuneration, endorsed by the tax ruling at issue, was lower than the remuneration that LuxOpCo would have received under arm’s length conditions.
111. Moreover, the General Court held in paragraph 585 of the judgment under appeal with regard to the third subsidiary finding that, as inappropriate as the ceiling mechanism may have been, and although it was not provided for in the 1995 version of the OECD Guidelines, the Commission had not demonstrated that that mechanism had an impact on the arm’s length nature of the royalty paid by LuxOpCo to LuxSCS.
112. Overall, the General Court concluded that the Commission had not succeeded in demonstrating the existence of an advantage through any of its three subsidiary findings (paragraphs 537, 548 and 586 of the judgment under appeal).
2. Assessment
113. The second ground of appeal is also unfounded, as the General Court’s judgment also proves to be correct on that point, albeit for different reasons. As is apparent from the explanations in point 53 et seq. of this Opinion, the decision at issue ought to be annulled because, apart from anything else, the reference system relied on had been determined incorrectly. Instead of Luxembourg law (Article 164(3) of the LIR), only the 1995 OECD Transfer Pricing Guidelines, in use in the years before the tax ruling was issued in 2003, and the 2010 and 2017 OECD Transfer Pricing Guidelines, published after that date, were applied.
114. Only in the event that the Court of Justice were to consider itself bound by the choice of the incorrect reference system, because Luxembourg and Amazon failed expressly to dispute that choice in their actions, would it be necessary to examine the second ground of appeal in more detail.
115. However, even in that situation, the Commission’s argument would have no prospect of success, since in the decision at issue – as the General Court correctly found – the Commission did not demonstrate that the tax ruling manifestly departed from the reference system and thereby conferred a selective advantage on Amazon.
116. The first two autonomous findings demonstrated no manifest infringement of the arm’s length principle under the OECD Transfer Pricing Guidelines. By contrast, the Commission pointed out – correctly, in my view – in the decision at issue (recital 575 et seq.), in the context of the third supplementary finding, that limiting the licence holder’s returns by including a ceiling of 0.55% of annual turnover was incompatible with the determination of appropriate remuneration by means of transfer pricing. Since unconnected independent entities do not usually arrange royalty payments between themselves in such a way that the licence holder never has to pay tax on more than a certain amount (0.55% of annual turnover in the present case), that element – indeed in my view manifestly – does not satisfy the arm’s length principle.
117. That is also what the General Court found in paragraph 575 et seq. of the judgment under appeal. As the General Court also correctly found (paragraph 578), that methodological error (even if manifest) was not sufficient to establish the existence of an advantage, since, even after the ceiling mechanism was applied, the remuneration remained within an arm’s length range. Consequently, the General Court held that the Commission had not succeeded in demonstrating the existence of an advantage.
118. Those remarks are not vitiated by any error of law. In accordance with settled case-law of the Court of Justice, it is for the Commission to prove the existence of ‘State aid’ as provided for in Article 107(1) TFEU and thus also to prove that the condition of granting an advantage to the beneficiaries is fulfilled. (42) That requires the Commission, in the present case, to demonstrate that Amazon received an advantage. However, the selective advantage only arises from the tax ruling conferring more favourable treatment compared to normal taxation. The Commission has to demonstrate such more favourable treatment.
119. The Commission must also carry out an overall assessment taking into account every significant element of the case in question which enables it to establish whether the recipient undertaking would not have received such relief already from the normal tax system (reference system). In the area of transfer pricing, that depends on the Commission being in a position to calculate the ‘correct’ transfer price.
120. The Commission is correct in its objection that that is very complex. However, it submits once more that not every misapplication of national tax law can simultaneously constitute a selective advantage, (43) but only manifestly erroneous findings in the tax ruling which affect the amount of tax due.
121. If the calculation of the royalty endorsed in the tax ruling would, in spite of its questionable methodology, have been lower than the usual transfer price, then there has been no more advantageous treatment of Amazon in comparison to normal taxation. After all, a higher normal transfer price would have resulted in an even lower tax burden. The General Court was correct in criticising the absence of such a comparison. In that respect, the second ground of appeal is also unfounded.
C. Conclusion
122. In conclusion, both of the Commission’s grounds of appeal are unfounded. It has emerged that the judgment under appeal is correct in its outcome – albeit for reasons other than those on which the General Court relied. Apart from anything else, the decision had to be annulled because the reference system (the OECD Transfer Pricing Guidelines instead of Luxembourg law) relied on by the Commission was incorrect.
123. Moreover, the tax ruling does not contain any manifestly incorrect recognition – too favourable to the taxpayer – of the amount of the royalty payment. Even if the inclusion of a ceiling for the licence holder’s taxable income is manifestly incompatible with the method of calculating royalty payments usual between independent entities, the Commission failed to demonstrate in the decision that that also conferred an advantage.
VI. Costs
124. In accordance with Article 184(2) of the Rules of Procedure of the Court of Justice, where the appeal is unfounded, the Court shall make a decision as to the costs.
125. Under Article 138(1) of the Rules of Procedure, which is applicable to appeal proceedings by virtue of Article 184(1) thereof, the unsuccessful party is to be ordered to pay the costs if they have been applied for in the successful party’s pleadings. As Amazon and Luxembourg did apply for costs in their pleadings, the Commission must be ordered to pay the costs incurred by Amazon and Luxembourg in the present appeal.
126. Article 184(4) of the Rules of Procedure of the Court of Justice provides that, where, without having brought the appeal itself, an intervener at first instance has participated in the written or oral part of the proceedings before the Court of Justice, the latter may decide that it is to bear its own costs. In the present case, Ireland, intervener at first instance, without being the appellant, participated in the oral procedure before the Court but has not applied for the Commission to be ordered to pay the costs. In those circumstances, it must be held that it is to bear its own costs relating to the appeal proceedings. (44)
VII. Conclusion
127. In the light of the foregoing, I propose that the Court of Justice should:
(1) Dismiss the appeal.
(2) Order the European Commission to bear its own costs and to pay the costs incurred by the Grand Duchy of Luxembourg, Amazon EU Sàrl and Amazon.com, Inc.
(3) Order Ireland to bear its own costs.