News Update State Aid & Transfer Pricing

A closer look at the General Court's annulment of the European Commission's EUR 250 million tax recovery order for Amazon
17 May 2021
17 May 2021

On 12 May 2021, the General Court (the "Court") annulled the European Commission's (the "Commission") decision finding that Luxembourg granted tax benefits to Amazon that constituted unlawful State aid.

In the Commission's view, Amazon distorted competition by paying substantially less tax than other businesses. Consequently, Luxembourg was ordered to recover the illegal aid amounting to around EUR 250 million before interest. The Amazon Group and the Grand Duchy of Luxembourg ("the Applicants") appealed the Commission's decision to the Court, which has now ruled that none of the Commission's findings were sufficient to demonstrate the existence of an advantage under State aid rules.

The judgment follows the earlier judgments in the Fiat, Starbucks, and Apple cases. In these judgments, the Court confirmed that applying State aid rules to tax rulings and the arm's length principle ("ALP") can determine the presence of a selective advantage, while at the same time, place a heavy burden of proof on the Commission.

Please see the appendix for a summary of the Court's main arguments ruled in favour and against the Commission.

The Commission's decision

In its investigation, the Commission focused on a tax ruling that Luxembourg granted to the Amazon Group on 6 November 2003 (the "tax ruling"). The tax ruling confirmed the corporate income tax ("CIT") treatment of two Amazon companies established in Luxembourg: Amazon Europe Holding Technologies SCS ("SCS") and Amazon EU Sàrl ("OpCo"). SCS is a Luxembourg limited partnership of which the partners are US entities of the Amazon group. OpCo is a wholly owned subsidiary of SCS.

From May 2006 to June 2014, SCS held certain intangibles, i.e., technology, software, marketing, and customer data (the "IP") of the Amazon Group. SCS granted an exclusive license to OpCo for the use of these intangibles, for which OpCo paid a royalty to SCS in return. The tax ruling endorsed applying the transactional net margin method ("TNMM") to determine the level of OpCo's annual royalty to be paid to SCS in return, using OpCo as the tested party. The accompanying transfer pricing report (the "2003 transfer pricing report"). substantiated this transfer pricing arrangement. The royalty payments were deducted annually from OpCo's Luxembourg profits.

The tax ruling confirmed (i) that SCS was not subject to Luxembourg CIT due to its status as limited partnership and (ii) the TNMM as the most appropriate methodology for determining the annual royalty to be paid by OpCo.

According to the Commission, the ruling endorsed a royalty which was not in line with the arm's length principle as it artificially reduced OpCo's CIT base in Luxembourg. In the Commission's view, almost three quarters of Amazon's profits were unduly attributed to SCS, where they remained untaxed.

The Court's judgment

The Applicants appealed the Commission's decision, on which appeals the Court adjudicated jointly.

In line with the Starbucks, Fiat, and Apple judgments, the Court confirmed the Commission's competence to assess the compatibility of tax rulings with the State aid rules. Moreover, the Court confirmed the ALP to be an adequate standard for determining a selective advantage if the national tax rules make no distinction between integrated companies and national companies. Two elements must be taken into account: (i) the ALP's approximative nature and (ii) a margin of error for the inaccuracies is inherent when applying the methodology.

The Court also confirmed the margin of appreciation of the Member States concerning the choice of transfer pricing method. The test applied by the Court is whether the Commission has met its heavy burden of proof in demonstrating that the "alleged" methodological errors do not result in a reliable arm's length outcome of the company's taxable income. Subsequently, the Court assessed the Commission's conclusions in two lines of reasoning, which we discuss below.

The Court agreed with the Commission that the relevant reference framework was the general framework for CIT (Law of 4 December 1967 on income tax) that does not distinguish integrated undertakings from standalone undertakings for the purposes of their obligation to pay the CIT.

Commission's primary line of reasoning
1. Choice of the tested party
Temporal relevance of the OECD Guidelines
First of all, the Commission considered the functional analysis of OpCo and SCS as set out in the 2003 transfer pricing report was incorrect and could not result in an arm's length outcome. According to the Commission's own functional analysis, the party performing unique and valuable functions was OpCo and not SCS. In its view, therefore, SCS should have been selected as the tested party instead.The Commission referred to the 1995 OECD Transfer Pricing Guidelines ("OECD Guidelines"), which states that the tested party is the one that has the less complex functional analysis. It also referred to the 2017 OECD Guidelines which indicates that a mere passive holder of intangibles cannot be considered the more complex entity. In addition, the Commission also focused, amongst other things, on the DEMPE (Development, Enhancement, Maintenance, Protection, and Exploitations functions) concept which appeared only in the 2017 OECD Guidelines 2017.

The Court ruled that the only OECD Guidelines available at the time of the adoption of the tax ruling were the 1995 OECD Guidelines. In so far as the Commission relied on the 2010 version of the OECD Guidelines, that version is not considered to be relevant, unless it merely provides a useful clarification without further elaboration. Furthermore, the Court ruled that the 2017 OECD Guidelines cannot be considered relevant in the present case, since they were published after the relevant period and in so far as the provisions contained therein have evolved significantly in relation to the 1995 version. In conclusion, the Court restricted the temporal relevance of later versions of the OECD Guidelines.

The Commission argued that all its findings are based on the 1995 OECD Guidelines and 2010 and 2017 versions were referred to only for further clarification purposes. With this argument, the Commission took the position, among others, that the DEMPE concept is not new, but clarifies the existing guidance. However, the Court dismissed this argument indicating that the DEMPE concept cannot be regarded relevant for the present case, as the concept was only introduced in the 2017 OECD Guidelines. It remains to be seen how this conclusion will impact dispute resolution in transfer pricing cases going forward as many transfer pricing practitioners, as well as the OECD, typically view the later versions of the OECD Guidelines (2010 and 2017) including the DEMPE concept as a further clarification of the existing guidance for applying the ALP.

Functional analysis
The Court considered the Commission's functional analysis of SCS to be unfounded, as it underestimated SCS' functions and the risks which it bore in relation to the intangibles. In particular, the Commission failed to take into account that SCS made available intangibles for which no comparables existed on the market and which, therefore, could be considered unique and valuable. According to the 1995 version of the OECD Guidelines, the party to which the TNMM is applied should be the party for which reliable data on the most closely comparable transactions can be identified. This will often be the party that is the least complex. In addition, the 1995 OECD Guidelines indicate that the party owning unique and valuable assets should not be used as the tested party for applying the TNMM; another party to the intercompany party should be used instead. Therefore, the Court ruled that the Luxembourg tax authorities could not be criticised for accepting that SCS was not used as the tested party.

Interestingly, the Court ruled that even if the Commission's assertion that SCS was a mere passive holder of the intangibles were to be accepted, the Commission erred in considering that SCS should have been chosen as the tested party because it failed to demonstrate that it was easier to find comparables for SCS than for OpCo.

2. Remuneration
Furthermore, the Court assessed the Commission's additional arguments relating to the remuneration for SCS if it were considered the tested party. The Commission concluded that the arm's length remuneration for SCS should have consisted of two components: (i) the re-charge of the pass-through costs it bore in relation to the licensing agreements and (ii) an arm's length mark-up of 5% on the costs of external services ("maintenance costs") related to the intangibles.

As for the first component, the Court ruled that the amount of the royalty should have reflected the market value of the intangibles made available under the licensing agreement. The mere passing on of costs corresponds only to the reimbursement of the costs incurred by SCS for the purposes of developing the intangibles and does not reflect the market value of the intangibles. Therefore, the Court considers the Commission's claim to be erroneous, as it does not correspond to a market outcome.

With regard to the second component, the Court states that the 5% mark-up on the maintenance costs corresponds to the mark-up generally observed for the provision of low value adding intra-group services. However, the Court rules that SCS did not provide such services, since the functions associated with maintaining ownership of intangibles cannot be treated in the same way as the provision of low value adding services.

In short, the Court ruled that the Commission erred in carrying outs its functional analysis of SCS and determining the appropriate net margin applicable to the intercompany transaction in the present case.

The Court's assessment of the Commission's secondary line of reasoning
The Commission argued that – even if the functional analysis of SCS had correctly been applied in the 2003 transfer pricing report – the profit allocation endorsed by the tax ruling was prone to at least three methodological errors that led to a deviation from an arm's length result.

1. The application of the TNMM as the most appropriate transfer pricing method
The Commission argued that OpCo did not exclusively perform day-to-day management functions, but that it instead made unique and valuable contributions to the intangibles. Therefore, the Commission considered that if SCS was indeed assumed to also make unique and valuable contributions, as was concluded by the Court, the profit split method on the basis of the contribution analysis should have been relied on instead.

After its remarkably thorough assessment of the functions performed, assets used, and risks assumed by OpCo, the Court concluded that the Commission's functional analysis of OpCo cannot be considered to be "entirely convincing". While the Court ruled that the Commission successfully demonstrated that OpCo's functions exceeded mere day-to-day management functions, it failed to demonstrate that all of OpCo's functions were both unique and valuable.

The Court assessed the Applicants' arguments that the Commission failed to demonstrate that OpCo's remuneration would have been higher had the profit split method indeed been used. In that respect, the Court ruled that the Commission cannot merely state that another transfer pricing method (i.e. the profit split method) would have resulted in a higher remuneration for the entity under review (i.e. OpCo). Instead, the Commission should have also quantified the impact of applying a different transfer pricing method to demonstrate the existence of an advantage, which it failed to do.

2. The choice of the profit level indicator
The Commission challenged the profit level indicators used for remunerating OpCo. It argued that a profit level indicator based on total costs (i.e., mark-up on total costs) should have been selected instead of the Berry ratio (based on operating costs). The Commission's claim was based on the mere observation that the tax ruling used the Berry ratio as the profit level indicator, whereas the 2003 transfer pricing report inconsistently used total costs as the profit level indicator. The Commission claimed that since total costs is a broader base than operating expenses, OpCo's taxable income would have been higher had total costs been relied on.

Once more, the Court ruled that the Commission failed to show that the alleged methodological error (i.e. in the choice of the profit level indicator) had not only led to a different remuneration, but also to a reduction in OpCo's tax burden.

3. The ceiling mechanism endorsed by the tax ruling
The Commission considered that the inclusion of a ceiling mechanism in the contested tax ruling, under which OpCo's remuneration could not exceed 0.55% of its annual sales, was inappropriate and conferred an advantage to OpCo which led to a decrease in its taxable income.

The Court agreed with the Commission that including a ceiling mechanism cannot be economically justified nor do the 1995 OECD Guidelines provide for such a ceiling mechanism. The Commission was therefore correct to claim that including such a ceiling mechanism constituted a methodological error. However, the Court ruled that this finding alone is not sufficient to establish the existence of an advantage. The Commission also failed to demonstrate how applying the ceiling mechanism led to a remuneration for OpCo which could be considered not to be in line with the arm's length principle.

In short, the Court concluded that merely stating that methodological errors were made is not sufficient. The Commission should have also demonstrated that the methodological errors led to an unjustifiable reduction in the taxable profit and, thus, in the tax burden borne by OpCo. Once again, the Commission's analysis seems to suffer from the same errors as were considered in its reasoning in the Apple case.


The Court's judgment in the Amazon case is a new blow for the Commission. With this judgment, the Court has given another clear signal to the Commission that it must take its burden of proof seriously and clearly demonstrate a selective tax advantage. Criticising the transfer pricing methods applied by the tax authorities is not sufficient. This aligns with the earlier judgments of the Court in the Starbucks and Apple cases. It seems that the Commission lost the cases mainly because of alleged errors in presenting a solid and convincing "counter" functional analysis. As in the Apple case, the alleged inconsistencies and errors included in the Commission's functional analysis are evident.

In cases where the Commission is not able to convincingly demonstrate that its criticism regarding the choice of the transfer pricing method applied or the specific application of the transfer pricing method clearly results in a preferential tax treatment, it becomes extremely difficult for the Commission to successfully challenge the specific choices made by the Member States in applying the selected transfer pricing method on the basis of the State aid rules. In such cases, the test whether the Commission has met its burden of proof is much stricter than in cases where the selective advantage of the tax ruling can be demonstrated based on the derogation of the reference framework, such as the Engie case in which the Court upheld the Commission's finding on the same day as its Amazon judgment. Nevertheless, the Fiat case demonstrates that it is possible for the Commission to meet the burden of proof concerning the erroneous application of the transfer pricing method.

In order to increase its chances of winning similar cases, a key takeaway for the Commission should be focusing on the following steps which are the most essential part of a transfer pricing analysis, rather than on various alleged somewhat "trivial" errors (e.g. the selection of the profit level indicator or the inclusion of ceiling mechanism).

  • Step 1 - perform a solid and convincing functional analysis;
  • Step 2 - select the most appropriate method based on the outcome of the functional analysis (and stick to your method); and
  • Step 3 - assess whether the method applied for the tax ruling deviates from the method in Step 2 in a manner that exceeds the Member States' margin of appreciation.

Only in cases where Step 3 results in the conclusion that the tax ruling is based on a manifest error in the application of the TP rules resulting in a selective advantage, the tax ruling is likely to result in unlawful State aid.
Written by:

Key Contact

Transfer Pricing | Counsel
+31 20 605 67 67
Greetje van Heezik

Key Contact

Advocaat | Counsel
+32 2 507 98 10
+32 4 7586 8593

Key Contact

Tax Lawyer | Associate
+31 20 605 61 55
+31 6 3080 4964