General Court on Starbucks and Fiat: confirms EU State aid control on tax rulings but places heavy burden of proof on the Commission

On 24 September, the General Court of the European Union (General Court) issued two landmark judgments following appeals against the decisions of the European Commission (Commission) in the Fiat and Starbucks cases. In these cases, the Commission had found that Luxembourg and the Netherlands had granted unlawful State aid to Fiat and Starbucks respectively, through a number of tax rulings that artificially reduced the amount both multinational groups of companies would normally pay in tax. As a result, Luxembourg and the Netherlands were ordered to recover the aid conferred on Fiat and Starbucks, consisting of EUR 20 million and EUR 30 million respectively in unpaid taxes.

With its judgments, the General Court upheld the Fiat decision, but annulled the Starbucks decision. Although the General Court confirms the Commission's competence to apply the State aid rules to tax rulings, it underlines the Commission's burden of proof with regard to the existence of an unlawful tax advantage resulting from the transfer pricing method endorsed by the tax ruling. The General Court confirms that the Commission must take into account the approximate nature of the arm's length principle (ALP) and the margin of appreciation of the Member States with regard to the choice of transfer pricing method.

In the likely event that the 'losing' parties in both procedures lodge an appeal, the final verdict of the Court of Justice of the EU can still overturn the judgments of the General Court.

The Commission decisions

The Starbucks case

The Starbucks case concerns an advance pricing agreement (APA) for the Dutch coffee roasting entity in the European Starbucks group (SMBV). SMBV buys its coffee beans from a Swiss group entity, for which it pays a remuneration. The APA establishes the taxable remuneration of the Dutch entity and provides that any profits generated in excess of that remuneration are to be paid out as a royalty to the UK-incorporated Alki LP.

Against this background, the Commission applied the ALP and found that the APA granted a selective advantage to Starbucks by lowering its taxable profit in the Netherlands in a way that is not comparable to standalone companies whose taxable profits are determined by transactions concluded on market terms.

The Fiat case

The Fiat case concerns the remunerations to be paid within the Fiat group to entities that provide intra-group financial services. In its decision, the Commission applied the ALP and found that the relevant rulings resulted in a selective advantage to Fiat. This was because the tax rulings established a taxable profit base for Fiat that was far lower compared to the taxable profit of comparable non-integrated companies in Luxembourg. The Commission decided this amounted to State aid.

The judgments of the General Court in Starbucks and Fiat

Confirmation of the general State aid approach

The General Court annulled the Commission decision in Starbucks but confirmed the Commission's approach in Fiat. Despite reaching different outcomes, both judgments confirm the fundamental principles applied by the Commission for assessing whether the tax rulings result in State aid.

First, the General Court confirms that although 'normal' taxation is defined by the national tax rules and the actual existence of an advantage must be demonstrated by reference to this, a tax measure which affects the tax base that is taken into account by the tax authorities may come within the scope of State aid law (Article 107(1) TFEU). In doing so, the General Court explicitly rejects suggestions that the Commission has, through its decisions, engaged in tax harmonisation in disguise.

Second, in line with the first point, the General Court confirms that the 'normal' national tax scheme forms the basis for the State aid assessment. If the national tax rules do not distinguish integrated undertakings from standalone undertakings for the purposes of their tax liability, the purpose of that law is to tax the profit of the integrated undertaking as though its transactions were carried out on market terms.

Third, if the national tax rules make no distinction between integrated companies and national companies, the Commission may use the ALP as an instrument to assess whether the tax ruling reflects the market conditions that apply to transactions of standalone companies. For that purpose, the Commission is also allowed to select the transfer pricing method it considers most suitable for obtaining a market-based outcome. However, the General Court points out that the Commission must explain why it chose its methodology.

Finally, the General Court confirms that national tax authorities have a margin of appreciation, but underlines that this margin of appreciation does not mean that the outcome of applying the ALP cannot be subject to State aid control. The Commission is obliged to take into account the approximate nature of the ALP, which means that the outcome may deviate to some extent from actual market conditions. If the outcome of the applied transfer pricing method leads to a taxable base that deviates more from the market conditions than the inaccuracies that are inherent to the approximate nature of the ALP, State aid control applies.

Burden of proof in Starbucks case not met by the Commission

In its Starbucks judgment, the General Court makes clear that is not sufficient for the Commission to successfully challenge the transfer pricing methods applied or to allege that they were incorrectly applied without demonstrating that this has actually resulted in a more favorable tax outcome compared to standalone companies.

In particular, the Commission found that the APA had erroneously endorsed the use of the Transactional Net Margin Method (TNMM) and instead, in its own assessment, the Commission applied the Comparable Uncontrolled Price (CUP) method. As a result of its own analysis, by applying the CUP method, the Commission considered that the amount of the royalty paid by SMBV (the taxpayer) should have been zero and SMBV paid a price for the coffee to its Swiss group entity which the Commission considered was too high.

The Commission identified methodological errors in the application of the TNMM as applied by SMBV and endorsed in the contested APA. However, the General Court held that the Commission would have had to demonstrate that these methodological errors identified under the TNMM did not allow a reliable approximation of an arm's length outcome and that they led to a reduction of the burden.

We note that in its decision, the General Court allows SMBV to apply the most appropriate transfer pricing method, i.e., the TNMM, as selected by SMBV, and does not require that this transfer pricing method be replaced or supplemented with another transfer pricing method, i.e. the CUP method as put forward by the Commission. As such, the General Court seems to adhere to paragraphs 2.2 and 2.11 of the OECD TP Guidelines (2010 version) stating that no transfer pricing method is suitable in every possible situation and the ATP does not require the application of more than one method for a given transaction.

The General Court mentioned that the OECD TP Guidelines are merely guidelines for the transfer pricing method to be applied and those lay down a strict rule on the identification of the tested party. This implies that applying a transfer pricing method to another tested party (possibly the wrong one) would still make the outcome at arm's length. However, in our view, the Commission could have strengthened its position by having referred to the OECD guidance that both sides of the intercompany transaction should be analysed, based on paragraph 3.20 of the OECD TP Guidelines (2010 version). These guidelines in summary state:

In order to select and apply the most appropriate transfer pricing method to the circumstances of the case, information is needed on the comparability factors in relation to the controlled transaction under review and in particular on the functions, assets and risks of all the parties to the controlled transaction, including the foreign associated enterprise(s). Specifically, while one-sided methods (e.g. cost plus, resale price or transactional net margin method which are discussed in detail in Chapter II) only require examining a financial indicator or profit level indicator for one of the parties to the transaction (the “tested party” as discussed in paragraphs 3.18-3.19), some information on the comparability factors of the controlled transaction and in particular on the functional analysis of the non-tested party (paragraph 3.20).

Having considered paragraph 3.20 of the OECD TP Guidelines, the Commission could have potentially considered the application of the profit split method as the most appropriate method based on paragraphs 2.4, 2.59 2.109, and 2.113 of OECD TP Guidelines (2010 version). In particular, paragraph 2.113 states that:

A further strength of the transactional profit split method is that it is less likely that either party to the controlled transaction will be left with an extreme and improbable profit result, since both parties to the transaction are evaluated. This aspect can be particularly important when analysing the contributions by the parties in respect of the intangible property employed in the controlled transactions. This two-sided approach may also be used to achieve a division of the profits from economies of scale or other joint efficiencies that satisfies both the taxpayer and tax administrations.

Paragraph 2.113 is important because while one side of the transaction dealing with SMBV is involved in sourcing beans resulting in economies of scale (SCTC in Switzerland), the other side of the transaction dealing with SMBV is considered the owner of the intangibles (Alki LP in the UK).

In summary, SMBV could reasonably have been asked by the Dutch tax authorities to also assess the functionality and remuneration of the other, foreign and non-tested parties to the transaction, Alki LP and SCTC.

Confirmation of Commission decision in Fiat case With regard to the Fiat tax rulings, the General Court was convinced by the Commission's findings that the arrangements for the application of the TNMM endorsed by the tax ruling were incorrect and resulted in an advantage for Fiat. In this respect, the General Court concluded that the Commission correctly found that the arrangements for the application of the TNMM endorsed by the tax ruling at issue were incorrect and could not result in an arm's length outcome. Although the General Court accepted the application of the TNMM in Starbucks – despite methodological errors in applying it – it did not accept this method in the Fiat case. As such, whether the TNMM is to be accepted by the Court in any future State aid decisions is something to consider and to look forward to.