Market leading insight for tax experts
View online issue

Apple: the Commission fails to overcome its evidential burden

printer Mail
Speed read
The General Court’s decision in the Apple case (Cases T-778/16 and T-892/16) shows the difficulties the European Commission faces in proving selective tax advantages that may constitute unlawful state aid. As was established in Portugal v Commission, ‘the very existence of an advantage may be established only when compared with “normal” taxation.’ In Apple, the crux of the dispute concerned the application of Irish rules on the profits properly attributable to and taxable on the Irish branches. The Commission’s decision, that a selective advantage had been granted, was not based on the actual activities of the branches. The Commission wrongly adopted the ‘exclusion approach’, attributing to the branches what it considered was not attributable to the US head office.

The European General Court’s decision in the Apple tax case (Ireland v European Commission (Apple Sales International) Cases T-778/16 and T-892/16) is a paradigm of the difficulties the European Commission faces in proving selective advantage through tax rulings that may constitute unlawful state aid. The Commission cannot rely on a general arm’s length principle; and its evidential burden is high.

Salient facts

The Commission’s decision concerned two Irish subsidiaries (ASI and AOE) of Apple Inc managed and controlled in the US, so their tax residence at the material times was not in Ireland. As Irish corporations, they were not US residents either. Each subsidiary had a branch in Ireland. The essential issue concerns the profits properly attributable to those branches and whether they were undertaxed selectively, giving rise to unlawful state aid.

The subsidiaries and Apple Inc agreed to share costs and risks relating to R&D concerning Apple’s products; a royalty free licence for Apple’s IP, permitting the manufacture and sale of Apple products throughout the world, except North and South America; and for Apple to provide marketing services to ASI for a fee based on Apple’s costs plus a mark-up.

The two Irish branches were:

  • ASI – responsible for procurement (of Apple branded finished products from manufacturers), sales and distribution activities to customers outside the Americas and providing aftersales support. Importantly, however, the contracts with manufacturers of, and customers for a large proportion of products sold by ASI were negotiated and signed by Apple Inc, or the directors of ASI, outside Ireland. ASI’s branch was either not involved in the critical decision making at all or had subsidiary functions, such as a monitoring role; and
  • AOE – responsible for manufacture and assembly of specialised ranges of computers such as iMac desktops and MacBook laptops, which were supplied to related parties for the EMEIA region.

A core feature of the rulings given by the Irish Tax Authority (ITA) in 1991 and 2007 was that the subsidiaries’ taxable profits from the Irish branches were based on specified proportions of the branches’ operating costs, rather than sales.

In essence, the Commission argued that the ITA’s failure to allocate profits to the branches, above those the ITA had accepted in the rulings, on the basis of the licences granted to the subsidiaries amounted to a selective advantage granted to Apple and hence state aid. Following the court’s ruling, executive vice-president Margrethe Vestager stated: ‘As a result of the rulings, in 2011, for example, Apple’s Irish subsidiary recorded European profits of US$22bn (c.a. €16bn) but under the terms of the tax ruling only around €50m were considered taxable in Ireland’ (emphasis added). (Note that, in relation to that example, although the balance of profits was not taxed in the US in 2011, it was taxable on a deferred basis; and it is understood it was taxed in the US following a change of US law in 2017.)

EU state aid rules that override member states’ fiscal sovereignty

The Commission claimed that the tax rulings were incompatible with article 107(3)(c) of the TFEU and it had not been notified under article 108(3), such that they constituted unlawful state aid.

It is well established that although direct tax is not harmonised in the EU, tax measures emanating from the state may constitute state aid where, in particular, they grant an undertaking (i.e. the subsidiaries) a selective advantage, either directly or indirectly. The burden of proof is on the Commission.

As was found in Portugal v Commission (Case C‑88/03), ‘the very existence of an advantage may be established only when compared with “normal” taxation’. An economic advantage arises ‘where it mitigates the burdens normally included in the budget of an undertaking’. Even though it is not strictly a subsidy, it is ‘similar in character and has the same effect’.

But what is normal taxation, i.e. what is the appropriate comparator? And when is an advantage selective? The Commission essentially concluded that:

  • Ireland had renounced tax revenue, whereby an advantage was conferred on the subsidiaries.
  • As the subsidiaries operated in all EU member states, the tax rulings were liable to affect trade in the EU.
  • The rulings having been given only to Apple subsidiaries were selective in nature, especially in comparison to other companies in a comparable situation. Neither Ireland nor Apple argued that the selective advantage was justified.
  • The rulings derogated from the ordinary rules of corporation tax in Ireland, those rules being ‘the reference framework’ and benchmark of ‘normal taxation’. That framework was said to provide grounds for applying the arm’s length principle. The Commission further claimed that article 107(1) required the arm’s length principle to be applied, even if it was not incorporated in the reference framework.

The normal taxation benchmark

Under the Irish tax system, resident companies pay tax on their worldwide profits (excluding domestic distributions) and capital gains. Non-resident companies only pay Irish tax if they carry on a trade in Ireland through a branch or agency (TCA 1997 s 25). Branches are taxed on all of their trading income arising directly or indirectly through them, income from property or rights used by or held by or for them, and capital gains attributable to them. That was broadly in line with the regimes of most OECD countries. The objective of that regime was to tax profits from activities in Ireland, whether carried on by resident or non-resident companies, ‘integrated or stand-alone’.

As the Commission had accepted that the ordinary rules for taxing corporate profits applied in Ireland formed the reference framework, the court held it did not err in its selection of the reference framework.

The crux of the dispute concerned the application of s 25 and the profits properly attributable to the branches and taxable in Ireland. The Commission’s approach perhaps unduly focused on the arm’s length principle, rather than the meaning and effect of s 25. In particular, there did not appear to be sufficient focus, consequently evidence collected, on the activities actually carried on through the branches.

The arms’ length principle

The court held that in the absence of EU rules, the Commission could not claim that there was a freestanding obligation to apply the arm’s length principle (namely the extant conditions between independent market operators) arising from article 107. Corporate taxes not being harmonised in the EU, the Commission did not ‘have the power independently to determine what constitutes “normal taxation” of an integrated undertaking while disregarding the national rules of taxation’.

However, the court accepted that article 107 gives the Commission the right to check whether the level of profits allocated to branches under national rules ‘corresponds to the level of profits that would have been obtained if the activity of the branches had been carried out under market conditions’ where the national rules essentially make provision for that.

There was an uncontested opinion, based on Irish case law, that the application of s 25 required assessment of the actual activities carried on by the branches and the value of those activities. Ireland had confirmed to the court that ‘the value of activities actually carried out by the branches is to be determined according to the value of that type of activity on the market’. Although s 25 did not expressly incorporate an arm’s length principle, that principle was part of Irish law and involved any necessary adjustments in line with the OECD’s transfer pricing guidelines. Therefore, if the value of a branch’s activity under the ruling was 80 but the arm’s length value was 100, it could have been possible to show an advantage conferred by the rulings.

Each branch being part of the same legal entity, the subsidiary, the Commission had relied on the authorised OECD approach. That approach reflects an international consensus regarding profit allocation to permanent establishments. Although Ireland had not adopted the authorised OECD approach, it confirmed that the valuation of the branch activities was based on objective analysis of the facts. The court found that there was considerable overlap between the application of s 25 and the authorised OECD approach, such that they were essentially the same, so the Commission could not be criticised for its reliance on the latter.

Where did the Commission go wrong?

In essence, the Commission made errors in applying s 25, the arm’s length principle and the authorised OECD approach in its primary line of reasoning.

In order to apply s 25 and the authorised OECD approach, the Commission had to identify the assets, functions and risks allocated to the branches and the functions actually performed by them; and those activities had to be valued. The Commission failed to base its decision on the actual activities of the branches. Instead, it had wrongly adopted the ‘exclusion approach’, which was inconsistent with s 25. The Commission identified the functions performed by the subsidiaries as a whole and presumed they had been performed by the branches when they could not, in the Commission’s view, be allocated to the head office of the subsidiaries.

The Commission allocated the profits from the intellectual property (IP) to the branches on the basis that the subsidiaries had no employees capable of managing the IP, without establishing whether the branches had actually managed the IP. Functions and risks related to the IP had been identified. Those functions and risk were at the ‘heart of the Apple Groups’ business model’. The Commission failed to establish, with supporting evidence, that the branch managers actively managed those functions or risks; this was fatal. By contrast, Apple’s evidence indicated that personnel around the world and in the US were involved in managing those functions and risks outside Ireland and the branches essentially provided support for the R&D function. The ‘centre of gravity’ of the Apple Group was in the US; and the court found that the strategic decisions, in particular concerning the development of the Apple IP, appeared to be in the US. Further, the subsidiaries only had one director based in Ireland.

The same deficiencies in the Commission’s approach were found in relation to ASI’s procurement activities.

In relations to ASI’s sales activities, Apple’s evidence that the branches did not have marketing staff was unchallenged. Sales and distribution activities undertaken were, according to unchallenged reports produced for Apple and Ireland, routine functions performed in accordance with instructions from executives in the US. Although ASI’s after sales support activities were in Ireland and important to the Apple brand, they did not warrant the IP being allocated to the branches.

Although it was accepted by Apple that AOE’s manufacturing activities in Ireland involved processes and expertise which provided some benefit for Apple’s IP, they did not justify allocating Apple’s IP to the branches especially as the contested tax rulings took account of the contribution made by AOE’s branch to the Apple IP.

The Commission thus failed to show that Ireland had granted Apple an advantage for the purpose of article 107.

The Commission’s secondary approach: its challenge to the profit allocation method

The Commission’s second line of attack was that the method adopted in the rulings undervalued the branch profits. The method was said to be one sided, rather than testing profits levels achievable by parties at arm’s length; it was based on costs, rather than income; and the level of returns it produced were too low. The court disagreed with some of those criticisms (in particular, rejecting the Commission’s view that choice of the wrong ‘tested party’ resulted in an advantage). It accepted that there had been some ‘methodological errors’ (basing the ruling on ‘very concise’ and ‘fairly vague’ descriptions of the functions of the Irish branches), but they were not sufficient on their own to show an advantage had been conferred. The Commission also had to show that those errors led to a reduction in the chargeable profits and the tax burden on the branches, against the burden they would have borne under the normal tax rules.

This aspect of the court’s decision goes into some detail of the applicable OECD transfer pricing rules. Essentially:

  • A one-sided profit allocation method involves selecting an appropriate base (costs, sales or assets), an appropriate operator (the tested party) and assessing whether the impugned undertaking obtains an advantage in comparison to the profits the tested party would obtain from comparable transactions in the free market. The comparative analysis must still be by reference to the functions undertaken by the impugned undertaking. The court disagreed with the Commission that mere selection of the branch (which, it pointed out, need not under the OECD guidelines necessarily have been the least complex entity, but must be a reliable comparator) as the tested party had led to a decrease in the subsidiaries’ taxable profits. Moreover, the Commission failed to undertake the necessary comparative evidence-based analysis on the functions performed by the branches.
  • Both sales and operating costs could be an appropriate profit level indicator under the OECD guidelines, which the Commission had not itself ruled out. Once again, the functions performed by the branches had to be identified, with their associated risks and assets, and those functions had to be valued. The choice of profit level indicator was not fixed for any type of function, provided the indicator was relevant and reflected the value of the function in question. Again, the Commission had relied on the exclusion approach. Its approach did not demonstrate that the sales were more appropriate for valuing the functions actually performed by the branches. The Commission, essentially relied on assertions, without showing that the value of operating costs taking into account in the rulings were not proportional to the value of the functions performed by the branches, or that the costs taken into account were too low. Although inconsistencies were identified in the approach adopted by Apple and ITA, methodological errors were again trumped by the Commission’s failure to show that those errors had actually led to a reduction in the tax burden of the branches in comparison to the tax burden they would have had under the normal tax rules.
  • Although Apple was criticised for not having provided explanations to justify the level of returns adopted in the rulings, another methodological error, the Commission’s alternative analysis was also criticised for failing to demonstrate that the tax actually paid by the branches was less than that which should have been paid under normal tax rules. Apple and ITA had submitted reports, albeit, as ex post evidence, that the profits allocated to the branches were within arm’s length ranges. The court found that even if the conclusions in the reports were invalid, it could not be inferred from this that the rulings had led to reduction in the subsidiaries’ tax liability in Ireland and that did not alter the burden of proof on the Commission to show that an advantage had been conferred by the rulings. The Commission also failed in challenging the reliability of the reports; and the Commission’s own conclusions essentially confirmed the conclusions that the profits fell within arm’s length ranges.

Under the rulings, the proportion of costs adopted for determining the taxable profits decreased above specified thresholds. This discretionary feature of the rulings could have supported findings of the grant of selective advantage. However, the threshold adopted was never reached. No advantage could be found from a part of the rulings that did not engage.

Where does this leave us?

The Commission’s failure to show, by reference to evidence, that the Apple subsidiaries had obtained a tax advantage in Ireland makes it difficult to identify a basis for appeal to the CJEU. Although a self-standing arm’s length principle was found not to exist, that did not prove to be the critical point on which the Commission lost, which would appear to prevent any appeal on that basis alone. However, given that the court found that Apple and ITA made a number of methodological errors, it remains to be seen whether the Commission bases an appeal on the court’s holding that such errors were not sufficient to assist the Commission (though it would seem hard to appeal the General Court’s finding that the Commission had not shown that those errors on their own resulted in a tax advantage).

Practitioners should note that the Commission’s opening decision (explaining why it was opening an investigation) quoted at length from an Irish government note of a discussion with Apple in 1990 in which Apple is said to have ‘mentioned by way of background information that Apple was now the largest employer in the Cork area with 1,000 direct employees and 500 persons engaged on a sub-contract basis’ before moving on to point out that it was ‘reviewing its worldwide operations and wishe[d] to establish a profit margin on its Irish operations’ before putting forward a profit figure for which Apple’s representative ‘confessed that there was no scientific basis’. The Commission failed to demonstrate that Apple was in fact granted a selective advantage, but one can see why that note might well have given rise to serious suspicions on its part that Apple had (in blunt terms) obtained an unprincipled tax advantage by pointing to the jobs that depended on it. Those dealing with EU tax authorities on complex matters need to bear in mind that references to the positive economic consequences of a favourable ruling can seriously backfire. Further, as pointed out above, the General Court noted that the original material on which the tax ruling was based was seriously deficient: although the Commission was in fact unable to show that those deficiencies resulted in an advantage in this case, there will be other cases where that can be shown without too much difficulty. It is therefore important that taxpayers ensure that favourable tax rulings are grounded on solid factual material and legal argument that addresses the tax issues, and that there is no risk of any perception that the ruling has been influenced by extraneous considerations.

Although this case concerns state aid, tax practitioners have much to gain from reading it about the rules on the taxation of branches and permanent establishments.