Market leading insight for tax experts
View online issue

BEPS: possible EU law issues

printer Mail
Speed read

In 2014, the OECD has put out a number of discussion documents regarding action 1 (digital), action 2 (hybrids), action 6 (treaty abuse/access) and action 13 (country by country reporting). In the run up to the OECD Committee on Fiscal Affairs meeting this month, this article highlights the possible EU law issues with each of the OECD proposals. Where possible, it makes suggestions as to how those issues could be addressed, so that EU/EEA members of BEPS won’t have to modify the proposed action outcome.

This article seeks to identify issues of possible concern from an EU law perspective regarding the BEPS proposals made to date.

Discussion documents have been released under action 1 (digital), action 2 (hybrids), action 6 (treaty abuse/access) and action 13 (country by country (CbC) reporting).

Whilst BEPS is not an EU initiative, 23 of the 42 original BEPS countries are members of either the EU or EEA. So if there is to be a uniformly applicable proposed solution regarding a particular action, it should be EU law compliant. Otherwise, more than half of the BEPS countries would have to modify the proposal to comply with EU law, resulting in at least two outcomes for the same action.

State aid issues

The digital discussion document mentions a number of possible options including ‘significant digital presence’, a ‘virtual permanent establishment (PE)’ and a withholding tax only on ‘digital transactions’. If these options would result in tax on a digitised sale where there would be no tax on a comparable physical sale, under EU law there may be state aid issues regarding selective measures.

In Ferring SA (C-53/00), the CJEU held that a French 2.5% sales tax imposed only on sales of medicines by French pharmaceutical laboratories constituted illegal state aid to wholesale distributors of medicines, insofar as it exceeded the costs of a regulatory regime requiring the wholesalers (but not the laboratories) to maintain a stock of medicines for delivery within 24 hours in their area.

Taken with the Gibraltar 2011 CJEU decision (C-106/09 P and C-107/09 P), this suggests that it is not always necessary to determine a ‘reference framework’ and then identify a derogation from that. This is in contrast to the ‘normal’ position set out in the Commission’s draft notice on the notion of state aid, dated 17 January 2014 (see especially paras 126–137).

The hybrids discussion documents comprise a domestic document and a treaty issues document. The former, running to 80 pages, is being rewritten down to a promised 30 pages. In this article, it is assumed that the essence of the domestic proposals will nonetheless be preserved. Specifically, the primary rule (p 18) is that for hybrid transfers, hybrid instruments and hybrid entities where there is a deduction but no taxation (D/NI), the payer must disallow the payment.

In Eurowings (C-294/97), the CJEU held (para 43) that Germany could not justify its 50% addback for trade tax purposes of lease rental payments made by a German lessee in respect of planes leased from an Irish IFSC based lessor, by reference to ‘the fact that the lessor established in another member state is there subject to lower taxation’. This referred to the former 10% IFSC rate of corporation tax, but could also be read as encompassing a wider prohibition of an EU/EEA member state adopting a rule at least as regards hybrid transfers to, hybrid instruments issued to, or hybrid entity payments made to another EU/EEA member state entity (D/NI), which is triggered by the non-taxation in the other EU/EEA member state.

In addition, Thin Cap GLO (C-524/04), as confirmed by Itelcar (C-282/12), suggests that any anti-hybrid rule must allow for commercial justification in order to be EU compliant.

In the author’s view, and having regard to the understood purpose of banking regulated capital securities (RCS) as being tier 1 capital, but bail-in-able if a capital reconstruction is necessary, the proposed exclusion of RCS from the hybrids proposals is justifiable, either on non-comparability grounds (as compared to unregulated issuers of hybrid instruments) or on public policy grounds (see Paint Graphos (C-78/08 and C-80/08)).

The proposals countering treaty abuse

The treaty abuse/access document proposes both the adoption of (basically) the current US limitation on benefits (LoB) article and a purpose-based overall anti-treaty shopping rule or ‘GAAR’.

There are arguably major EU concerns with the US LoB as proposed. This is unsurprising, as US treaty policy is developed entirely – or at least largely – without regard to the freedoms in the TFEU. Once again, if it is still a key aim of the OECD to achieve a uniform action 6 response, there needs to be an acknowledgement of the constraints which the TFEU imposes on the 23 EU/EEA countries in the original BEPS 42 countries.

Before discussing the various US LoB tests, we need to consider the D case (C-376/03), in which the CJEU held that the Netherlands was entitled to deny a German national the benefit of an annual €0.5m exemption from Dutch wealth tax on Dutch real estate, notwithstanding that the Netherlands granted that exemption to Dutch residents and also to Belgian residents via article 25(3) of the Belgian/Dutch tax treaty. In other words, in its tax treaty network, an EU (or EEA) member state does not, notwithstanding the TFEU freedoms, have to have regard to the most favoured nation (MFN) principle. On the other hand, in cases such as Saint-Gobain (C-307/97), the CJEU held that Germany (and other member states) had to extend the benefit of its tax treaty network, including treaties with third countries, to the German PE of a French company, in the same way as to a German subsidiary. What follows is therefore based on the analysis that where a treaty grants a treaty partner a benefit, that must be done in a way which complies with the TFEU freedoms. This is distinct from requiring the source EU/EEA state to grant the same benefit to a second EU/EEA country company, having already granted that benefit to a first EU/EEA country company.

Firstly, the limitation in the article X(2)(c)(i) requirement that a parent company’s shares be ‘primarily traded on ... stock exchanges located in the contracting state of which the company is a resident’ appears to be contrary to the analysis in the Royal Bank of Scotland (C-311/97). In that case, a higher corporation tax rate levied on the Greek operations of a foreign listed bank (or foreign non-banking company), as compared with a Greek bank (or other Greek company) listed on the Athens stock exchange, was held to be an unjustifiable breach of the freedom of establishment.

Secondly, the article X(2)(c)(ii) subsidiary test and the article X(2)(e) base erosion test require that ownership of the tested company can only be traced via companies resident in either treaty partner contracting state. This is arguably contrary to CJEU case law such as Papillon (C-418/07) and the SCA (C-39/13 and C-41/13) decisions on the tracing of ownership for tax grouping via any EU (or EEA) country company.

Thirdly, the article X(3) active trade or business test should, in the author’s view, allow qualifying activity to be undertaken in any EU (or EEA) member state, having regard in particular to the CJEU case law regarding R&D incentives not being consistent with TFEU freedoms if tied to activity in the granting member state (Baxter Healthcare C-254/97 and Laboratoires Fournier C-39/04).

Fourthly, the Open Skies cases (C-476/98 etc.) suggest that from an EU law perspective, a ‘derivative benefits’ test is not just a ‘nice to have’ but a ‘must’. But we need to contrast the CJEU decision in ACT Class IV (C-374/04), with regards to D (see above).

In the Open Skies Commission infringement cases, the Commission successfully challenged eight member states regarding their bilateral aviation agreements, e.g. with the US. Most of these agreements provided that if the EU national carrier was acquired as to more than 50% by non-nationals, then it lost its landing rights in the third country, e.g. at JFK. So the ‘Bermuda 2’ UK/US agreement provided that, had BA been acquired by (say) Lufthansa, BA would have lost its JFK (and other US) landing rights.

By contrast, in ACT Class IV, the CJEU held that the anti-avoidance provision in the UK/Dutch treaty, which removed the right to a half tax credit refund on dividends received by a Dutch resident company from its UK subsidiary if controlled by a company which wouldn’t have been entitled to such a refund had it held the underlying UK subsidiary shares directly, was a proportionate measure consistent with D.

Dr Tom O’Shea (Queen Mary University of London and IALS) pointed out in his book EU law and double tax conventions (pp 202, 205–216) that LoBs denying treaty benefits with regard to nationality (Open Skies) – which are in breach of EU law – have to be distinguished from LoBs with regard to residence, which may be acceptable as in ACT Class IV. However, in Dutch Ship Registration (C-299/02), the CJEU rejected restrictions with regard to both EU/EEA nationality and residence.

In the author’s view, the action 6 proposal for a US LoB without necessarily incorporating a ‘derivative benefits’ test – whereby source country treaty benefits would be granted if the company resident in the second state is controlled by, say, seven or fewer persons, ‘equivalent beneficiaries’, each of whom would enjoy equivalent benefits under their own country’s treaty with the source state – is first and foremost an LoB as opposed to a MFN situation. This is because in the absence of a derivative benefits test, the source country would be denying a company resident in the second state benefits, where the source state has granted equivalent benefits to companies in the countries of residence of the seven shareholders of the claimant company. Therefore, the source state is treating the eight residence countries comparably as regards direct investment. Accordingly, and having regard to the definition of an equivalent beneficiary being effectively a combination of nationality and residence, the Dutch Ship case suggests we may be closer to Open Skies than to ACT Class IV.

Lastly, if possible the treaty GAAR must incorporate a commercial purpose justification test to be compliant with EU law (Thin Cap GLO/Itelcar as above). The ‘in accordance with the object and purpose of the relevant provisions of this Convention’ test does not, in the author’s view, satisfy this requirement.

The CbC information proposals are unlikely to give rise to EU law issues, unless they constitute a discriminatory information burden with regard to non-nationals of a source state (Safir C-118/96).

Where will the OECD go with all of this?

The clash of US treaty policy with the TFEU requires political as well as technical movement. But, although not in the US model, equivalent beneficiaries is something that features in many of the US’s existing treaties with EU partners, and this is hopefully therefore within reach. It is to be hoped that there will be acknowledgement and accommodation of many, if not most, of the other issues too, to achieve as uniform and EU law compliant outcome as possible.