|
Erica Howard |
|
Thijs Brans |
|
Barry Larking |
The risk that a company could constitute a taxable permanent establishment (PE) of another member of its group is not new. In fact the possibility has been recognized in versions of the OECD Model Tax Convention going back to the early 1940s. This issue has also been the subject of sporadic judicial rulings, culminating most recently in the controversial decision of the Italian Supreme Court in the Philip Morris case in May 2002 (Corte Suprema di Cassazione, No 7682/02). That case called into question a number of traditionally held views as to when a group company providing internal management services can constitute a PE. As explained below, the OECD has since taken action to address this matter. Perhaps the reason why there have not been more such cases is the prevalence of the assumption that, even if a PE were to exist, there would be no net profit to attribute to it, and therefore little point in arguing that a PE did exist.
However, the message contained in the OECD's latest report on global trading is that this assumption has a limited lifespan. (Discussion draft on the attribution of profits to permanent establishments (PEs): Part III (enterprises carrying on global trading of financial instruments) (Discussion Draft) issued on March 4 2003).
This article examines firstly the technical basis for the assertion that global trading structures using group companies can give rise to a PE, and secondly the issues that arise when considering allocating a residual profit to such a PE.
What the Discussion Draft says
In its initial analysis of the global trading business environment, the Discussion Draft points out (in paragraph 25) that a variety of legal structures and forms are used, with some businesses trading "exclusively through PEs, others through separate legal entities (which may act in their own right or as dependent agents of other entities), whilst others use a combination of PEs and separate legal entities". The significance of the words in parentheses does not become clear until the end of the 56-page document has nearly been reached.
The Discussion Draft explains there that, in cases where a PE arises from the activities of a dependent agent, "the host jurisdiction will have taxing rights over two different legal entities - the dependent agent enterprise (which is a resident of the PE jurisdiction) and the dependent agent PE (which is a PE of a non-resident enterprise)" (see paragraph 257). The Discussion Draft does not go into the technicalities of when a PE can arise from the activities of a dependent agent (or, for that matter, what constitutes a dependent agent for these purposes), but merely asserts that it is "quite common for functions associated with a global trading business to be undertaken by dependent agents within the meaning of article 5(5) of the OECD Model Tax Convention (a dependent agent PE)". The first part of this article considers the technical basis for this assertion.
However, perhaps the most important message comes in the penultimate paragraph of the OECD document where, after premising that the dependent agent is likely to be remunerated by way of a service fee, the OECD concludes: "In short, when attributing profits to the dependent agent PE, there are likely to be profits (or losses) over and above the arm's length service fee paid to the dependent agent enterprise". The issues surrounding this view are discussed in the second part of this article.
Is there a permanent establishment?
Under most tax treaties a PE can only be constituted in one of two possible ways. The first, by way of a fixed place of business through which the business of the non-resident enterprise is carried on, and the second - a deemed permanent establishment - through a dependent agent of the non-resident enterprise. From the comments made in the Discussion Draft, it seems that the main risk in the global trading context lies with the dependent agent PE.
Given their current interest in the possibilities of construing PEs in intra-group service situations, it is interesting that the OECD have recently produced a draft discussion paper on the group company PE issue entitled Proposed clarification of the permanent establishment definition (issued on April 12 2004). In general, this may be seen as a reaction to the general debate engendered by the previously mentioned Philip Morris decision of the Italian Supreme Court, which does little more than put the position back to where everyone thought it was before that decision, that is, that a subsidiary (or indeed any group company) may in principle constitute a PE of another member of the group, but only under the same conditions as apply to any non-related entity (it is possible for a third party to create a dependent agent PE of a non-related entity). Membership of a group is not, as such, relevant to the question.
To come within the standard OECD Model Tax Convention definition of a dependent agent PE, the following basic conditions would have to be satisfied: an agent (which can include a company) that is not "independent", is acting in the ordinary course of its business, habitually exercising an authority to conclude contracts in the name of the foreign principal, and those activities go beyond mere preparatory and auxiliary activities. If a person qualifies as a PE, it should not be overlooked that this status will apply "in respect of any activities which that person undertakes for the [non-resident] enterprise", that is, not just as regards the contracting activities.
It will be clear that before a global trading arrangement involving an associated enterprise will lead to the creation of a PE, a number of tests have to be satisfied. Whether or not they are satisfied will depend on how they are interpreted by the taxing state in the light of the specific facts and circumstances, and the weight given to the guidance in the Commentaries to article 5 of the OECD Model.
That the facts and circumstances are likely to vary from case to case seems clear from the -as the OECD itself observes - "almost limitless number of different business structures that firms engaged in global trading can employ". However, the problems seem likely to be the greatest under the "Centralized Product Management" model, described by the OECD in its Discussion Draft, that envisages a transfer of market risk from the originating (or marketing) location to the central trading location. Whether or not this gives rise to a dependent agent PE depends on what functions are performed by the marketing and trading operations, and where the credit and market risk management is carried out.
Although the Discussion Draft provides some guidance as to what is understood by the terms "marketing" and "trading", there is still ambiguity. In practice, identifying the functions performed by each part of the organization will require case-by-case analysis. However, the Discussion Draft does indicate a number of possible marketing activities that can be relevant for PE status. These include, in particular, negotiating the terms of the transaction (including the price based on the parameters set by the trader), obtaining internal clearances, and concluding the transaction.
A thorough analysis of the treaty conditions for dependent agent PEs would go beyond the scope of this article. However, activities along the lines of the above would seem to have at least some of the ingredients needed for dependent agent PE status. In this respect, it might be noted that the current OECD interpretation of concluding contracts in the name of the foreign principal (as expressed in the Commentaries to the Model Tax Convention) is that this covers conclusion of contracts that are binding on the foreign principal. The previously mentioned discussion paper issued on April 12 2004 also comments on the "contracting authority" issue, and indicates that a formal power of representation is not required. However, the main area of discussion in the context of global trading is likely to be the independence question. The control exercised over the "marketing" location is likely to be a major factor here, including possibly the way in which credit limits are set and managed. A relevant consideration in this regard is that under the OECD's current views as expressed in the Commentary to the Model Tax Convention, limitations as to the scale of business which may be conducted by the agent are not relevant to dependency: the latter is determined by the extent to which the agent exercises freedom in the conduct of business on behalf of the principal within the scope of the authority conferred. Another factor of relevance is likely to be the extent to which the marketing location is dedicated to serving the central trading location, that is, the absence of other principals.
It will be apparent from the above that much will depend on the particular facts and circumstances, as well as the interpretation given to the above provisions by the relevant taxing authority. However, the risk clearly exists that tax authorities will consider they have grounds for arguing the existence of a dependent agent PE.
What profits should be allocated?
As already mentioned, the consequence of dependent agent PE status, as the OECD sees it, is that the country where the agent (or, in terms of the above global trading example, the marketing operation) is based will be able to tax not just the arm's length remuneration it receives, but also an additional amount of profit attributable to the PE of the non-resident principal (or central trading operation).
An alternative view is that no additional profit can be attributed to the PE since the gross income attributable to it is exactly matched by the - deductible - agent's fee. This view is based on the assumption that, in effect, the functions, assets and risks assumed by the PE are identical to those assumed by the agent (these factors, traditionally regarded by the OECD as key to determining arm's length profits, remain central to the Discussion Draft analysis). The OECD's approach appears to be that these factors are not identical. In particular, the Discussion Draft argues that in the dependent agent PE situation, the risks arising from the transaction arise in the non-resident enterprise (and not the agent) and, to the extent they arise from the functions performed by the dependent agent or by the non-resident enterprise in the PE jurisdiction, those risks and the capital needed to support them should be attributed to the PE. In the OECD's view those risks and capital need to be remunerated on an arm's length basis with the result that the PE would show a net profit over and above the fee paid to the agent.
While this analysis is interesting, it leaves unresolved a major question. That is why, in the case of an independent agent carrying out the marketing function (and hence not creating a dependent agent PE), the host state would accept a total tax base equal to the agency fee but, just because of the loss of independence, would require the agency fee plus a residual PE profit. This would make sense if the loss of independence were attributable to a transfer of risk and capital from the agent to the non-resident principal sufficient to justify the additional profit allocation. Although risk can be a relevant factor in determining dependence or independence (see OECD Commentary to article 5, paragraph 38) this is unlikely to be the cause of the loss of independence in the global trading context. After all, even in the independent agent case, where there is a centralized product management model, the market and credit risk will still be borne by the central trading location and not by the agent. If anything, the result of the loss of independence is likely to lead to a decrease in the arm's length fee (given that less independence is likely to involve more limited business activity for the agent). While this might conceivably lead to some shifting of risk to the principal, this cannot involve a shifting of market or credit risk (nor therefore a significant shifting of profits) since these risks would already lie with the principal.
One explanation for the OECD's approach is that since the PE is the product of a legal fiction (the dependent agent is merely deemed to be a PE), the necessary consequence of triggering this legal fiction (by the loss of independence) is that functions, assets, and risks should be deemed to be attributed to the PE, as if the agent did not exist as a separate enterprise. The arm's length fee for the agent is, however, then deducted from the income attributed to the dependent agent PE in arriving at its assessable profits/losses.
The OECD's advocated approach to dependent agent PEs raises significant practical issues of concern. In the first place, it will require global trading businesses to track down "accidental" PEs in numerous jurisdictions and make sure the appropriate compliance procedures are followed or run the risks of interest and penalties for failing to do so (this problem arises irrespective of whether additional profit is, or is not, attributable to the PE). A flexible approach by tax administrations to this problem by permitting a single return to be filed on behalf of both the agent and the PE would perhaps go some way to alleviating this problem.
Secondly, by introducing PEs into the tax picture, the scope for dispute between the different taxing jurisdictions is multiplied. This is readily apparent from the problems that traditionally arise between home and host state in computing the tax base, where PEs are involved. From the clear absence of consensus as regards capital allocation methods in the OECD's most recent report on profit allocation to banking PEs (Discussion draft on the attribution of profits to permanent establishments (PEs): Part II (banks), reissued on March 4 2003), it seems that these problems are more likely to increase rather than decrease.
Use of OECD analysis
While there is technically nothing new in treaty terms in respect of the above analysis, it seems that the OECD's views as expressed in the Discussion Draft may be used by tax authorities to challenge existing global trading structures and, indeed, other structures involving limited risk associates. However, as the OECD has not yet finished its deliberations on this issue, there still may be time for business and practitioner community comments to be incorporated into the draft. While it may go beyond the OECD's authority, a recommendation to OECD member states of a flexible compliance approach (for example, to avoid double filings) would be welcome. A more radical approach would be for OECD member states to agree to limit the dependent agent PE approach to non-treaty situations, so providing some practical comfort to taxpayers, while at the same time addressing member states' concern to protect their revenues. We await with interest the next release of the OECD's Discussion Draft.
For further information please contact:
Erica Howard
KPMG LLP (UK)
Tel: +44 20 7311 5259
Email: erica.howard@kpmg.co.uk
Thijs Brans
KPMG Meijburg & Co. (The Netherlands)
Tel: +31 20 656 1353
Email: Brans.Thijs@kpmg.nl
Barry Larking
KPMG Meijburg & Co. (The Netherlands)
Tel: +31 20 656 1465
Email: larking.barry@kpmg.nl
The views and opinions are those of the author and do not necessarily represent the views and opinions of KPMG LLP (UK) & KPMG Meijburg & Co. All information provided is of a general nature and is not intended to address the circumstances of any particular.
