This paper sets out our comments on the Public Discussion Draft published on 31 October 2014.


  1. The Discussion Draft on artificial avoidance of permanent establishment (PE) status deals with four main areas where there are concerns that a business operating across borders can avoid becoming subject to tax outside the jurisdiction where it is resident by structuring its business to prevent the creation of a taxable presence:
  • commissionaire and similar arrangements;
  • exemptions for specific activities (such as preparatory or auxiliary activities);
  • fragmentation of contracts; and
  • insurance.
  1. It must be noted at the outset that business may, for legitimate business reasons, choose to adopt structures that may be potentially affected by the Action 7 proposals. For example, commissionaire structures are not solely used for tax avoidance, but are a tried and tested means for introducing products to new and wider markets. Similarly, groups of companies often establish separate entities to conduct separate business functions for good commercial reasons, and activities do not cease  to be “preparatory” or “auxiliary” simply because they are conducted by a related party.  The proposal to deny the specified activities exemption for related parties implies that a “cohesive operating business” should be taxed as a single entity – that is, it seems, in essence, a move towards a form of unitary taxation.
  2. Also, coordination is required with other BEPS Actions that cover similar ground. Artificial structures that attempt to rely on double tax treaties can be attacked under anti-avoidance rules (for example, the principal purpose test proposed in Action 6); and allocating income and profits between related parties should be undertaken through correct transfer pricing (within Actions 8 to 10), rather than treating a related party as a PE when a third party would not.
  3. Finally, any changes should not lead to additional uncertainty. If implemented, several proposals in the Discussion Draft are likely to lead to areas of greater subjective interpretation and potential for disputes. In particular, widening the tests for a PE are also likely to lead to more situations when an apportionment of profits is required, yet the Discussion Draft indicates that no changes are needed to the existing rules for attributing profits to a PE, but those rules themselves contain subjective elements that can be difficult to apply in practice.


Commissionaire and similar arrangements

  1. It has long been possible to do business in a jurisdiction without a physical presence there, and without the presence of a dependent agent – that is, without a taxable presence in traditional terms. New business structures, such as internet commerce, are in many ways little different to traditional forms of business, such as mail order.
  2. Under commissionaire arrangements, a third party (the commissionaire) is appointed in a jurisdiction to sell products on behalf of a business that is not otherwise established there (a manufacturer, say).  Typically, the commissionaire acts in its own name as regards customers, but bears little risk in relation to the goods, which continue to be owned by the manufacturer and will be returned to the manufacturer if they are not sold. The commissionaire is usually remunerated by way of  a commission on sales, reflecting the relatively limited commercial risks it bears.
  3. The apparent concern with commissionaire structures is that the commissionaire is usually not a dependent agent of the manufacturer (and in particular does not conclude contracts “in the name of” the manufacturer), and the commissionaire does not own the goods.  As a result, only the small commission is taxed where the sale is made (and where the customer typically resides), and none of the (relatively larger) profit made by the manufacturer is taxed there.  However, it should be noted that the manufacturer would usually be subject to tax in full on its profits in the jurisdiction where it is established.
  4. The concept of a commissionaire exists in many civil law countries.  It is often not well understood in common law countries, where a commissionaire is usually interpreted as a form of undisclosed agent.  For example, a UK “commissionaire”, selling in its own name on behalf of a third party principal, would typically be regarded as an agent of the principal, concluding contracts that bind the principal. A commissionaire in the UK that is not independent – that is, it is not dealing with its principal on an arm’s length basis – could then be treated as a PE of the principal in the UK.
  5. The Discussion Draft suggests that such structures  are “artificial”. However, there may be good, legitimate business reasons to appoint a commissionaire, rather than a manufacturer choosing to set up its own sales and distribution business or to appoint a full sale agent. A local intermediary may understand the market better, have better local contacts, and provide some protection from local legal liabilities, but the commissionaire may prefer not to expose itself to the full commercial risk of buying and reselling the product.
  6. In principle, there is no reason why a corporate group may not choose to carry on a commissionaire activity “in house” in a dedicated subsidiary. Determining income and profits on transactions between related parties should be undertaken through correct transfer pricing, rather than by treating a related party as a PE when a third party would not. The activities of related parties interacting on a proper arm’s length commercial basis should not be aggregated or attributed to each other. If the arrangements are not on arm’s length terms, that is a mater for transfer pricing.


  1. Page 11 to 14 of the Discussion Draft contains four alternative but somewhat similar proposals, A to D.
  2. Proposal A seeks to replace the current and clear “conclude[s] contracts” test with a vaguer one of “engag[ing] with specific persons in a way that results in the conclusion of contracts”.
  3. Proposal B is similar to proposal A, but avoids the vagueness of the Proposal A test of “engagement resulting in conclusion of contracts” by replacing it with a test of negotiating material terms. However, there is no guidance on what elements of a contract should be considered “material”.
  4. Proposal C is similar to proposal A, but extends the concept of concluding contracts to include contracts which are not in the name of the principal, but which are for the account of and at the risk of the principal. Unlike Proposal B, it retains the uncertainties inherent in the test of “engagement resulting in conclusion of contracts”.
  5. Finally, Proposal D is similar to Proposal B – negotiation of material terms, not a causal “resulting in” test – but brings in the element of proposal C on risk and reward.


  1. In Proposals A and C, it remains unclear how to establish a causal connection between the activities of an intermediary and the conclusion of contracts.  Is it simply sufficient that the customer was introduced by the intermediary, with all negotiation and the final decision to conclude the contract occurring outside of the jurisdiction? As proposed, it seems that pure marketing or referral arrangements could be caught – is that the intention? For that reason, we prefer the “material terms” formulation in Proposals B and D, but with modifications.
  2. Under all four proposals, an intermediary would be deemed to be a PE, unless similar activities conducted through a fixed establishment of the principal would not create a PE. This proposal begs the question of whether there is an establishment in the first place, and essentially treats an intermediary as a fixed establishment of the manufacturer.
  3. All four proposals also include an exclusion for an intermediary acting independently in the ordinary course  of its business of representing several principals, which will continue to be treated as not creating a PE.  The exclusion does not apply where a commissionaire only acts for one customer, even if exclusivity is agreed as a commercial matter between wholly independent parties.  Is it really the case that a sales distributor ceases to be independent if is only sells goods for one principal?
  4. In our view, only one significant change is required, to deal the technicality of whether contracts are “in the name of” an enterprise.  We would suggest adapting part of Proposals C and D, replacing the phrase “conclude contracts in the name of an enterprise” with the phrase “conclude contracts in the name of an enterprise or which, by virtue of the legal relationship between that person  and the enterprise, are for the account and risk of the enterprise”.
  5. We do not see a need for an intermediary to be deemed  to be a fixed establishment, nor for the causal test in Proposals A and C. In particular, any element of artificiality can be dealt with by other means, such as anti-avoidance rules, and provisions that are not on arm’s length terms should be dealt with through transfer pricing.


  1. The second area considered in the Discussion Draft is the exemption for specified activities, which allows taxpayers to conduct some limited activities in a jurisdiction to facilitate its business, without being exposed to tax there on a portion of its overall profit, on the basis that the limited activities have a relatively low value. However, there is a concern that some activities traditionally treated as “preparatory” or “auxiliary” could be significant components of some businesses.
  2. Proposal E on page 15 will make sure that all activities – including storage, display or delivery of goods, or a purchasing or information gathering function – can only benefit from the exemption if they truly are “preparatory” or “auxiliary”. We have no objection to this suggestion, which should limit the exemptions to their intended targets.
  3. Proposals F, G and H are alternatives to Proposal E, and would see targeted changes to the specified activities list to remove delivery, and/or purchasing goods, and/or collection of information.  Provided that such activities are actually “preparatory” or “auxiliary”, we see no reason to remove them from the list of specified activities, which is why we prefer Proposal E.
  4. To further consider the splitting of a single “cohesive business operation” in between separate related entities. As is noted on page 19, the commentary on Article 5 already considers separate fixed establishments of the same entity (which must be considered separately, and not aggregated, if they are separated locally and organisationally).
  5. Proposal I would prevent an enterprise relying on the specified activities exemption if it or an associated enterprise already has PE in that jurisdiction, and the activities are “complementary functions that are part of a cohesive  business operation”.
  6. Proposal J is similar to Proposal I, but also applies where neither entity already has a PE but the overall activity resulting from aggregating the activities of the related enterprises is not of a preparatory or auxiliary character.
  7. It is one thing to aggregate the activities of a single taxable person, but quite another to aggregate the activities of separate entities. The proposal to deny the specified activities exemption for related parties implies that a “cohesive business operation” should be taxed as a single entity – that is, it seems, in essence, a move towards a form of unitary taxation. The proposals contain no clear guidance on when separate activities, conducted by separate companies on an arm’s length basis, constitute a “cohesive business operation”. We would suggest that these concerns are best dealt with through anti-avoidance rules and transfer pricing, not by amending the definition of a PE.


  1. The third area considered in the Discussion Draft is the splitting of the activities of contractors, etc., between related companies, so that each entity does not conduct the activity for more than 12 months. This would appear to be an area that could be covered by the proposals on treaty abuse in Action 6. For that reason, we prefer Proposal L on page 23, instead of adding the separate targeted anti-avoidance rule suggested in Proposal K.


  1. Finally, there seems to be a concern that insurance companies can engage an independent agent to collect premiums from a jurisdiction, or cover risks in a jurisdiction, without the insurance company being taxed there. It is  not clear to us that a special rule is required to deem an insurance company to have a PE in a situation where another business would not have a PE. For that reason, we prefer Proposal N, which would apply the usual rules, rather than the new rule suggested in Proposal M.