A PYMNTS Company

Online Platforms: Retailers, Genuine Agents or None of the Above?

 |  June 20, 2013

Jun 20, 2013

CPI Europe Column edited by Anna Tzanaki (Competition Policy International) presents:

Online Platforms: Retailers, Genuine Agents or None of the Above? by Matthew Bennett (Charles River Associates)

Introduction by Europe Column editor Anna Tzanaki

In our June edition of the Europe Column, Matthew Bennet (CRA) explores the interesting case of online platforms and their treatment under EU competition law. The article highlights the limitations of any formalistic approaches such as using the Genuine Agency exception to address innovative patterns and business models, and suggests that such an approach risks hampering new business models whilst not actually preventing the competitive harm that may arise. In the end, what should be determinative is not the legal characterization of a (vertical) structure but “whether the arrangement is likely to restrict competition outside of the agreement” using the words of the author. In this regards, the reason why online platforms have drawn so much attention from antitrust regulators lately is the use of retail MFNs, rather than because there is an esoteric question of whether the platform is called a retailer or a provider of distribution services. The author argues that economics can help focus on the right questions and create clear and solid policy rules when assessing the competitive implications of the conduct of online platforms.

The article provides an excellent analysis of all these considerations. Enjoy reading!


There has been a resurgence of legal interest in Genuine Agency agreements on the back of several high profile cases in online platforms including E-books, Amazon and Hotel Booking. To date economics has played little role in this debate, however this memo shows that economics can play a valuable role in helping to disentangle some of the contradictory case law and guidance regarding what constitutes a Genuine Agent. Furthermore, economics suggests that in concentrating on whether online platforms are Genuine Agents, we may be missing the bigger picture of what role an online platform actually performs.


In the last year there has been a resurgent interest in the legal concept of ‘Genuine Agency Agreements’. This has followed several high profile cases in the area of online platforms including the online hotel booking cases in both the UK and Germany, the e-books cases before the European Commission and the US DOJ, and the recently announced cases regarding Amazon and ‘online retailing’ in Germany and the UK respectively. All of these cases appear to have a number of interesting aspects in common. First, they all involve the sale of goods or services online. Second, they all involve the use of Most Favoured Nation clauses (“MFNs”) that constrain the relative levels of online retail prices. Third, the firm with which the consumer actually deals is not the firm that chooses the retail price. For example, although consumers buy their e-books from Apple, Apple does not choose the price of e-books, publishers do. Likewise, consumers buy from Amazon’s website, but Amazon does not choose the price of third-party products on its website, its third-party sellers do. Finally, whilst a consumer goes to Expedia to search and find hotels, Expedia is not choosing the price of hotels on its website, it is the hotel that does so.

This third issue raises a seemingly innocent question of how one defines a retailer. Are online platforms retailers? Are they genuine agents of manufacturers (and hence exempt from competition law) or are they something else altogether? Whilst this question may seem like a rather esoteric and philosophic debate best left to academics, its conclusion has potentially far reaching consequences for the growth of online markets, online business models and how the MFNs are practically treated. For example, if Expedia is defined as a retailer, by not choosing its price itself and leaving the price decision to hotels, Expedia is now party to Retail Price Maintenance. RPM is a hardcore infringement under Article 101(1) TFEU, assumed to be detrimental to competition and carrying with it substantial fines and public castigation – as the publishers and Apple have been forced to face. It also implies a fundamental revisit to their business model by the firms engaging in RPM practices. However, if Expedia were a genuine agent, then the fact that hotels choose price and not Expedia is no longer assessable under Article 101(1) – therefore there is no infringement, no fine, no public pillorying and most importantly, no need to change their business model.

Faced with such starkly contrasting alternatives, it is no surprise that the platforms have argued vociferously that they are not retailers but are merely ‘genuine agents’, simply undertaking their principals’ wishes. With so much at stake, the legal community has refocused on this lesser known area of competition law – Agency agreements. To date the discussion has been legal, with economic analysis playing little part in what constitutes a ‘Genuine Agency Agreement’ or why such an exception from competition law exists. In this paper we attempt to redress this balance by first, setting out some initial economic underpinnings for Genuine Agency law. Second, we use these underpinnings to provide insights into some of the key legal questions regarding what actually constitutes a Genuine Agent. Finally, we caution against seeing everything within a Genuine Agency ‘lens’: it may well be that online platforms are neither retailers (infringing RPM) nor genuine agents (being exempt from Article 101(1)), but are simply platforms, providing retailers with the ability to market and distribute their products.


The eminent economist Ronald Coase first posited that firms exist in order to minimise transaction costs, as the transaction costs of working outside of a firm structure increase, the internal structure of a firm becomes a more efficient way of organising labour and capital than the market.1 Williamson later expanded this theory of the firm to include contracts, explaining that the firm and the market lie on different ends of a continuum with different types of contracts and vertical constraints lying between.2 This concept of firms choosing their structures as a result of the different benefits and transaction costs leads naturally to a viewpoint that competition law should strive to be ‘organisationally neutral’ in so far as possible.3 Indeed moving away from an approach based on the form of the agreement towards an approach based on the effect of the agreement is one manifestation of ‘organisational neutrality.’

Firms should be free to choose the most efficient structure if the competitive effects of their different structure choices are the same. For example, if a distribution agreement and full vertical integration have the same competitive outcome on the upstream and downstream markets, in general both should have identical intervention standards. Without such competitive neutrality, firms may choose to embark on less efficient firm structures simply to exploit the difference in treatment under competition law. This concept of organisational neutrality has been a driving force behind many of the reforms in competition policy over time – including the gradual convergence of the Non-Horizontal Merger Guidelines and the Guidelines on Vertical Restraints. Over time there has been a concerted effort to ensure consistency across competition instruments precisely to prevent organisational arbitrage.

This concept of organisational neutrality provides a rationale for the Genuine Agency exception. In order to appraise an Agency agreement, one could obviously look at all the specifics of the agreement to determine if each clause was pro- or anticompetitive. This would take time and may result in some clauses (such as RPM) being found infringing Article 101(1). Faced with risk that the entire agreement could be declared void, firms will have a strong incentive to vertically integrate. However, if it was known that the distribution agreement created a relationship that was truly analogous to a vertically integrated structure (i.e. a Genuine Agency agreement), then it could simply be treated and evaluated as a vertically integrated structure. Provisions that only relate to the principal and agent, such as RPM, just as in a vertically integrated structure, would not fall within Article 101(1). Such an analysis would fulfil the concept for organisational neutrality and provides a possible short-cut to the need to look at the individual clauses between the principal and the agent. This provides the economic rationale for the first limb of the legal test which states that where an agent, although having separate legal personality, does not independently determine his own conduct on the market, but carries out the instructions given to him by his principal, the prohibitions set out in Article 101(1) will not apply to the relationship between the agent and the principal.4

Of course, just because an agreement structure is analogous to the creation of a vertically integrated firm does not mean there are no competition issues with the agreement. From the Non-Horizontal Merger Guidelines it is clear that the creation of vertically integrated firms can cause harm – even if harm is much less likely than in horizontal mergers. Therefore, one will still want to examine whether there are restrictions, which have an impact outside of the vertically integrated structure. This suggests that a General Agency exception should be narrowly scoped. For example, clauses that grant the principal exclusivity to the agent’s distribution network may still be potentially problematic if there are few competitive alternatives to that agent’s distribution network, in the same way that one would consider the incentives to foreclose for a vertically integrated firm. This provides the economic rationale for the second limb of test for Genuine Agent agreements, where provisions which concern the relationship between the principal and the agent may infringe Article 101(1) in so much as they either lead to foreclosure on the relevant market, or facilitate collusion – for example by collectively excluding others from using the same agents.5


So when is the Principal-Agent relationship sufficiently analogous to a ‘single economic unit’, and hence when they can be treated as if they are a single vertically integrated entity? The key economic question is whether the agent will make the same decision as the principal, or whether it will make its own, possibly diverging set of decisions independently of the principal. Or putting it in economic terms, we are interested in the extent to which the principal’s and agent’s incentives are aligned, thus ensuring that the two parties act as a single economic unit rather than two parties with potentially conflicting goals.6 This criterion allows us to differentiate between Genuine Agency agreements, and ‘sham’ agency agreements implemented simply to bypass RPM law. RPM in a standard wholesaler relationship splits the incentives of retailer and manufacturer. The retailer has ownership of the good, but it is the manufacturer who controls the price – thus creating the scope for divergence of incentives between the parties. However, in a Genuine Agency agreement situation, both ownership of the product and control of the price remains with the principal.7 Unless the ownership is retained by the principal, the agreement looks less like a Genuine Agency agreement and more like a sham agreement to bypass RPM – a question that US law has long considered.8 This explains why ownership of the product is a key consideration for the evaluation of the Principal-Agent relationship. However, although ownership is key to the question of Genuine Agency, it is not, as we discuss below, the only factor considered.

Impact of Risk on Genuine Agency

Risk has long played a key role in the question of Genuine Agency, unsurprisingly given its close correspondence with the concepts of ownership and control. Without corresponding control, ownership of a product is a cost with a significant risk if it cannot be sold. More generally, if the agent holds all the risk from undertaking the controlling principal’s requests, it is more likely to make decisions, which are independent of the principal. Conversely, if the agent is compensated for any risks that it undertakes on behalf of the principal, it will not have the incentive to take independent decisions from the principal. For example, the principal may want the agent to lavishly promote its product, by providing and training knowledgeable sales staff. But providing such sales support is inherently risky – the agent needs to sell enough of the product in question and make enough money on those sales to ensure that it can provide such service profitably. Indeed even if ex post the agent makes money on the product, this does not necessarily mean that there was no risk at the time of the initial decision. The key is whether the agent at the time would have made a different decision regarding the optimal level of sales support than the principal. Whilst the ECJ’s decision in DaimlerCrysler underlined the importance of risk, it also made it clear that risk was not the only factor that matters.

Impact of multiple principal-agent relationships on Genuine Agency

The same analysis can be used regarding the question of whether an agent who has relationships with multiple agents can truly be a Genuine Agent. This is an area, which is particularly opaque in EU competition law given the potentially conflicting views between the Commission and the Courts. Whilst the Commission in its Guidance on Vertical Restraints states explicitly that “it is not material for the assessment whether the agent acts for one or several principals”, nevertheless this appears to be contradicted by the Court of Justice in VVR v Sociale Dienst which stated that a travel agent must be considered independent if “he sells travel organised by a large number of different tour operators and a tour operator sells travel through a very large number of travel agents.”9 From an initial economic viewpoint it would appear that the Court’s analysis is correct. When a firm acts as an agent to multiple, competing principals, it will be difficult to perfectly satisfy all their wishes simultaneously. Competing principals are likely to make competing demands on their agents, for example as to the extent that a principal’s brand is promoted by the agent. Unless there is a clear mechanism in place to deal with such conflicts, competing demands will require the agent to make its own independent decision regarding which principal’s instruction to prioritise over others. By the same reasoning as before, if the agent is making such independent decisions, then, absent other factors, it will be more difficult to argue it will act as a single economic unit with the principal and hence is a Genuine Agent.

Bargaining Power

One interesting question is whether one can look at the bargaining power, and therefore the types of conditions that the agent can impose on the principal, as relevant to the question of whether it is a Genuine Agent. In this context it has been posited that if the agent can impose disadvantageous arrangements on the principal, then it may be harder to state that the agent is acting as a subsidiary of the principal – the concept of the ‘tail-wagging-the-dog.’ Whilst such an argument may warrant closer attention, from an initial economic viewpoint this seems to confuse the question of whether the arrangement can be truly called a Genuine Agency once it has been entered, and the ex-ante conditions imposed by the agent for entering into the arrangement. In this regard the downstream party may well be in a stronger bargaining position before entering into the agreement and therefore may be able to extract concessions from the principal in return for becoming its agent. However, once the conditions have been decided and the agent has entered into the arrangement, the agent may well act as a single economic unit with the principal and hence constitute a Genuine Agent. Therefore, in considering bargaining power one must be careful not to conflate any bargaining power before the agreement with the ability to act independently once the agreement has been entered into.


So far so good, whilst the case law and policy are somewhat opaque, there seems to be a clear economic underpinning for the Genuine Agency exception to Article 101(1). Nonetheless, a difficulty arises when one starts to attempt to widen the Genuine Agency analysis outside of standard retailers and into the realm of distribution platforms. With the advent of the Internet we have seen the significant growth of the business model in which disparate buyers and sellers are brought together on a common platform. In this sense the platform is not a typical retailer or seller whom the Genuine Agency analysis has historically considered.

As a result, an overly narrow interpretation of the Genuine Agency analysis may well suggest that these platforms are not Genuine Agents and therefore should not be exempt from Article 101(1). This is not just relevant for online platforms and indeed one can illustrate the concern with a simple example taken from the traditional world of bricks and mortar. Consider the application of the Genuine Agency test to a shopping mall. Suppose a fictitious shopping mall (“The Mall”) holds a large number of shops, including competing goods retailers (John Lewis, Debenhams etc), clothes retailers (Next, M&S etc) and other such stores. The shops all rent space from The Mall on a fixed monthly basis. The Mall, like many malls, also advertises in the newspaper to attract customers, as its success depends on the number of customers it can attract. These advertisements may include advertisements for specific shops, and may even involve advertising promotions within specific shops – “come to The Mall, this week Dixon’s have a massive sale on LCDs”. Is The Mall a Genuine Agent? Applying the test one may conclude not. First, it is making specific and risky investments – through the advertising of retailer stores, and indeed their products. Second, The Mall has multiple principals – retailers, which are competing with each other and hence have competing incentives. Therefore, the relationship between any given shop and The Mall may not be seen as analogous to a single vertically integrated firm creating unity of conduct. However, does this imply that The Mall must choose the product prices for the stores within it or risk infringing RPM and hence Article 101(1)? This would seem to make little sense. Fundamentally, The Mall’s and the shops’ incentives are aligned, they both want to attract customers. Nor is ownership and price control divorced, the shops control both the goods that they sale, and the price at which they sell it. The Mall is simply a platform to enable the shops to retail their products.

Whilst the example above may appear trivial, it is surprisingly hard to draw a clear distinction between it and the case of online platforms. Taking the recent OFT hotels case as an example, one might, for the same reasons as in the shopping mall scenario, argue that the online booking agents would fail a narrowly defined Genuine Agency test. However, one could equally argue that online travel agents are merely platforms that provide distribution services to facilitate hotels to sell their rooms directly to consumers. Both the online travel agent and the hotels, which list on it, arguably have aligned incentives – getting as many customers onto the platform. Nor is ownership and price control divorced – the online travel agent does not take ownership of the hotel rooms, nor does it chose the prices of hotel rooms on its website. Thus, if agreements between online travel agents and hotels do not meet the Genuine Agent test, why does the shopping mall fulfil it, or if it doesn’t but yet still does not infringe RPM, why is this the case when the online travel agent does? This highlights the potential danger of drawing the Genuine Agency test too narrow in scope – whilst it may or may not be correct that online travel agents are not true agents, without clearly delineating what makes an online platform a distributor, and not a retailer, other online platform business models may shrink because of the competition law risk, to the detriment of consumers and innovation.


Focusing on this potential Genuine Agent conundrum risks missing the wood from the trees (to overextend our trees metaphor). One of the main reasons competition authorities became interested in the platform cases discussed in the introduction, was due to their use of retail MFNs. However, the advantage of taking an economics-based approach to such MFNs is that it does not rely on being able to define precisely whether the distributor is an agent or a retailer. Suppose that the online travel agent was indeed a Genuine Agent, and merely provided platform services to hotels, this does not fundamentally change our analysis of the retail MFNs. A retail MFN from Online Agent A still forces the Hotel to set its price using Agent A’s platform for less than or equal to the price that it charges on Agent B’s platform.

In such an agreement context one may still have a concern regarding the impact that such a restriction has on the pricing incentives of the online travel agent platforms. In particular, such a constraint potentially limits the extent to which the Online Agent B can use price as a means of competing with Online Agent A. As a result, B can no longer offer a commission discount to the Hotel in order to encourage the Hotel to reduce its prices thereby driving more traffic to Online Agent B. Of course there may also be an efficiency rationale for such an agreement. If Online Agent A provides a higher quality platform to the Hotel and invests more in attracting users, the Hotel may be tempted to exploit Online Agent A’s investment initially, but then once it has established a customer contact, may direct these customers to the cheaper Online Agent B for subsequent purchases. In this sense the Hotel and Online Agent B may be able to free ride on Online Agent A’s platform investments. However, whilst this agreement may or may not be anticompetitive depending on the potential harm and benefits, this consideration has little to do with the question of whether the online travel agents are ‘Genuine Agents’ or not. The benefits and the harm will exist even if they are considered Genuine Agents under a formalistic legal standard. In this regard one must ask the question of whether too much attention is provided to the first limb of the Genuine Agency test and not enough attention is paid to the second limb – whether the agreement restricts competition by foreclosing competition or facilitating coordination.


The challenge for Competition Policy going forward is to design clear policy rules which both ensure that platforms are not treated as retailers subject to the Genuine Agency test, but also that retailers do not simply label themselves as platforms, thereby simply creating a Platform exception in place of the Genuine Agency exception. Whilst there is to date very little case law on this issue, from an initial view it would appear that one of the main determinants is which party has the commercial relationship with the customer. In the case of the shopping mall it is clear that customers have the relationship with the stores and not the mall.

Second, it is worth keeping in mind that the legal scope of the Genuine Agent exception appears to have been drawn relatively narrow, a scope which is arguably in line with the economics. The fact that a Principal-Agent relationship is analogous to a vertically integrated firm structure does not necessarily mean that the structure will not harm competition. This might suggest that one should perhaps worry less about the question of whether the arrangement is a Genuine Agency and spend more time looking at whether the arrangement is likely to restrict competition outside of the agreement. In this respect practices such as MFN contractual guarantees or price guarantees may, depending upon the facts of the case and the market context, restrict competition even if the parties are within a Genuine Agency relationship.

In summary, there is much at stake within this interesting area of competition law, and whilst economics has been relatively slow to input into the discussion, it is an area in which economics can provide both useful and cautionary insights.

(Click here for a PDF version of the article.)

1 Coase, Ronald H. (1937), ‘The Nature of the Firm’, Economica 4 (16): 386–405.
2 Williamson, Oliver E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications, New York: The Free Press. See Zhang (2012) for a more detailed discussion of the contractual economics and legal underpinnings for Contingency Contracts and Genuine Agency: Zhang, Angela Huyue, ‘Toward an Economic Approach to Agency Agreements’ (September 1, 2012), Journal of Competition Law and Economics (Forthcoming).
3 We state ‘strive’ because there may be instances when a deliberate decision is made to create an inconsistency between different structures. The main example is horizontal price agreements. Even though a merger of two firms with 5% market share may have negligible impact on competition, horizontal collusion may be so unlikely to create benefits that one may simply want to ban it outright because there are no likely efficiencies from which firms may be deterred.
4 Case T-325/01 DaimlerChrysler [2005], at paragraph 88.
5 Guidelines on Vertical Restraints [2010] OJ C 130/01, paras 19-20.
6 In this respect there is a close analogy to the literature on the efficiency rationale for vertically integration. When firms are separate, they may have divergent interests regarding the optimal pricing level (‘double marginalisation’), the level of effort provided in marketing the upstream product, or the possibility of hold-up. Vertical integration internalises these differences and ensures that rather than maximising individual profits, joint profits are maximised,
7 Zhang (2012) advocates the use of ownership and control as a means to delineate the problem of not allowing firms to contract their way out of RPM through the use of consignment contracts (see footnote above).
8 See General Electric, and Morrison v. Murray Biscuit.
9 VVR v Sociale Dienst, at para 20.