by Peter Ormosi[1] 

Self-preferencing is usually defined as when a large digital platform treats its own (vertically integrated) products more favourably than those of its competitors. Such behaviour may be harmful to consumers.[2] Versions of this definition have now been embodied in EU and UK legislation which also makes such behaviour illegal for gatekeepers or businesses of strategic market status. However, is this definition of self-preferencing too narrow in that vertical integration may not be a prerequisite for essentially similar strategies to be profitable and similarly harmful? Or is it too broad because it does not specify consumer harm as a condition of illegality, even though self-preferencing often does not harm consumers?

A bit more detail on the legal definitions

The EU Digital Markets Act designates digital platforms as ‘gatekeepers’ if they provide an important gateway between businesses and consumers in relation to core platform services. Article 6(5) states that “The gatekeeper shall not treat more favourably, in ranking and related indexing and crawling, services and products offered by the gatekeeper itself than similar services or products of a third party. The gatekeeper shall apply transparent, fair and non-discriminatory conditions to such ranking.” Similarly, para.20(3)(b) of the draft UK Digital Markets, Competition and Consumers Bill, empowers the CMA to prohibit a firm with strategic market status (i.e. with market power or strategic significance) from: “using its position in relation to the relevant digital activity, including its access to data relating to that activity, to treat its own products more favourably than those of other undertakings.”

What makes these definitions too narrow?

There are numerous cases where the platform is not integrated into supply but the effect of its behaviour is the same as self-preferencing because a platform can profit from restricted supplier competition in many different ways. In Fletcher et al (2023)[3] we group such business practices into three categories:

• Favouring suppliers that confer a higher margin: Suppose there are two products x and y, offered by suppliers A and B respectively, and x would be more relevant for the consumer, but y represents higher revenue for the platform. This could happen, for example, where the platform receives a higher commission from the supplier of y. Examples include recommendations on streaming platforms and online travel agencies.[4] If the platform steers end-users towards suppliers that award the platform a greater margin (for example by manipulating the ranking of recommendations), it could lead to suppliers competing to give the platform a bigger margin. This increased cost would then feed into higher prices for end-users.[5] For example in the Trivago case, the ACCC found that when providing users with holiday recommendations, Trivago’s algorithm gave more weight to those suppliers that paid a higher payment fee (cost per click). Trivago’s own data showed that higher-priced room rates were ranked at the top (the most prominent offer or Top Position Offer) over alternative lower-priced offers in two-thirds of their listings.[6]
• Supplier bargaining power: Another conduct that’s similar in its effect to self-preferencing, is where some larger suppliers have bargaining power with respect to the platform and are able to impose contracts that influence product ranking or product recommendations on a platform. A relatively little-discussed implication of network effects is that certain suppliers can effectively become ‘must-have’. Without their presence on a given platform, end-users on the other side of that platform would switch to an alternative platform. This could in turn lead to other suppliers leaving, and so on. Such critical suppliers have substantial bargaining power and can potentially utilise this to require preferential treatment by a platform. This requirement can be direct, but it can also be indirect. For example, some music streaming services have minimum payment guarantees with the three major record labels, each of which has substantial bargaining power. At the margin, such minimum payment guarantees may be expected to incentivise the streaming services to favour major label music over independent music.[7]
• Wider strategic reasons: Finally, a platform may have wider strategic reasons for distorting supplier competition. For example, a firm which offers an ecosystem with many different services within it may wish to keep end-users within its ‘walled garden’. As such, even if it does not itself provide a particular product, it may be more inclined to recommend a third-party product that lies within the walled garden than one which would take end-users outside it. For example, Google’s mobile search service (at one stage) gave preference in its rankings to content which was cached on Google’s own AMP servers (AMP originally stood for ‘Accelerated Mobile Pages’). This may be – as Google claimed – because Google could then be sure of the download speed and quality of such content. However, it might also have reflected Google’s preference to keep end-users within the Google ecosystem. Similar considerations may apply in relation to Amazon giving preference in its rankings to third party suppliers that use its ‘fulfilled by Amazon’ service.

A platform engaging in any of the above conducts effectively self-prefers, even though the favoured product is not its own vertically integrated product. It is possible that policymakers had in mind an effects-based approach intended to include these business practices.[8] However, even if this is the case, it will likely take lengthy and costly legal battles for the seemingly narrow legal definitions to be tested (and broadened) in court.

What makes the legal definitions too broad?

On the other hand, one could argue that the legal definitions are too broad in that they fail to qualify an important condition of welfare-reducing self-preferencing, the harm to consumers. It is true that a platform’s incentives are not necessarily aligned with consumers. For example, in a field experiment on a video-on-demand system, Zhang et al. (2021) estimate the effect of using a profit-maximising platform relative to an end-user welfare maximising platform. They show that when facing consumers who exhibit a lower price elasticity of demand towards products placed in salient slots a profit maximising platform can increase its profit by hurting both consumer surplus and total welfare.[9] Similarly, Fletcher et al. (2023) show that the preferences of the platform over choice of its recommender system model can be the precise reverse of the preferences of end-users.[10]

But it is also true that there are many ways a platform can limit supplier competition and this can often align with consumers’ preferences. For example, if the way products are displayed by a platform disfavours third-party products that are inferior, it could technically violate self-preferencing provisions, but may still benefit consumers. Objective measures of product quality might reasonably be included in an unbiased ranking algorithm.

It is possible that leaving out reference to consumer harm as a condition of unlawfulness is a sign of turning away from a consumer welfare standard, and both the EU and the UK want to allow these instruments to be able to pick up behaviour that only harms other suppliers (and not the consumer). It would be an unfortunate precedent if these provisions were used to save certain competitors even if disfavouring them would not harm (or even benefit) consumers.

[1] The views expressed in this post are the sole responsibility of the author and cannot be attributed to Compass Lexecon or anyone else.

[2] See, for example, Padilla, J., Perkins, J., & Piccolo, S. (2022). Self‐Preferencing in Markets with Vertically Integrated Gatekeeper Platforms. The Journal of Industrial Economics, 70(2), 371-395.

[3] Fletcher, A., Ormosi, P. L., & Savani, R. (2023). Recommender systems and supplier competition on platforms. Journal of Competition Law & Economics, 19(3), 397-426.

[4] See Bourreau, M., & Gaudin, G. (2022). Streaming platform and strategic recommendation bias. Journal of Economics & Management Strategy, 31(1), 25-47; and Hunold, M., Kesler, R., & Laitenberger, U. (2020). Rankings of online travel agents, channel pricing, and consumer protection. Marketing Science, 39(1), 92-116.

[5] More generally, see Armstrong, M., & Zhou, J. (2011). Paying for prominence. The Economic Journal, 121(556), F368-F395; and Inderst, R., & Ottaviani, M. (2012). Competition through commissions and kickbacks. American Economic Review, 102(2), 780-809.

[6] https://www.accc.gov.au/media-release/trivago-to-pay-447-million-in-penalties-for-misleading-consumers-over-hotel-room-rates

[7] Antal, D., Fletcher, A., & Ormosi, P. (2021). Music streaming: Is it a level playing field?. Competition Policy International Antitrust Chronicle, 2(2).

[8] An effects-based rather than form-based approach to exclusionary conduct has been confirmed by European courts in several cases such as Intel, Qualcomm, or Google Android, and has been explicitly laid down in the new Amending Communication to interpret the 2008 guidance on Article 102 TFEU https://competition-policy.ec.europa.eu/antitrust/legislation/application-article-102-tfeu_en

[9] Zhang, X., Ferreira, P., Godinho de Matos, M., & Belo, R. (2021). Welfare Properties of Profit Maximizing Recommender Systems: Theory and Results from a Randomized Experiment. MIS Quarterly, 45(1).

[10] Fletcher, A., Ormosi, P. L., Savani, R., & Castellini, J. (2023). Biased Recommender Systems and Supplier Competition. Available at SSRN 4319311.