Competition agencies and courts have increasingly had to deal with multiplatform businesses – and have started to incorporate economic insights on their operation into their decisions. Nonetheless, many questions concerning the design of antitrust analysis involving platform businesses remain unsettled.
This article, available here, develops three basic principles for conducting the antitrust analysis of multisided platforms in light of economic learning, as follows:
Section II explains how multisided platforms increase welfare by reducing transactions costs and resolving externalities among economic agents.
Platforms lower transaction costs by bringing potential traders to a common place for interacting, thereby solving a collective action problem. The economics literature often relies on simple indirect network effects to explain how two-sided platforms create value. Positive indirect network externalities arise because the presence of additional numbers of the right counterparties increases the likelihood of good exchanges. In practice, however, the externality issues addressed by platforms are broader and subtler. Platforms also often deal with negative network externalities (e.g. intrusive adds or disruptive behaviour), by developing governance structures – involving rules, enforcement, and punishment – to limit behavioural externalities which reduce platform value. Managing this panoply of externalities, which extend well beyond positive indirect network effects, is central to how multisided platforms reduce transaction costs and increase the value available to their participants and themselves.
Section III presents three normative principles for policy interventions.
The basic antitrust framework for analysing business conduct applies just as well to multisided platforms as it does to single-sided businesses. However, these platforms create new challenges in terms of measuring consumer welfare. Traditional businesses serve many different kinds of customers and sometimes even charge them different prices. What makes two-sided platforms different is that not only are there two distinct types of customers, but the demand by those customers are interdependent, and their welfare is linked. As a result, one needs to take feedback effects into account. One should also take into account that: (i) platforms maximise consumer welfare when they generate more trades and when the total consumer surplus for each trade is bigger; (ii) competition among platforms can also increase consumer surplus by competing down the share of the gains to trade taken by the platform.
In the light of this, the author proposes three principles that should govern enforcement: (i) The analysis should consider the joint surplus of consumers on both sides of a platform, unless there is a policy reason for placing a lower relative value on one type of consumer. (ii) Since customer welfare on each side of the platform are affected by the interdependencies between customer groups, policymakers need to account for feedback effects from both groups – since ignoring those feedback effects could affect the welfare of the groups being considered. (iii) While there may be opportunities for policymakers to improve matters, platforms have financial incentives to maximize the overall value of the platform, since that is the source of their earnings. In other words, platforms have incentives reduce externalities among their participants. That is reflected in the extensive use of governance systems by platforms which deal with cross-side negative externalities.
Section IV applies the standard error-cost framework for competition policy design to multisided platforms.
One must choose how to account for the existence of various types of customers when analysing multisided platforms – either by making explicit choices or through default. Decision-makers could, for example, decide to restrict relevant antitrust markets to one type of customer, or decide to take both types of consumer into account. Two-sided platforms provide services that help two distinct types of customers engage in mutually advantageous trade – so courts and competition authorities would need to examine the impact of anticompetitive practices and mergers by reference to gains to trade and the volume of trade for platform participants. At the same time, this requires decision-makers to think about how to address disparate effects on different types of customers or platform sides (e.g. can pro-competitive effects for some customers outweigh anticompetitive effects for other types of customers), or how to treat the complexities of interdependent demand.
The economic theory of multisided platforms cannot by itself provide definitive guidance, as other practical issues such as the administrability of rules, balancing error costs, and adhering to antitrust statutes must be taken into account. Nor can precedent that did not address these two-sided issues, or did not do so in light of the new economic learning, provide a clear path.
Section V considers competing approaches to analysing multisided platforms, by reference to the US Supreme Court’s American Express case.
I have discussed this case on multiple occasions, so I will not rehash it here. The case was considered under the rule of reason, which involves a multi-step inquiry under US law according to which the burden shifts between plaintiffs (to prove anticompetitive effects or that less anticompetitive conduct was possible) and defendants (to prove efficiencies, and that they outweigh anticompetitive effects). Much of the debate over the appropriate principles for analysing a restraint by a two-sided platform boiled down to how, if at all, effects of a competitive restraint should be considered – in particular by focusing solely on one side of the market (i.e. merchants) or both sides of the market (i.e. including cardholders as well). The answer to this ultimately depended on market definition. If the relevant antitrust market included both customer groups (i.e. a market for card transactions comprising merchants and card-users), the anticompetitive effects analysis in the first step of the rule of reason would focus on both. If the relevant antitrust market only included the merchants (i.e. a market for the provision of card services to merchants), one could argue that the effects on cardholders should be considered as a possible procompetitive efficiency at a second stage, or not at all (since they could be said to occur out of market). Focusing on the importance of indirect network effects and the interdependent pricing structure for two-sided platforms in assessing market definition, the majority decided that the relevant market in this case was a ‘transaction market’ which included both merchants and cardholders. As a result of this market definition, the analysis of competitive effects had to consider the impact of the restraint on the welfare of both types of customers – and a price increase only on the merchant side did not suffice to establish anticompetitive effects.
The Court’s decision in American Express ultimately turned on how much weight to give the different types of platform participants. However, there was no substantive discussion of the rationale for placing more, less, or the same weight on each type of participant. Largely because of how the case was argued before the Court, the determination of how to count consumer welfare was subsumed in a debate over market definition. From a public policy standpoint, it is hard to see the rationale for this approach – which can prove more controversial in future cases and requires additional discussion on how to think about consumer surplus for the purpose considering different ways of inferring anticompetitive harm.
On the other hand, it is now clear that the US, EU and China have all concluded that two-sided features are relevant for assessing whether restraints are anticompetitive. The issue for antitrust enforcement is no longer ‘Should we consider both sides?’ but rather ‘How should we consider both sides, and what analytical tools should we deploy for doing so?
This is a paper prepared for Very Distinguished People, as it was based on a keynote address delivered by the author at the 2018 CRESSE Conference. It is sophisticated, and the proposed normative principles are eminently sensible; and I would expect nothing else from the author. In this regard, I particularly enjoy how the author brings to the fore an intuition I have had for a while, namely that the fundamental questions that multisided platforms raise for competition law revolve around our understanding of consumer welfare – without prejudice to the obvious technical challenges that this then creates, which are neatly discussed in the two other pieces I am reviewing below.
Given his extensive publication history on this topic, it was to be expected that the author would emphasise the efficiencies created by platforms, and the risks of not taking into account the effects of a conduct on both sides of platforms. On its face, This position seems obvious – after all, it is common ground that anticompetitive effects should be balanced with efficiencies. However, such a conclusion can only be reached with such certainty if one avoids dealing with the underlying normative questions regarding whose consumers’ welfare we should focus on, or whether antitrust should focus on total or consumer welfare. These are questions that competition law has been able to broadly avoid, but which digital markets bring to the forefront. This is not to say that the author should have addressed these questions – in effect, he seems to advocate that such questions should be dealt with head-on.