It is commonly accepted that market definition is more complex in two-sided markets than in normal (single-sided) markets. A proposal to simplify this exercise is to distinguish between transaction and non-transaction platforms. Two-sided transaction platforms such as payment card systems, online marketplaces and auction houses, are characterised by the presence and observability of a transaction between the two groups of platform users, so that the platform operator can impose a per transaction charge or two-part tariff (for joining and using the platform). In contrast, non-transaction platforms, including most media platforms, have no such transaction between the two sides. It follows that, while in non-transaction markets one must define two (interrelated) markets, while a single market encompassing both sides should be defined for transaction platforms.
The author argues here that this distinction is inapposite, particularly in the context of the hypothetical monopolist test. This article addresses the various theoretical and practical arguments put forward in support of the distinction between transaction and non-transaction, and explains why none of these justify a different approach to market definition for both types of platforms, as follows:
Section 2 reviews the theoretical foundations of how market definition should be pursued in multi-sided markets.
The aim of market definition is to identify those products that are close enough demand substitutes to impose a price constraint on each other. Assuming an underlying general demand system with multiple products, where demand for each product depends on its own price and on the price of all other products, the hypothetical monopolist test defines the relevant market as the smallest group of products (and geographic area) in which a hypothetical, profit-maximising monopolist would impose a SSNIP. In normal (single-sided) markets, only substitute products are of interest. The situation is more complicated where complements are involved, since authorities have to make a decision of whether to include such complements in the tested products (e.g. after-markets). The literature does not prescribe a single approach to complements, and the author holds that this decision necessarily involves an element of judgement and should be guided by the specific competition concerns and the facts of the case.
This challenge concerning complements is similar to that raised by two-sided platforms – i.e. should we consider both sides of the platform or just one? With multi-sided platform, a hypothetical monopolist test can produce a number of results: (i) the monopolist will be able to increase prices above 5-10% on both sides of the market (i.e. the platform constitutes the market); (ii) the monopolist will not be able to increase prices above 5-10% on either side of the market (i.e. the relevant market must be expanded); (iii) the monopolist will be able to increase prices above 5-10% on one side of the market, but not the other. The implications of this last result for market definition are not obvious. Given that two-sided platforms—whether competitive or monopolistic—tend to have a skewed pricing structure across the two sides, the price increases on either side that result from hypothetical monopolisation may also be asymmetric. In practice, market definition in this case will depend on the competition concern that market definition is aimed to shed light on. If the competition concern relates purely to one side, an approach based on the smallest-market principle that looks into that area of market power will be informative. However, if the competition concern is about conduct that relates to both sides of the platform—as will often be the case given the close interaction between the two sides— then this market definition test may not yet provide the full picture of competitive constraints on the platform
Section 3 evaluates various factors that are said to justify taking a different approach to market definition for transaction and non-transaction platforms.
None of the considerations above is influenced by the platform being transaction or non-transaction. However, there are other arguments that could justify adopting such a distinction. For example:
- In transaction platforms there is such a thing as a total platform price—that is, if a transaction platform sets a transaction price across the same dimension (e.g. number of transactions or percentage of transaction value) on both sides, the sum of these prices can be seen as the total per-transaction price. However, even for non-transaction platforms (and for transaction platforms that charge users on both sides across different dimensions), one can look at the platform revenues on both sides and infer some kind of quantity weighted average or composite price for the platform as a whole.
- The simplification of analysis that the distinction is supposed to bring also does not hold, since the distinction between transaction and non-transaction platforms is not clear-cut. The literature distinguishes between transaction platforms where the platform can observe and charge a transaction, and platforms that promote interactions which it cannot observe and charge for. This distinction is not accurate. For example, price-comparison websites deploy a number of business models, and it is unclear which can be classed as transaction platforms and which as non-transaction. Nor is the concept of ‘observable transaction’ that useful in the digital sphere – e.g. clicking in an advertisement is observable, but that does not mean that platforms where this happens will be transaction ones. As such, it is unclear how this distinction can accurately capture the competition dynamics inherent in market definition.
- It is said that transaction and non-transaction platforms face different types of competitors. In particular, non-transaction platforms may face single-sided competitors on either side, whereas transaction platforms are said to compete only with other two-sided platforms. However, this differentiation by type of competitor cannot be a decisive factor for market definition. As there can be a spectrum of interactions between the two sides of a platform (from ‘mere’ interactions to transactions), by the same token there can be a spectrum of ways in which users on either side of the market interact with each other, not all of which are through two-sided platforms. Two-sided platforms may compete with different types of ways or channels through which users on both sides interact.
- Situations where the externality flows in only one direction are more likely to arise for non-transaction platforms (e.g. when advertisers care about the number of users in a media platform, but not the other way around) than for transaction platforms. In many such circumstances, the market can be defined for one side of the platform while discounting the other. However, in reality both sides of the platform must still be taken into account, since the hypothetical monopolist will still set its profit-maximising prices on both sides simultaneously, and one should take into competitive constraints on both sides. In any event, this does not justify a distinction between transaction and non-transaction platforms for market definition, which should be pursued in the usual way.
Section 4 concludes by articulating a policy proposition for market definition in two-sided markets.
The hypothetical monopolist sets its profit-maximising prices on both sides, as a function of price elasticities and externalities between the sides. It follows that two-sided platforms compete on both sides of the market at the same time. They set low prices on one side to draw away customers on that side from other platforms. By doing so, they also draw customers on the other side away from other platforms due to the externalities. Both sides of a platform thus compete with both sides of other platforms, and relevant markets must include both sides. This simple proposition holds for transaction and non-transaction platforms alike. It follows that whether the platform is a transaction or non-transaction platform should be disregarded when defining markets.
This is an interesting paper that reviews the fundamentals of market definition and applies them to multisided markets. It is a bit beyond my skillset, but I think I still understood the argument, so I can recommend it to anyone interested in the topic regardless of economic nous.
At the same time, I am not sure I fully understand the paper in its entirety. For example, I understand why, following the US Supreme Amex judgment, people may want to dispel the idea that the transaction/non-transaction provides a clean answer to how to define multisided markets. At the same time, I found it surprising that the paper does not devote any attention to whether the distinction should be a good place to start as regards certain platforms – even though that may be a consequence of the author’s firm rejection of the distinction in the first place. Further, I am not sure I fully understood the author’s argument about what the appropriate scope of a product market is as regards platforms. It is clear that interactions between both sides of a platform need to be taken into account. But then the author seems to simultaneously hold that it is possible to start market definition merely by looking at one side of a platform, and that ‘In all instances (…), the HMT still leads to one market definition for the hypothetical platform monopolist for the competition analysis in question, not two separate market definitions for each side.’ Both cannot be correct, but maybe I am misunderstanding the argument. A last thought I had was that the article did not engage with arguments that market definition is not that important for multisided markets – as developed in the article below.