Drawing from the economics of two-sided markets, we provide suggestions for the definition of the relevant market in cases involving two-sided platforms, such as media outlets, online intermediaries, payment cards companies, and auction houses. We also discuss when a one-sided approach may be harmless and when instead it can potentially lead to a wrong decision. We then show that the current practice of market definition in two-sided markets is only in part consistent with the above suggestions. Divergence between our suggestions and practice is due to the failure to fully incorporate the lessons from the economic theory of two-sided markets, to the desire to be consistent with previous practice, and to the higher data requirements and the higher complexity of empirical analysis in cases involving two-sided platforms. In particular, competition authorities have failed to recognize the crucial difference between two-sided transaction and non-transaction markets and have been misled by the traditional argument that where there is no price, there is no market.
In most economic models the relevant market is simply assumed. It is therefore an issue to which most academic economists do not devote much attention. In practice, however, market definition is of great importance for antitrust and regulatory cases. A mistaken definition of the relevant market might, for instance, lead an antitrust authority or a court to block a welfare-enhancing merger, or to allow a welfare-detrimental one. Similarly, a regulatory decision imposing a set of requirements on an incumbent might be socially inefficient if the incumbent faces sufficient competition. In the case of an appeal, the recognition of a wrong market definition is often sufficient for a court to reject the whole analysis and to rule in favor of the appellant, irrespective of any other argument brought up by the antitrust authority or the regulator.1 Market definition is therefore the foundation stone on which an antitrust case or a regulatory intervention is built.2
Although the 2010 U.S. Merger Guidelines seem to move away from market definition and propose to replace it by the measurement of the upward pricing pressure (UPP),3 it remains to be seen the extent to which, even in the United States, courts will adopt this new concept and allow an antitrust authority to block a merger without having previously defined the relevant market.4 Moreover, market definition seems bound to remain—at least for a few more years—the first step in merger cases in other jurisdictions, such as the European Union. Finally, even if UPP were to replace market definition in merger cases, market definition would remain crucial in order to establish the existence of a dominant position in cases of abuse of dominance, or in order to justify regulatory intervention.
The main purpose of market definition is to identify the products that exert competitive pressure on the products sold by a particular firm or firms, be they firms that plan to merge, firms suspected of anticompetitive abuse, or firms that are candidate subjects of regulatory intervention. Market definition is therefore an attempt to define a group of products, which are substitutable to such an extent that the firms producing them can be perceived as competing against each other, and which therefore constrain each other's ability to increase prices.
The most rigorous conceptual tool used to define the relevant market is the “small significant non-transitory increase in price test” (SSNIP test).5 This test defines the narrowest market (collection of products plus geographic area) in which a single firm could sustainably raise prices or otherwise exercise market power.
Yet the SSNIP test was designed for a single-sided market. Drawing from the economic literature on two-sided markets, we will argue that, in a two-sided market, the test cannot be applied in its traditional form. We will then discuss how the same logic underlying the traditional test can be extended to a two-sided setting, while taking into account the two-sided nature of the market.
Relatedly, because in a two-sided market a firm sells two different products or services to two different groups of consumers (the so-called two “sides” of the market), two additional questions naturally arise: whether one or two relevant markets need to be defined, and whether the presence of another side of the market should influence market definition. We will argue that market definition should always take into account both sides of the market, but that whether one or two relevant markets need to be defined depends on the type of two-sided market.
We will then compare our suggestions to the approach taken by competition authorities in recent decisions. Although our sample of cases is not exhaustive, it includes some well-known two-sided markets cases handled by the European Commission, as well as some less well-known cases handled by national competition authorities. We consider these cases to be indicative of the issues that may arise when defining two-sided markets.
We will show that divergence between our suggestions and the competition's decisional practice is due to three main reasons: (1) the failure to fully incorporate the lessons from the economic theory of two-sided markets, (2) the desire to be consistent with previous practice, and (3) the higher data requirements and the higher complexity of empirical analysis in cases involving two-sided platforms. In particular, the failure to incorporate the lessons from the theory of two-sided markets manifests itself through the failure to recognize the crucial difference between two-sided transaction and non-transaction markets, and through the use, in cases concerning two-sided platforms, of the argument that where there is no price, there is no market.
The article proceeds as follows. Part II defines two-sided markets and makes a crucial distinction between two-sided transaction and non-transaction markets. The following three parts draw from the economics of two-sided markets to derive suggestions for market definition and analyze the recent decisional practice of competition authorities by examining cases in two-sided markets in light of these suggestions. In particular, Part III considers whether competition and regulatory authorities should define one or two markets, and why. Part IV explains why authorities should take into account both sides of the market in assessing the relevant market. Part V discusses whether and how authorities should perform the SSNIP test to define the relevant market. Throughout these three parts, we compare our suggestions with current practice. Part VI concludes.
TWO-SIDED TRANSACTION AND NON-TRANSACTION MARKETS
According to David Evans,6 a two-sided market is a market in which a firm acts as a platform: it sells two different products to two groups of consumers, while recognizing that the demand from one group of consumers depends on the demand from the other group and, possibly, vice versa.7 In other words, the demands on the two sides of the market are linked by indirect network effects,8 and the firm recognizes the existence of (that is, internalizes) these indirect network effects. The buyers of the two products, however, do not internalize these effects, which are therefore called externalities. Note that, as recognized also by Jean-Charles Rochet and Jean Tirole, this makes a two-sided market different from markets for complementary products, in which both products are bought by the same buyers, who, in their buying decisions, can therefore be expected to take into account both prices.9
Prominent examples of two-sided markets include (1) media markets, where firms sell content and advertising space, (2) payment cards markets, where firms sell the use of a card to buyers and that of a point-of-sale (POS) terminal to shops, and (3) online intermediaries, which sell their services to buyers and sellers. In media markets, advertisers' demand for advertisements on a media outlet increases with the number of consumers of content (viewers, readers, listeners, and the like), while viewers, readers, and listeners might also be positively or negatively affected by the quantity of advertising. Similarly in payment-card markets, the more cardholders there are, the higher the demand from shops, and vice versa. Card issuers, such as American Express or VISA, are well aware of this relationship between the two demands they face. Similarly, online intermediaries such as eBay know that the more buyers that visit their website, the more likely it is that sellers will use their services and vice versa. In fact, the most common business model on the Internet, as shown by the success of Google or Facebook, is to attract users with various free services and “sell” them to advertisers.
Different types of two-sided markets can be distinguished. Most important for the analysis of market definition in a two-sided market is the distinction proposed by Eric van Damme, Lapo Filistrucchi, Damien Geradin, Simone Keunen, Tobias Klein, Thomas Michielsen, and John Wileur between two-sided transaction and non-transaction markets.10 Two-sided non-transaction markets, such as most media markets, are characterized by the absence of a transaction between the two sides of the market. Even though an interaction is present, it is usually not observable, so that a per-transaction (or per-interaction) fee or a two-part tariff is not possible.11 Two-sided transaction markets, such as payment cards,12 are instead characterized by the presence and observability of a transaction between the two groups of platform users. As a result, the platform is not only able to charge a price for joining the platform, but also one for using it—that is, it can ask for a two-part tariff.13
Two-sided transaction markets are characterized by both membership externalities (or indirect network effects) and usage externalities. Membership externalities, which are also present in non-transaction markets, arise from joining the platform (buying a newspaper, placing an advertisement in a newspaper, holding a payment card, having a point-of-sale terminal, listing your product at an auction, or attending an auction), whereas usage externalities arise from using the platform (for example, paying or accepting payment with a card, selling or buying a product at an auction). As the value of joining the platform depends on the number (or more generally the demand) of customers of the other side, the benefit of using the platform similarly depends on the demand for usage by the other side. For instance, assuming that a customer holds a card and a shop has the corresponding point-of-sale terminal, even if this customer wants to pay by card, the merchant must be willing to accept that card for that particular transaction, and vice versa. Once again, these externalities are not internalized by the users of the platform—that is, the cardholder and the merchant. For instance, suppose a given merchant would benefit from being paid by card because she would not need to go to the bank to deposit cash and she would not have to face the risk of being robbed. A cardholder would not take that into account when offering to buy in cash or by card. He would only consider his own convenience.
As pointed out by Rochet and Tirole, in a two-sided market, where two products or services are sold to two groups of customers, one can distinguish the price level and the price structure.14 The price level is (roughly) the sum of the two prices, while the price structure is (roughly) the ratio of the two prices.15 In this respect, Rochet and Tirole define two-sided markets as follows:
In other words, according to Rochet and Tirole, for a market to be two sided, it is sufficient that the price structure is non-neutral.
A market is two-sided if the platform can affect the volume of transactions by charging more to one side of the market and reducing the price paid by the other side by an equal amount; in other words, the price structure matters, and the platforms must design it so as to bring both sides on board.16
As explained by Lapo Filistrucchi, Damien Geradin, and Eric van Damme, at first sight this definition might seem very different from the one proposed by Evans, but in practice it is just broader.17 For the price level to be non-neutral, it must be impossible for the side that pays more to pass through the difference in its cost of interacting to the side that pays less. If a complete pass-through were possible, the price structure chosen by the platform would not matter. The platform would not control the relative price charged to the two sides.
However, a complete pass-through can only take place if there is a transaction between customers on both sides of the market, such as a payment card transaction, or a transaction in a virtual marketplace (for example, eBay). In markets where there is no transaction between end-users of the platform, no pass-through between the two sides can take place. Thus, given the presence of externalities between the demands on the two sides, the non-neutrality of the price structure necessarily holds and the platform has perfect control of the relative prices charged to the two sides.18
In the next three parts, we will discuss how a competition authority should define the relevant market in cases involving two-sided platforms, and how the distinction between two-sided transaction and non-transaction markets is crucial for the definition of the relevant market. We will compare our suggestions with current competition authorities' practice and show that one of the main reasons why theory and practice diverge relates to the failure to recognize the crucial difference between two-sided transaction and non-transaction markets.
DEFINING ONE VERSUS TWO MARKETS
As discussed above, the two sides of a two-sided market are, by definition, linked by the presence of indirect network effects. As a result, firms can be viewed as platforms that need “to get both sides on board”19 in order to do business. The question then arises whether there is only one market that needs to be defined, or if there are two.
One might be tempted to argue that, when one side of the market does not pay, only one market should be defined, namely, the one with paying customers. After all, we are used to thinking that, in a market, one party buys a product from another party. However, one of the first and most important contributions of the theory of two-sided markets is that giving away a product for free may be a profit maximizing strategy for a firm, even for a monopolist. The reason is that by giving away a product for free, a platform boosts the number of people receiving that product. Although it loses money on one side, it may recover this loss on the other side, making higher profits overall than if it were to sell on both sides at a positive price.20 This is, for instance, the case with free newspapers, payment cards, and discotheques where women, but not men, get in for free. This is one reason why the argument “no price, no relevant market” does not apply to a two-sided market. Another reason is that, as further discussed below, platform behavior on the paying side of the market is likely to affect customers on the other side. For instance, a merger between free-to-air TV channels may raise the price of advertising but also affect viewers.
The question can thus be rephrased as whether there are two markets to be defined or only one market encompassing the two sides. For instance, when analyzing a merger among newspapers, the question is whether there is a market for newspapers or there is a market for advertising (on newspapers) and a market for the dissemination of news. Similarly, in a case involving payment cards, the question is whether there is a single market for payment cards services, or a market for payment cards services to cardholders and a market for payment cards services to merchants.
One of the consequences of defining only one market is that a firm would be either on both sides of the market or on none. Defining instead two interrelated markets would allow a platform to be on one side of the market but not on the other.
Whether one or the other outcome is preferable depends on the type of two-sided market under consideration. Everyone would probably agree that a payment card company such as American Express is either in the relevant market on both sides or on neither side, for the reason that either the transaction between the buyer and the merchant takes place using American Express services on both sides, or it does not take place through American Express. The analysis of a merger between two payment card platforms should thus consider whether cash or PayPal exert competitive pressure on payment card companies on both sides of the market.
However, in a case involving newspapers, a product might be in the relevant market on the advertiser's side but not on the readers' side.21 For instance, suppose that people do not regard TV and newspapers as substitutes because they read newspapers on the metro going to work and watch TV at home in the evening. However, if advertisers are interested in reaching each person only once during a day, they will tend to regard TV and newspapers as substitutes. TV would then be in the same relevant market as newspapers on the advertiser's side but not on the readers' side. The analysis of a merger case involving newspapers should then consider that TV exerts competitive pressure on newspapers in the market for advertising, but not in the market for content.
We submit that the crucial element distinguishing a newspaper market from a payment cards market is that, in the latter, a transaction is present between so-called end-users—that is, between customers on the two sides of the market. Whether one should define a single market or two interrelated markets depends on whether we are dealing with a two-sided transaction market or a two-sided non-transaction market.22
To sum up, our first suggestion is
Suggestion 1: In two-sided non-transaction markets, two (interrelated) markets need to be defined.
In two-sided transaction markets, only one market should be defined.
In fact, as already mentioned, in two-sided transaction markets end-users on the two sides can be charged both a fixed and a per-transaction fee. In practice, the former is the price paid for joining the platform, the latter is the price paid to use the platform. In most cases—consistent with current practice regarding after-markets25—one should define a single system market, comprising both the primary market for membership and the secondary market for use, since it is usually the case in two-sided transaction markets that “the likely expenditure on the secondary product is a relatively high proportion of the primary product's price”26 and customers “anticipate the cost of future necessary purchases of the secondary market product when buying the primary product.”27 The only differences with respect to a single-sided market are the presence of indirect network effects between the primary markets on the two sides and the fact that the secondary market is a transaction market linking the two-sides.28 As discussed above, these differences imply that a single system market cannot but comprise both sides of the market.
One may wonder whether defining a single market in the case of two-sided transaction markets would lead to a relevant market comprising products that are not substitutes for each other. For instance, a single market for services provided by rental agencies would include both services to the tenants and services to the landlords. A landlord trying to rent a flat would not usually be interested in visiting flats that are up for rental. Similarly, for payments card services, a cardholder would not usually consider getting a POS terminal instead of a payment card, and a merchant would not consider getting a payment card instead of a POS terminal.
Yet, in all these cases, defining a single market implies defining the market for services to a transaction. The product that is offered is the possibility to transact through the platform, be it the rental agency or the payment card company. Candidate substitute products are not only other platforms that offer, to both sides, the possibility to transact, but also non-intermediated transactions, such as a direct rental or a cash payment.
Moreover, as noted by William Rooney and David Park, the suggestion to define only one relevant market for cases involving two-sided transaction markets is consistent with some of the case law in the United States.29 More precisely, although relevant markets are usually defined as comprising substitutable products, single markets have also been defined in cases of complementarities in demand, for instance in banking services or in hospitals services.30 In United States v. Grinnel, the U.S. Supreme Court addressed the point explicitly:
The District Court treated the entire accredited central station service business as a single market, and we think it was justified in so doing. Defendants argue that the different central station services offered are so diverse that they cannot . . . be lumped together to make up the relevant market. For example, burglar alarm services are not interchangeable with fire alarm services. . . . We see no barrier to combining in a single market a number of different products or services where that combination reflects commercial realities. To repeat, there is here a single basic service—the protection of property through use of a central service station—that must be compared with all other forms of property protection.31
Although, as explained in Part II, two-sided markets are different from markets for complementary products, the argument above would seem to extend to two-sided transaction markets, where the “single basic service” is in fact facilitating the transaction.
The Market for Payment Cards
An interesting case of a two-sided transaction market is the payment card market. In such a market, two types of payment card systems can be identified: three-party payment card systems and four-party payment card systems.32 In the former system, a payment card company directly provides POS services to merchants and issues cards to buyers.33 In the latter system, banks and other members of a payment card association issue cards and provide POS services.34 Hence, in general, a cardholder buys card services from a bank, while a merchant buys them from another bank.35 As a result, when a cardholder buys a product from a shop, the transaction needs to be cleared at the level of the two banks. Membership of the payment card system implies a guarantee that the transaction is cleared, so that the merchant receives the payment of the cardholder. The system works similarly to traditional checks. However, while the latter have always been cleared at par (that is, at no price) and the costs of clearing have been charged indirectly to the buyer and the seller (for example, through higher fees on bank accounts), in four-party payment systems an interchange fee is often paid per transaction from one side to the other—in most cases from the bank of the merchant to the bank of the cardholder. Clearly, such an interchange fee constitutes a component of the marginal cost of serving customers on its side of the market for the bank that pays it—in most cases the bank of the merchant. As such, on the one hand, it constitutes a lower bound for the per transaction fee charged to the merchant for the transaction; on the other hand, it allows the price structure to be influenced. Most payment card associations traditionally set a default interchange fee, which is applicable unless the banks involved reach a different agreement. We will see in the next part that, as the use of payment systems has increased, the setting of interchange fees has become a typical case of concern for antitrust authorities. As a result of regulatory and competition policy interventions, payment card associations transformed themselves into payment card companies, giving rise to a fifth system participant. Hence, a distinction is often made between four-party and five-party systems.36
According to our suggestion, irrespective of whether the system is a three-party system or a four-party system, a single relevant market must be defined. Such a relevant market should include transactions taking place through that payment system and may or may not include transactions through other payment systems and possibly direct transactions. In such a relevant market, three-party and four-party systems may coexist if they are substitutable from the point of view of both the buyers and the sellers.
However, in a four-party system, it is possible to distinguish inter-brand (or, better, inter-system) competition and intrabrand (or, better, intra-system) competition.37 The former takes place when different payment systems compete for clients; the latter takes place when different banks compete to issue cards or provide POS terminals of a given payment system.
In a four-party payment system, a payment card association sells its card services through a bank or another financial institution. The situation is thus reminiscent of a market where different car producers compete for clients through retailers, which may sell cars of different brands. In such a market, traditionally both inter-brand and intra-brand competition have been assessed with regard to a relevant market defined on the basis of consumers' preferences.38 The difference is that, in the case of payment cards services, the market is two-sided and the payment card association needs to sell to both cardholders and merchants. Hence, inter-system and intra-system competition need to be assessed taking into account both sides of the market. This assessment can be carried out with regard to a relevant market, defined, according to our Suggestion 1 above, as the market for transactions that take place through the payment system.
We now discuss whether our Suggestion 1 is in line with past and current practice in cases involving two-sided platforms. As already mentioned, among two-sided transaction markets are those for payment cards services, auction houses, video game consoles, and operating systems. Media markets, on the other hand, are two-sided non-transaction markets. We discuss here cases relating to both types of markets, starting with transaction markets.
There are a number of recent cases in which the competition authorities correctly defined only one relevant market when platforms operating in two-sided transaction markets were involved.
In Travelport/Worldspan,39 the Commission explicitly acknowledged the two-sided nature of the market by noting that a Global Distribution Service40 is a two-sided platform allowing Travel Service Providers and Travel Agencies to interact.41 The European Commission defined only one relevant product market for “electronic travel distribution services through a GDS,”42 which again appears correct given that the GDSs are two-sided transaction platforms.43
Although the Commission did not reach a definitive conclusion as to the exact scope of the relevant product market44 in Google/Doubleclick45 it correctly identified the two-sided nature of some online advertising intermediation services.46 The Commission did not envisage the existence of separate markets for advertisers and publishers. On the contrary, the Commission assessed the reality of an intermediation market, which it considered separate from the direct sales channel. Even though the decision does not provide an explicit explanation for why a single market for intermediation services was defined, the definition of a single market appears correct in light of the transactional nature of these intermediaries.
Although it correctly defined a single market for online intermediation, the Commission missed the point that other two-sided markets were at stake. The Commission considered the online advertising market, but did not recognize that the websites hosting such advertising, including those run by Google itself, are two-sided non-transaction platforms that cater not only to advertisers, but also to website viewers, and in some instances to content providers.47 Arguably, the Commission should have also defined one or more non-transaction markets for users of (Google) websites,48 or at least discussed why it thought that such additional markets were not relevant to the decision. Had it done so, it might have realized that the acquisition of Double-Click potentially allowed Google to acquire information on online users' data that could be used to improve targeting of advertising also on its own websites. Interestingly, such an issue was raised in the dissenting statement of Commissioner Jones Harbour in the decision of the Federal Trade Commission on this merger.49
At the level of national competition authorities, the most explicit discussion of whether one or two markets need to be defined is found in Bloemenveiling Aalsmeer/FloraHolland.50 In this case, the Dutch competition authority, NMa, defined only one relevant market for the trade in ornamental horticultural products, which comprised both sales through the auction houses and sales through alternative channels including direct sales. The NMa first explicitly acknowledged the two-sided nature of the market. It also recognized that it was important whether the two sides of the market for auction services for ornamental horticultural products were considered to constitute separate relevant product markets or whether they were to be included in a single relevant market. Interestingly, the NMa then referred to the manner in which the relevant market was defined in Visa International to justify its decision to define a single market.51 As in the case of credit card networks, where the European Commission recognized that demand from both consumers and retailers must be taken into account (since the choice of payment method is determined by both sides jointly), the NMa considered that, for auction services of ornamental horticultural products, both the demand from buyers and the supply by growers needed to be taken into account when determining demand for the services offered by auction houses and defining the relevant product market. Even though some of the NMa's arguments would seem to imply that for all two-sided markets a single market encompassing all sides should be defined, the definition of only one relevant market seems again correct in light of the fact that auction services are a two-sided transaction market.
The Market for Payment Cards
While in the above discussed cases involving two-sided transaction markets only one relevant market was defined, in cases involving four-party payment card schemes—unlike earlier decisions in the United States and Europe that correctly defined a single relevant market—competition authorities in Europe defined more than one market.
Long before the theory of two-sided markets had been developed, the District Court for the Southern District of Florida, in National Bancard Corp. v. VISA,52 defined the relevant market in which Visa operated to be the market for all payment systems. More recently, however, in United States v. VISA U.S.A.,53 the District Court for the Southern District of New York and the Court of Appeals of the Second Circuit54 defined two relevant markets—a market for credit cards and a market for credit card network services, the latter including, for example, authorization, settlement, and clearance of transactions.
In Visa International55 and Visa International-MIF,56 the Commission appears to have correctly defined a single market for transactions by payment cards between cardholders and merchants based on both merchants' and cardholders' preferences,57 and mainly discussed inter-system competition with regard to this “upstream” market.58 In these cases, the Commission also identified two-additional “downstream” markets: the issuing and acquiring markets,59 where “intra-system” competition takes place. Although intra-system and inter-system competition could have both been assessed with regard to the single market for transactions, the Commission's approach would have been consistent with established practice in cases involving aftermarkets if it had explicitly claimed that: (1) the issuing and acquiring markets were the primary (interrelated) markets and the single market for transactions was the secondary market; (2) primary and secondary markets needed to be considered separate relevant markets.
Unfortunately, in the more recent MasterCard60 and Visa Europe cases,61 the Commission substantially changed its approach to market definition. Again, it distinguished two types of markets. However, this time, the “upstream market” or “network market” was identified with the market in which competition between different payment systems takes place to persuade financial institutions (rather than merchants and cardholders) to join their network and on which they provide services to these institutions. The Commission considered this the “upstream market” or “network market,” where inter-system competition takes place. The other was, as in the previous decisions, the market in which competition between financial institutions (usually banks) for card-related activities (issuing and acquiring) takes place. Within this downstream market, the Commission defined two relevant markets, one for acquiring and one for issuing services.62
Notably, in MasterCard, the Commission recognized the two-sided nature of the market but explicitly refused to define a single market encompassing both card-holders and merchants, stating that: “Two-sided demand does not imply the existence of one single ‘joint product’ supplied by a ‘joint venture.’”63 The argument was threefold. First, the definition of a single market would not be able to take into account the complex vertical structure of the market and would not therefore be suitable to assess intra-system competition.64 Second, such an approach would not acknowledge that the relevant product is not only payment services, but also other acquiring and issuing services.65 Third, such an approach would be inconsistent with previous practice in cases concerning two-sided platforms, such as newspapers.66
In our opinion, all three arguments are unconvincing. First, the Commission's desire to be consistent with previous practice in the newspaper market overlooked the fact that while payment cards are transaction markets, newspaper markets are two-sided non-transaction markets. As we argued above, this difference warrants a differential treatment in market definition. Second, the additional issuing and acquiring services (such as accounting records on sales by payment cards offered to merchants) are mostly ancillary to the main service of allowing the transaction. Third, and more important, by refusing to define a market for transactions, the Commission completely misrepresented the “complex vertical structure of the market.” In fact, the Commission seems to propose a structure of the market similar to the five-party system described in Figure 5, but without the final transaction among end-users. In practice, when defining the relevant market, the Commission focused only on the adoption decision, assuming it is made by banks and not by customers, and completely disregarded the importance of the decision by end-users to use a card for a transaction.
The consequence of focusing on banks rather than end-users was that the Commission failed to discuss, for instance, whether end-customers multi-home in adoption67—that is, whether cardholders have more than one card or merchants have more than one POS terminal. The discussion would have been relevant because, to the extent that customers multi-home in adoption, the adoption of the system by a customer does not guarantee that the system is used and generates revenues from such use. In such a situation, identifying the market for adoption as the market in which competition takes place is misleading. The case is similar to that of telecommunications operators which offer phone services, potentially competing to sign up customers and to make customers phone through their system (only here the market is two-sided). In the presence of customers' demand for multi-homing in adoption, for example, if cardholders desired more than one card or people subscribed to different operators, then competition for having customers use the system would be fierce.68 In such a situation, failing to define a market for use implies failing to take into account major competitive constraints faced by the firm.
By focusing on the adoption decision, the Commission could not properly assess whether four-party systems such as Visa and MasterCard competed for use against each other, against three-party systems such as American Express, or against national debit card schemes.
Finally, focusing on the adoption decision and disregarding the usage decision appears at odds with the fact that most of the revenues of payment cards derive from transactions, and ignores that the central issue of the case was the interchange fee, which is a per transaction fee. Implicitly, the Commission argued that such a fee affected pricing decisions by banks (for example, the merchant discount) without affecting end-users choices.
Similar to the practice at the EU level, in the MasterCard case,69 the U.K. Office of Fair Trading (OFT) explicitly refused to define a single market encompassing both cardholders and merchants, noting that,
“[the] ultimate demand for credit and charge card transactions is two-sided, from merchants on one side and from cardholders on the other side . . . . However, the OFT does not accept the view of the Parties that the presence of two-sided demand makes it necessary to consider a single relevant market based upon the card scheme as a whole.70
Hence, while the OFT defined one market for acquiring services and one for issuing services, it did not define a network market.
In contrast, the OFT defined a third market for the supply of services provided between issuers and acquirers for the completion of purchases made using MasterCard. Such a market was called a “wholesale” market. In practice, the OFT argued that, in a four-party payment system, there is a market for the clearing of a transaction and related services between the issuing and the acquiring bank. The OFT considered the interexchange fee as the price an acquirer pays to an issuer for the latter to provide “the services necessary to complete the transaction.”71
In our opinion, there is no such market because once the transaction by the payment card of a given system has taken place, one of the banks involved holds a credit, the other holds a debit, and only these two banks can clear the transaction. In this respect, the OFT's argument that the case is not different from that of other wholesale markets where a contract for inputs has already been agreed upon is misleading: in the market under consideration, the demand for inputs (that is, for clearing services) is not only derived from the demand for output (that is, for transactions), but is also conditional on the output being sold (that is, on the transaction taking place) and, more important, the choice of the input supplier is determined by the parties that transact. The case is analogous to that of a customer who buys a car from a car manufacturer but himself determines from whom the manufacturer should purchase the inputs. In the absence of a previous agreement, the sale of the car would lock the manufacturer in a bargaining game with the suppliers of the inputs. The implied uncertainty would prevent the manufacturer from selling the car in the first place, particularly if inputs suppliers were so numerous that ex ante bilateral contracts with all of them were unfeasible. This is the situation in a payment cards system when the payment card association sets a fallback interchange fee that is valid in the absence of bilateral contracts between the issuing and the acquiring banks.72 Interestingly, a similar conclusion with regard to the absence of such a wholesale market was reached by the Commission in Visa International MIF.73
Overall, it would seem that, in most recent cases involving two-sided transaction markets, competition authorities have defined one single market, encompassing both sides of the market, thus in line with our Suggestion 1. It is only when four-party payment card schemes are involved that competition authorities fail to define a single market. This appears to be due to the complex vertical structure of the market, and also to the desire to be consistent with previous practice. However, it has resulted in the failure to take into account the role of cardholders and merchants in the choices of whether to use a payment system and of which payment instrument to use.
We now turn to the observed practice in cases involving two-sided non-transaction markets. As claimed above, in media markets such as newspapers or TV, which are two-sided non-transaction markets, two interrelated markets—one for each side—should be defined.
There have been some cases in which the competition authorities correctly defined two separate markets. For instance, well before the theory of two-sided markets was formulated, the Supreme Court, in Times-Picayune Publishing Co. v. United States, acknowledged that “every newspaper is a dual trader in separate though interdependent markets; it sells the paper's news and advertising content to its readers.”74 This decision set a precedent and is cited, for instance, in Community Publishers, Inc. v. Donrey Corp., where both a market for readers and a market for advertisers was defined and considered interdependent.75 The court ruled that “[i]n this case, the relevant product market for antitrust purposes is the local daily newspaper. This market is in fact composed of two markets: one for readers and one for advertisers.”76
Turning to Europe, in the case GIMD/Socpresse,77 the Commission defined a market for readers and a market for advertisers. Although the exact scope of the relevant product market on the advertising side was left open, on the readers' side, two relevant markets were considered to be affected by the merger—the market for general and political magazines and the market for economic and financial magazines.
Turning then to national competition authorities, in both the Holtzbrinck78 and the Springer/ProSieben/Sat179 cases, the Bundeskartellamt defined two different markets—one for readers and one for advertisers. Interestingly, in both cases, the scope of the relevant product market was different on each side. In Holtzbrinck, the market for readers was defined as the market for daily regional subscription newspapers in Berlin.80 In Springer/ProSieben/Sat1, the Bundeskartellamt made a distinction between over-the-counter and subscription newspapers, and defined the readership market as the national readership market for over-the-counter newspapers.81 In both cases, the market for advertisers was defined more broadly, as including both over-the-counter and daily subscription newspapers, as well as advertising-only publications. As discussed above, one of the reasons behind our suggestion to define two separate markets in case of two-sided non-transaction markets is the possibility that a product may compete with those of a two-sided platform on one side of the market but not on the other side.
Nevertheless, there have also been cases involving two-sided non-transaction markets where competition authorities failed to define a relevant market on each side. The main rationale for not defining a relevant market on both sides appears to have been the observation that the product is given away for free on one side of the market, and the claim that, as a consequence, there is in fact no market on that side.
For instance, in KinderStart v. Google,82 the Court for the Northern District of California reasoned that the claimants “cite[d] no authority indicating that antitrust law concerns itself with competition in the provision of free services. Providing search functionality may lead to revenue from other sources, but KinderStart has not alleged that anyone pays Google to search,” hence, “the search market is not a ‘market’ for the purposes of antitrust law.”83
Similarly, in BSkyB/Kirch Pay TV,84 the Commission defined two product markets—the market for pay-TV and the market for interactive digital TV—without distinguishing however the advertisers' and viewers' sides. More generally, it did not recognize the two-sided nature of these two markets and focused only on substitutability between the products for viewers. The main reason was the claim that PAY-TV stations did not sell advertising slots to advertisers.
While it is by now clear that PAY-TV stations can sell advertising, the claim that there needs to be a positive price (the so-called “trade relationship” in the language of the Commission) for the effects of a merger on that side of the market to be considered seems to have started a trend in decisions in the European Union regarding media markets. For instance, in Archant/Independent News and Media,85 the U.K. Competition Commission did not identify two distinct markets, namely a readership market and a market for the sale of advertising space. In fact, it did not define any readership market but focused solely on defining a market from the perspective of advertisers.86
In Springer/ProSieben/Sat1, the Bundeskartellamt only defined a TV advertising market and concluded that there was no market for private free-to-air (FTA) TV viewers to be defined. The argument was that it is only in the case of Pay-TV channels that there is a viewer market, as only there do viewers actually pay for receiving the program. Conversely, there is no viewer market for private FTA TV channels, which are financed through advertising.
Yet, if one side does not pay at all, as in the business model of free-to-air TV, the market is still two-sided. In fact, as we argued above, it is exactly because the market is two-sided that one side does not pay. As a result, it is clearly not possible to analyze pricing decisions and, more generally, firms' behavior on one side while disregarding the other side. In addition, choosing a priori to disregard the other side of the market, by not defining a relevant market for that side, implies that the market is defined as comprising only platforms that choose to charge the same side of the market. For instance, in Springer/ProSieben/Sat1, the claim that there is no relevant market if there is no price implies that the relevant market includes only FTA TV, without asking the relevant questions of whether FTA TV and pay-TV are substitutes for viewers—that is, without fully assessing product substitutability.
A similar issue arose in News Corporation/Premiere,87 in which the Commission only defined a market for the provision of pay-TV services to viewers, but failed to define an advertising market. The argument was based on the claim that pay-TV operators still largely relied on “revenues stemming from subscription fees and, to an insignificant extent, from advertising.”88 The claim that there is no relevant market to be defined on the advertising side because a pay-TV operator sells only limited advertising time implies that the relevant market can include only FTA TV, without asking the relevant questions of whether a pay-TV operator could at some point find it profitable to increase its share of advertising financing.
We will come back to the issue of market definition in media markets in the next part, discussing why such an ex ante assumption is wrong. For the moment, we highlight that, for all the reasons discussed above, two relevant (interrelated) markets—one for advertisers and one for viewers—need to be defined in TV cases, irrespective of whether one of the two sides does not pay.89
CONSIDERING BOTH SIDES OF THE MARKET
When one defines two different markets—one for each side of a two-sided market—the issue arises of whether one should look at each side of the market independently or jointly—that is, whether one should consider the role of the indirect network effects when defining the markets. For instance, in a merger among newspapers, the question is whether one should look at the advertising side when defining the relevant market for readers, and vice versa. The question also arises whether one should look at both buyers and merchants when one defines the market for (transactions by) payment cards.
We argue here, as also suggested by David Evans and Michael Noel, that it is necessary to take all sides of the platform under consideration.90 A platform in a two-sided market needs both sides “on board” and it is therefore competing for customers on both sides. More important, how much competition a platform faces in getting customers on one side also depends on its competitive position on the other side, and vice versa. We now explain why.
It is well known in the economic literature that product differentiation, whether vertical or horizontal, relaxes price competition.91 Similarly, on each side of a two-sided market, the degree of competition faced by a given platform depends on the degree of vertical and horizontal product differentiation on that side. For example, the level of competition faced by a newspaper on the advertising side depends inter alia on the number of its readers compared to other newspapers. For instance, if a newspaper has many more readers than its rival, one can expect a similar price increase on the advertising side to lead to a smaller loss in advertising than if the newspapers were closer to each other in terms of copies sold. One can argue that from the advertisers' point of view newspapers are vertically differentiated in the number of readers.92
Moreover, the level of competition faced by a newspaper on the advertising side is also likely to depend ceteris paribus on the demographic composition of its readers with respect to that of the readers of rival newspapers. Different advertisers might value some demographic groups of readers more than others. If so, newspapers can also be perceived as horizontally differentiated on the advertising side. In this case, horizontal differentiation relaxes price competition. In the extreme case, if the newspaper under consideration had all the “valuable readers” for some advertisers and the rival had no “valuable readers,” a given price decrease for advertising in the rival newspaper would not lead to any of these advertisers switching to the rival.
Both forms of differentiation should therefore be taken into account when looking for the set of products that exert competitive pressure on the product or products of the firm or firms of interest.
However, both horizontal and vertical product differentiation can be endogenous in a two-sided market. Consider again a newspaper market. We previously explained that from the point of view of advertisers, newspapers are likely to be vertically differentiated (because they have a different number of readers) and horizontally differentiated (insofar as they have different types of readers). But both the number and the type of readers also depend on the price charged to readers and—if readers are affected by advertising—on the price charged to advertisers, which determines the quantity of advertising in the newspaper. In addition, the number and type of readers is likely to be determined by the content of the newspaper. For instance, a sports newspaper might have a different number and a different type of reader than a financial newspaper, even if they charge the same cover price. Thus, product differentiation on one side (the advertising side in our example) may depend not only on prices, but also on product characteristics on the other side (the readers' side in the example above).93 In other words, horizontal product differentiation on one side is linked to both the price charged and horizontal product differentiation on the other side. Hence, the competitive constraints faced by a platform can be assessed only by taking into account both sides when defining the relevant market.
The risk of neglecting one side of a two-sided market is particularly high when the product on the overlooked side is priced at zero. In such a case, one may, at first sight, think that firms are not competing on that side. For instance, one often thinks of shopping malls as renting space to retailers, failing to see that they also compete for shoppers against other malls. Yet if a shopping mall experienced a drop in the number of shoppers, it is likely that this would lead to a drop in demand for space from the shops. The mall might then be forced to lower the price charged to shops, or experience a decrease in the number of shops and in corresponding revenues.
Similarly, one might think that phone directories compete only on the advertising side. However, if a phone directory experienced a drop in advertisers, it would likely suffer not only a direct drop in profits, but also an indirect drop in usage due to people finding less information in the directory. The latter would be due to the directories competing against each other for usage.
By failing to consider all sides in the definition of the relevant market, one would then ignore the real competitive pressure faced by the merging firms before and after the merger. This may result in a flawed market definition and ultimately in a mistaken decision.
It is only in the particular case of a two-sided non-transaction market with only one externality that one could safely perform a market definition exercise on that side of the market irrespective of the other side. For example, if one were to find that advertising has no effect on the readers' side of the market, one would need to take into account the advertising market when defining the readers' market, but could safely define the advertising market irrespective of the readers' market. In that case, whatever the choices of publishers on the advertising side, they will not affect the readers' side. Hence, the platform on the advertising side of the market will not behave differently from a firm in a single-sided market.
More generally, while one always needs to consider both sides in a two-sided transaction market, it is not always necessary to consider both sides when defining the relevant market in the case of a two-sided non-transaction market. In fact, in the latter case, it is only necessary to consider all other sides toward which the side under consideration exerts an externality, either directly or indirectly.94
To sum up, our second suggestion is
Suggestion 2: In two-sided markets, competition authorities should take into account both sides of the market when defining the relevant market, even in the case of two-sided non-transaction markets, when they should define two interrelated markets.
It is only in a two-sided non-transaction market with only one externality that one could safely perform a market definition exercise on the side of the market that exerts no externality on the other without considering the other side.
In most cases involving two-sided transaction markets, competition authorities correctly have taken into account competitive constraints on both sides of the market when defining a single relevant market.
An exception is Nabanco v. Visa in the United States, where indirect network effects or, more generally, externalities were not discussed. As already observed however, this judgment predates the development of the theory of two-sided markets.
The U.S. Visa and MasterCard judgment was, by contrast, one of the cases that helped develop the theory of two-sided markets by focusing the attention of economists on the payment cards market. In the decision, the two-sided nature of the market was not explicitly recognized, but externalities did play a role in the court's analysis. Indeed, it was claimed that “the merchants' demand for general purpose cards is derived from consumers' demand to use these cards.”95
The European Commission also recognized the need to take into account both sides of the market, even as it failed to define a single market encompassing both sides. Most notably, in Visa International-MIF, even though the decision did not explicitly mention that the market is two-sided, the European Commission recognized that the market is characterized by network externalities. It then argued that
the usage of different payment systems (and thus market shares) is determined by the inter-related decisions of consumers and merchants; . . . Consequently, in order that two different payment instruments be considered as substitutable and therefore included on the same relevant inter-system market, they must be substitutable for both consumers and merchants. If one or the other user of payment instruments considers two different payment instruments as not substitutable, then those two instruments are not substitutable on the inter-system market.96
In contrast, the Commission in MasterCard97 explicitly recognized the two-sided nature of the market and not only the existence of positive indirect network effects between issuers and acquirers.
Turning to national competition authorities, and to two-sided transaction markets other than the payment cards market, in Bloemenveiling Aalsmeer/FloraHolland, the NMa argued explicitly that, in defining the relevant market, “the buyers' side and the growers' side of the market must be assessed together. A change on one side of the market also has an effect on the other side of the market. The two sides of the market . . . cannot therefore be considered in isolation.”98
In two-sided non-transaction markets, however, the market definitions adopted by competition authorities are not always consistent with our Suggestion 2. While most decisions define two relevant interrelated markets, there nevertheless have been cases in which competition authorities forgot to discuss one of the two indirect network effects.
For instance, in cases involving newspapers or TV, while it is generally recognized that advertisers care about the number of readers a newspaper reaches or the number of viewers a TV channel has, competition authorities often do not discuss whether readers or viewers like, dislike, or are indifferent to advertisements. In practice, this is equivalent to an implicit assumption that readers or viewers are indifferent to advertising. This omission seems particularly relevant for the TV market, where it is well known that viewers dislike advertising.99
In BSkyB/Kirsch Pay TV, the European Commission did not discuss whether viewers like or dislike advertising, It thus made the implicit assumption that viewers are indifferent to advertising. In comparing pay-TV and FTA TV, it correctly noted that “[t]he fact that subscribers are prepared to pay considerable sums for pay-TV indicates that the latter is a distinguishable product with specific extra utility.”100 However, it did not ask the relevant question of whether this extra utility is due to higher quality of the non-advertising content or to the lower level of advertising. In other words, the European Commission did not take into account the possibility of a nuisance cost of advertising.101
Similarly, in Springer/ProSieben/Sat1, the Bundeskartellamt recognized that the number of readers is important for the demand from advertisers,102 but did not explicitly address the issue of the readers' attitude toward advertising. The Bundeskartellamt also recognized that the number of viewers, both in FTA TV and pay-TV, determines the channel's position in the TV advertising market. The Bundeskartellamt correctly claimed that the number of viewers was less important for public TV stations, as they are financed through license fees and face time limitations for advertisements. Again, it did not address the issue of the viewers' attitude toward advertising.
The Bundeskartellamt, however, took a different approach in Holtzbrink, where it considered both indirect network externalities to be positive. It assumed that more advertising enhances readership and argued that readers value advertisers not only per se, but because more advertising leads to higher revenues for the newspaper, which in turn allows the newspaper to offer higher quality journalism,103 leading to a so-called “advertisement-circulation-spiral.”104
In the United States, in Community Publishers, Inc. v. Donrey Corp., the court also assumed that readers attach a positive value to advertising; such value however would derive directly from the content of the advertisment rather than indirectly through a higher quality of the newspaper.105
In GIMD/Socpresse, although the European Commission took into account the fact that there are two customer groups, readers and advertisers, it did not explicitly recognize that magazines are two-sided platforms. The decision argued that advertisers chose a magazine depending on which type of readers they want to target, but ignored that the number of readers is also relevant for advertisers.106 In addition, it also argued that advertisers care about the image of the newspaper and that the image depends on the content of the newspaper. The European Commission thus claimed that the market could be defined on the basis of newspaper content on the advertising side as well.107
In fact, as we discussed above, the sign and size of the indirect network effects are crucial to understand the competitive constraints faced by a given firm on either side of the market. Disregarding such indirect network effects and analyzing one side of the market alone might lead to a wrong decision.
The failure to take into account properly the links between demands on the two sides can also be observed in the fact that many competition authorities distinguish between pay-TV and FTA TV based on the fact that the two adopt different business models.
For instance, in BSkyB/Kirch Pay TV, the European Commission already claimed that pay-TV and FTA-TV constitute separate product markets.108 It highlighted that the definition of these two separate markets was consistent with its past practice.109 In addition, it justified the distinction by noting that in the case of TV financed by advertising, a trade relationship was established only between the program supplier and the advertiser, while in the case of pay-TV a trade relationship is established between the supplier and the viewer via a subscription. According to the Commission, these different trade relationships lead to different conditions of competition.110
Also in News Corporation/Premiere, the European Commission reasoned that pay-TV and FTA TV were still distinct markets.111 In support of this market definition, it argued that pay-TV and FTA TV are financed differently and that it made a difference whether viewers had to subscribe to services in order to get access to programs, as in pay-TV, or whether programs could be received at no specified cost, as is the case for FTA TV.112 Only as a last reasoning in support of its market definition, the European Commission referred to the lack of substitutability between the two TV services for viewers, due to the difference in content.113
As already mentioned, we believe that concluding that pay-TV and FTA services constitute two distinct product markets is misguided: whether one side pays or the other (in the case of advertising-financed TV advertisers pay, and in pay-TV mainly viewers pay) should not determine the relevant market(s). FTA and pay-TV simply constitute different business models. As such, they represent a choice of the firm and not a feature of the market itself. This is even more evident now that digitalization has made encryption of TV signals easier, thus allowing for the exclusion of viewers who do not pay.114
We do not claim that pay-TV and FTA are substitutes for viewers. Rather, we argue here that the choice of financing mechanism is not linked to demand substitutability. It is only if pay-TV and FTA TV are not substitutes for viewers that the business model might be relevant to assess supply substitutability as claimed in News Corporation/Premiere.115 However, the arguments used in News Corporation/Premiere to justify the lack of substitutability between FTA TV and pay-TV for viewers are not fully satisfactory. It might indeed be the case that the type of content and program schedules differ between pay-TV and FTA services (where, for instance, premium content is often first broadcasted via pay-TV) so that pay-TV and FTA TV are not fully interchangeable from the viewers' perspective. Nevertheless, to support the claim that the two are not substitutes, one would need an empirical assessment of the relative importance that viewers attach to content (including lower advertising) versus price.116
In any case, even if FTA TV and pay-TV turned out not to be substitutes for viewers, arguably, one should still take into account the other side of the market, even in a case involving pay-TV stations. In fact, absent binding regulatory caps on advertising, the decision by a pay-TV station whether to rely on advertising financing or not, and the amount of such advertising, is likely to depend on an assessment of the costs and benefits of advertising financing. The costs would then depend on the degree of advertising aversion of TV viewers (that is, the strength of the indirect network externality from advertisers to readers) while the benefits would depend on the degree of competition on the advertising market for the viewers of pay-TV. For instance, if an advertiser wanted to reach viewers that watch both pay-TV and FTA TV (possibly because pay-TV is “complementary (rather than substitutable) to FTA TV,” as mentioned in News Corporation/ Premiere117), then pay-TV could be competing with FTA TV for advertising to those viewers. Vice versa, if viewers did not watch both pay-TV and FTA TV, then pay-TV would not be competing with FTA TV for advertising to those viewers.
One decision is worth mentioning in this respect. In BSkyB/ITV,118 the U.K. Competition Commission recognized that “the parties operate in a two-sided market. This means that the companies serve two distinct sets of customers—viewers and advertisers—which are interdependent.”119 Furthermore, the Competition Commission noticed that “the parties [however] operate different business models, with BSkyB deriving the majority of revenues from subscription fees and ITV from advertising. The BBC operates a third business model, relying primarily on revenue from its license fee.”120 Interestingly, and differently from the European Commission's analysis, the U.K. Competition Commission did not base its market definition on the difference in business models. Rather, it more correctly based its market definition on the viewers' ability to substitute between different retail offers. The U.K. Competition Commission acknowledged that the parties offered differentiated products.121 But, it correctly noted that
observed price differences are not necessarily an indication of a lack of substitutability in any market. In two-sided markets suppliers can compete with one another at different price points, given the ability to generate revenue in two separate markets. For example, FTA services may compete directly for viewers with pay services, with higher viewing figures indirectly generating higher advertising revenues.122
Accordingly, the U.K. Competition Commission defined the relevant market on the viewers' side as the U.K/ market for all TV, which included both pay-TV and FTA services.123 The Competition Commission concluded that FTA services provided a constraint to a monopolist of pay-TV attempting to raise prices. However, it noted that it is likely that the FTA offering is a closer substitute to a pay-TV package that includes only basic channels than to a pay-TV package that includes basic and premium channels. Thus, FTA services may be a relatively weaker constraint on those packages, which include premium channels.
Still, even in this decision, the Competition Commission did not take into account that viewers might dislike advertising. Hence, when discussing product differentiation, the U.K. Competition Commission missed the point that TV channels might also be differentiated on the amount of advertising broadcasted.124 As a result, once again, the role of the second indirect network effect was not taken into account.
To conclude, in most cases involving two-sided non-transaction markets, competition authorities do not always correctly take into account the interrelatedness of the two sides, and thus fail to consider all competitive constraints faced by a two-sided platform. As we will see in the next part, this also manifests itself when the SSNIP test is used to assess the relevant market.
THE SSNIP TEST AND THE HM TEST
A commonly used tool for market definition in a traditional single-sided market is the so-called small-but-significant-non-transitory increase-in-price test (SSNIP test), which defines the market as the smallest set of substitute products such that a substantial (usually five or ten percent) and non-transitory (usually one year) price increase by a hypothetical monopolist would be profitable. Starting from a set of candidate products for the relevant market, the SSNIP test is implemented by first simulating a given price increase above the current level125 by a hypothetical monopolist who owns just one product,126 and as long as that leads to estimated losses in profits, progressively increasing the number of products owned by the monopolist. When profits are not estimated to decline following a small but significant increase in price by the hypothetical monopolist, the set of products owned by the monopolist in the last simulation constitutes the relevant market.
The SSNIP test is often performed using Critical Loss Analysis (CLA), for which formulas are derived under the assumptions of constant marginal costs and either linear or constant elasticity of demand.127 Under these assumptions, performing a CLA is exactly identical to performing the SSNIP test.128
In any case, the idea behind the SSNIP test (and thus CLA) is that if the small but significant non-transitory increase in price is unprofitable, then there exists at least one close enough substitute to the product whose price is raised. If so, the two products should be in the same relevant market (and so on and so forth). Thus, both the SSNIP and CLA set an implicit benchmark for substitutability between products to be in the same relevant market.
In addition, the iterative procedure described above is designed to ensure that a relevant market is defined as the smallest set of substitute products on which a monopolist would find it profitable to increase prices by a small-but-significant amount, and therefore makes sure that the market is designed in such a way that a monopolist has market power, which is a basic requirement of economic theory.
Interesting issues arise when attempting to extend the SSNIP test to a two-sided market. First, given that in a two-sided market firms set two prices, one on each side of the market, the question is which price the hypothetical monopolist should be raising. Second, given that in a two-sided market there are indirect network effects between demands (and therefore profits) on the two sides, the issue is whether one should consider profits on one or both sides of the market.
Some articles—such as Evans and Noel's two articles published in 2005 and 2008, Evans's article published in 2009, and Renata Hesse's article published in 2007—warned against the application of a one-sided SSNIP test in defining markets when two-sided platforms are involved.129
The first reason is that, in a two-sided market, the traditional SSNIP test cannot be applied as it is usually conceived.130 As already noted, market definition should account for both sides of the market in order to correctly assess the competitive constraints faced by firms. The logic of the SSNIP test should thus be extended (and therefore the formulas for CLA) in order to account for the indirect network effects between the two sides of the market when judging the profitability of a price increase.
Considering a two-sided platform with sides A and B linked by positive indirect network effects, the application of a one-sided SSNIP test on side A would only account for the direct effect that a price increase will have on the demand and profits of side A. It will not account for the fact that a reduction of the number of customers on side A is likely to lead to a reduction of the number of customers on side B such that, if the price on side B is kept constant, there would be a loss in profits also on side B. It would also not envisage the fact that the smaller number of customers on side B will in turn reduce the demand of side A, and so on. Hence, it would also underestimate the loss in profits on side A.
Positive indirect network effects between the different sides of the platform reduce the profitability of any price increase. As there is always at least one positive indirect network effect, the risk of applying a standard SSNIP test, which does not account for feedback effects, is that in such cases the market will be defined too narrowly.
The second reason why the test should be modified is that, if one wants to use a SSNIP test (or CLA) in a two-sided market, one should follow the original rationale of the test: defining the market as the smallest set of products on which a monopoly would find it profitable (or profit-maximizing) to exercise market power by non-temporarily raising the price above the current competitive level (at least) by a small but significant percentage.131
In order to ensure that the test is based on the same rationale, the SSNIP test in a two-sided market should take into account the changes in profits on both sides of the market and all feedback between demands on the two sides of the market following the hypothetical monopolist' raise in price.
In a market characterized by a transaction between end users (for example, in the payment card market), the SSNIP test should be implemented by raising the price level (that is, the price of the transaction), allowing the monopolist to optimally adjust the price structure (that is, the ratio between the prices paid for a given transaction by the two sides).132
In a market without a transaction among end users (for example, in a media market), the test should instead be implemented by raising first the price on one side of the market, and then the price on the other side of the market, each time allowing the hypothetical monopolist to optimally adjust the price structure.133 Only if the market were found to be characterized by a single externality, then the traditional SSNIP test and single-sided formulas for CLA could be applied to define the market on the side that does not exert an externality on the other.
Note that while there is consensus in the literature on the fact that one should take into account changes in profits on both sides of the market and all feedback between demands on the two sides, it is by contrast subject to debate whether the hypothetical monopolist should be allowed to optimally adjust the price structure.134
In particular, Evans and Noel suggest new formulas to perform CLA in a two-sided market when one wants to define two interrelated markets.135 These formulas are derived under the assumption of linear demands and constant marginal costs, as is usually the case in single-sided markets, but also under the additional assumption that the hypothetical monopolist raises the price on one side while keeping the price on the other side fixed.
On the contrary, both Emch and Thomson136 and Filistrucchi137 claim that the hypothetical monopolist should be allowed to optimally adjust the price structure (that is, roughly, the ratio of the two prices) when it is asked to raise the price on either side or the price of the transaction. They point out that a real monopolist would adjust the price structure when asked to raise the price, so that if one wants to know whether a hypothetical monopolist in the market would find it profitable or optimal to raise price on one side by a given amount, then one should allow it to optimally adjust the price structure.
In addition, Filistrucchi138 highlights that the logic behind the traditional SSNIP test is to define a market as the smallest set of substitute products on which a monopolist would find it profitable in the EU (or profit-maximizing in the United States) to increase prices by a small-but-significant amount. It is therefore designed to ensure that the market is defined in a way that a monopolist has market power, which is a basic requirement of economic theory. In order to maintain the same rationale when dealing with two-sided markets, one should allow the monopolist to optimally adjust the price structure. Filistrucchi also proposes some CLA formulas to perform the SSNIP test in a two-sided non-transaction market under the usual assumptions of linear or isoelastic demand and constant marginal costs.139 He then argues that, while using the standard single-sided CLA formulas would lead to the definition of a relevant market that is too narrow, adopting the formulas proposed by Evans and Noel would lead to the definition of a market, which is too large.140 Indeed, not allowing the price structure to be adjusted optimally would overestimate the loss in profits due to the increase in prices, because by definition the optimal adjustment by the hypothetical monopolist will tend to reduce such a loss.
These reasons lead to our
Suggestion 3: In two-sided markets, competition authorities should employ a modified version of the SSNIP test (or of Critical Loss Analysis) that takes into account the two-sided nature of the market.
Ideally, in both cases one should allow the hypothetical monopolist to adjust the price structure.
in a two-sided non-transaction market, one should check profitability of a rise in price on each side of the market;
in a two-sided transaction market, one should instead check the profitability of an increase in the price level (that is, the sum of the prices paid for the transaction by the two parties).
Whether a relevant market defined too widely or too narrowly leads to the wrong decision about a merger would then depend on the case itself and on whether the merger was cleared or blocked.
If a given merger is found not to raise competitive concerns in a market defined according to the single-sided formula, a fortiori, it would not be found to raise competitive concerns in the larger market defined according to the correct two-sided formula.
Similarly, if a given merger is found to raise competitive concerns in a market defined according to a two-sided formula that does not allow the monopolist to adjust the price structure, a fortiori, it would be found to raise competitive concerns in the smaller market defined according to the correct two-sided formula.
In the other cases—that is, when a merger is blocked in a market defined too narrowly or a merger is cleared in a market defined too widely—an incorrect market definition might lead to an incorrect decision.
In cases when dominance needs to be assessed in order to evaluate a possible abuse or to decide about regulation of an incumbent, if the firm is not found dominant in a market defined according to the single-sided formula, a fortiori, it would not be found dominant in the larger market defined according to the correct two-sided formula.
On the contrary, if the firm is found dominant in a market defined according to the single-sided formula, it might not be found dominant in the larger market defined according to the correct two-sided formula.
These arguments lead to our
Suggestion 4: In a two-sided non-transaction market, a single-sided SSNIP test can still provide evidence on the lower bound to the relevant market.
Both in two-sided transaction and non-transaction markets, a two-sided SSNIP test that does not allow the hypothetical monopolist to optimally adjust the price structure can instead provide evidence on the upper bound to the relevant market.
In practice, only a few decisions involving two-sided platforms explicitly mention that a SSNIP test was conducted and report the results.141
In the U.S. Visa and MasterCard case, the court only discussed whether consumers would switch away from credit cards following a SSNIP by a hypothetical monopolist, but did not discuss merchants' reactions to such a price increase. This was consistent with the claim, referred to above, that merchants demand was derived from their clients' demand. Hence the SSNIP test was performed on one side as if it were a single-sided market.
Turning to the European Union, the Commission in MasterCard explicitly refused to carry out a SSNIP test on the sum of the transaction fees charged to merchants and cardholders as suggested by Emch and Thomson. It correctly argued that this would have been inconsistent with its refusal to define a single market encompassing both sides.142 It also refused to perform a single-sided SSNIP test on the acquiring market, due to the understandable concern that a SSNIP test in a case which concerned an agreement between banks might be prone to the cellophane fallacy.143
In particular, in arguing against a SSNIP test on the sum of the transaction fees, the Commission explicitly criticized the article by Emch and Thomson, claiming the model was too simplistic. First, according to the Commission, using their suggested SSNIP test would imply assessing four-party payment cards schemes as if they were three-party schemes. Second, their proposed SSNIP test was based on constant marginal costs.144
With regard to the latter argument, it is sufficient to recall that the SSNIP test is usually carried out in practice using CLA, and that the formulas used are based on constant marginal cost (and either linear or constant elasticity demand).145
The claim instead that the model of Emch and Thomson is too simplistic is quite misleading. In fact, in order to be able to analyze whether, for end-users, payment cards of a three-party system are substitutes for payment cards of four-party systems, one needs a test which can be performed if both types of platforms are present in the relevant market. Let us consider the case of a car market, where brand A is vertically integrated and brand B is not. If the question is whether brand A and brand B are substitutes for consumers, it would not make sense to have a test that applies only to brand A and one which applies only to brand B. The SSNIP test in a single-sided market aims to measure substitutability irrespective of the market structure. A two-sided SSNIP test should do the same.
Finally, we note that the necessity to define only a single market for transactions and the need to run a SSNIP test on the sum of the end-users fees does not derive from the model of Emch and Thomson being right, but from the arguments we presented above in support of our suggestions. Their model should be seen as an illustration of the point.146
Turning to national competition authorities, in OFT MasterCard,147 consistent with its decision to define a wholesale market, an issuing market and an acquirer market, the OFT considered the implications of a SSNIP test for each of these markets. Somewhat surprisingly, to define the boundaries of each of these markets, the OFT chose to perform the test by raising the interchange fee. For the wholesale market, this is consistent with its view that there exists a market for the clearing of transactions between issuing and acquiring banks. However, the choice to perform a SSNIP test on the interchange fee in the issuing and acquiring markets seems to depart completely from the way in which a SSNIP test is performed in single-sided markets. Consider, first, the acquiring side of the market: for the acquiring banks, the interchange fee is a marginal cost for each transaction. Hence, assuming there is indeed a necessity to define separate issuing and acquiring markets, the SSNIP test the OFT applied is a test that raised the marginal cost rather than the price. Therefore, substitutability between products was not measured as a result of a change in price but as a result of a change in marginal cost. Unless the rate at which increases in the marginal costs are passed through to prices are equal to one, in practice this is a much different test from the usual one. A 5- or 10-percent increase in the interchange fee may be profitable, thus leading to a narrow market definition of the acquiring market, only because the pass-through to prices is much smaller than one. Consider then the issuing side of the market: as recognized by the OFT, the interchange fee is a marginal subsidy to the issuing bank for each transaction. Inexplicably, the SSNIP test the OFT applied is a test that raised the marginal subsidy rather than the price to cardholders.148 Clearly, raising the subsidy did not cause a loss in customers, thus leading to a narrow market definition. Last, but not least, even assuming the need to define separate markets, the suggested SSNIP test did not properly take into account the role of indirect network effects, neglecting feedbacks between the two sides of the market.
A notable exception among cases involving two-sided transaction platforms is the decision on Bloemenveiling Aalsmeer/FloraHolland, in which the NMa discussed at length the application of CLA to define the relevant market and discussed the effects of the two-sidedness of the market, although it did not apply a specific two-sided test.149 In particular, the NMa started by explicitly endorsing a CLA approach to the definition of the relevant market.150 In fact, it ran a survey of buyers and sellers of flowers and plants at the auction houses, in which it asked how they would react to a “deterioration of various parameters of competition.”151 It first calculated the critical losses.152 Then it calculated the actual losses.153 It finally compared, for each side of the market, the actual losses and the critical losses and found that the actual losses were higher than the critical losses, both for growers and buyers. From this, the NMa drew the conclusion that the relevant product market was larger than the one for sales of ornamental horticultural products through auctions.154
Importantly, the NMa acknowledged that the CLA formulas it used did not account for the two-sidedness of the market.155 It assessed the impact of indirect network effects on the CLA and correctly concluded that the actual loss was in fact higher than the one that was calculated.156 The NMa failed to note however that the critical loss formula would also change as a result of the two-sidedness of the market.157
More generally, although the NMa correctly realized that the two-sidedness of the market should play a role in the definition of the relevant market, it did not fully realize the implications, in this regard, of the transactional nature of the market. According to our discussion above, in a two-sided transaction market the test should have been conducted by raising the price level—that is, the sum of the fees charged to sellers and buyers, possibly using formulas for critical loss and actual loss that allowed the hypothetical monopolist to adjust optimally the price structure (that is, the ratio of the fees). The NMa instead discussed the implications of two-sidedness for the definition of the relevant market as if the market were a two-sided non-transaction market.
Turning to two-sided non-transaction markets, in BSkyB/ITV, the U.K. Competition Commission adopted the SSNIP test approach to the definition of the relevant market, focusing on the substitutability of products from the end-user perspective, as well as supply-side substitution.158 As already noted, the U.K. Competition Commission recognized the two-sided nature of the market. It also acknowledged that two-sidedness made the application of a standard SSNIP test difficult. Hence, it decided to “use the conceptual framework of the SSNIP test, relying as appropriate on a variety of qualitative and quantitative evidence” to define the relevant markets.159 However, neither the decision to include both FTA TV and pay-TV in the same relevant market, nor the decision to include also Video-on-Demand (VoD) in the relevant market, refer to the SSNIP test explicitly.160 Only when a further extension of the relevant market to Internet TV and DVD was considered, was the SSNIP test explicitly recalled, although the discussion was in fact mainly qualitative.161 As a result, in practice, a quantitative SSNIP test was not really carried out, and the two-sided nature of the market played a role only in the qualitative argumentation.
Also in Archant/Independent News and Media, the U.K. Competition Commission used a SSNIP test to define the relevant market. However, as already mentioned, in that case the two-sided nature of the market did not play a role in the market definition exercise, and only the advertising side was analyzed.162 Hence, the issue of whether the two-sided nature of the market, and in particular the presence of indirect network effects, should affect the test was not touched upon. According to our Suggestion 4, the U.K. Competition Commission correctly defined the relevant market for advertising only if the implicit assumption that readers are indifferent to advertising was correct. Otherwise the market was defined too narrowly.163
All in all, none of the competition authorities appear to have applied a specific two-sided market formula to perform the SSNIP test. This may be due to the slight disagreement in the literature with regard to the right two-sided formulas, but it is due more likely to the higher data requirements and the relatively higher complexity of two-sided formulas.
We reviewed the literature on market definition in two-sided markets. In our opinion, there appears to be a consensus on the fact that the two-sided nature of the market should play a role when defining the relevant market. We showed that, with some noticeable exceptions, this is increasingly recognized also by competition authorities.
Based on the economics of two-sided markets, we highlighted a crucial distinction between two-sided transaction and non-transaction markets. We then provided suggestions for the definition of the relevant market in cases involving two-sided platforms. We also analyzed whether the current practice of market definition in two-sided markets is consistent with the above suggestions.
First, consistent with the objective of market definition, we argued that, in a two-sided non-transaction market, one should define two interrelated markets, and, in a two-sided transaction market, one should define a single market. For the latter type of market, we showed that this is consistent with the practice of market definition in recent cases, but for the former type of market, there appears to be some confusion in the competition authorities' practice. In particular, we highlighted a worrying tendency to define only one market when customers on the other side do not pay. Typically, this has resulted in a number of cases where the relevant market may have potentially been defined as too small.
Second, we argued that one should take both sides of the market into account not only when defining a single market, but also when defining two interrelated markets. Unfortunately, only a few decisions, even when recognizing the two-sided nature of the market, have done this correctly. Again, this might have led in some cases to the market being defined too narrowly.
Finally, drawing from the economics of two-sided markets, we argued that, in such markets, the SSNIP test cannot be applied in its traditional form. Although the same logic underlying the traditional test can be extended to a two-sided setting, the test needs to be redesigned in order to take into account the two-sided nature of the market. Indeed, due to the interrelatedness of the demands on the two sides of the market, the profit of a hypothetical monopolist who raises the price in a significant non-transitory way on one side of the market is linked to the profit on the other side. Moreover, since in a two-sided market there are usually two prices and profit-maximizing prices are interdependent, it is not a priori clear whether the hypothetical monopolist should raise only one price, both prices, or the sum of the two (that is, the so-called price level). In addition, if one believes in the first or third option, it is not self-evident whether the hypothetical monopolist should be allowed to optimally adjust the ratio of the two prices (that is, the so-called price structure).
Therefore, we also addressed these issues. We claimed that there is a consensus in the literature on the fact that profits on both sides of the market should be taken into account. In addition, consistent with the need to define two markets, in two-sided non-transaction markets the SSNIP test should be implemented by raising first the price on one side of the market and then the price on the other side of the market. Conversely, but consistent with the need to define only one market, in two-sided transaction markets the SSNIP test should instead be implemented by raising the price level (that is, the price of the transaction). We recognized there is some debate as to whether the hypothetical monopolist should be allowed to optimally adjust the price structure. Our claim is that the correct formulas should allow the hypothetical monopolist to do so, as discussed in Emch and Thomson, and Filistrucchi. We therefore disagree in this regard with Evans and Noel. In our opinion, while as recognized in the literature traditional formulas would lead to the definition of a relevant market which is too narrow, the formulas Evans and Noel propose would lead to a relevant market that is too large.
Maybe due to the disagreement in the literature, but also to the higher data requirements and the relatively higher complexity of two-sided formulas, while in a few cases competition authorities correctly acknowledged the implications of using traditional single-sided formulas, so far none of the competition authorities appear to have applied a specific two-sided market formula to perform the SSNIP test.
Finally, we discussed when a one-sided approach is harmless and when instead it can potentially lead to a wrong decision. We argued that only in the case of a two-sided non-transaction market, and only when one side does not exert an externality on the other side, can one proceed to define the relevant market on the first side irrespective of the presence of the other side (however, this would not be true for the second side, as the latter must necessarily exert an externality on the first side if the market is two-sided). We also argued that, in most cases, applying a one-sided approach, including a one-sided SSNIP test, would lead to the definition of a market that is too narrow. As a result, one might make a mistake only when the narrow market definition is a decisive element to block a merger, to establish an abuse of dominance or to impose a regulatory burden.