Digital markets, be it social media, online search or e-commerce, are becoming increasingly global. Although some regional and local players exist as well, it is striking how competition authorities around the globe are being confronted with similar or even identical antitrust questions concerning the same digital platforms – frequently the GAFAM (Google, Apple, Facebook, Amazon, and Microsoft). As one of the first steps in virtually every antitrust investigation, the delineation of a relevant antitrust market is particularly crucial, and this is also true in digital markets. In a recent working paper we discuss the application of market definition to digital markets from a comparative perspective, spanning Brazil, the European Union, and the United States.
II. Market Definition as the Basis of Competition Law: A Short Primer
Market definition is a core concept of competition law in Brazil, the European Union, and the United States. In all three jurisdictions, the main parameters of market definition are strikingly similar and market definition is understood to have both a quantitative and a qualitative side to it. In particular, antitrust authorities in all three jurisdictions have issued soft law guidance on delineating a relevant antitrust market – testament to the ongoing convergence on this important aspect of competition law.
Market definition is a central feature of Brazilian competition law that is explicitly referenced in Article 36 of the Competition Act. That provision prohibits anti-competitive agreements and anti-competitive single firm conduct. As one possible infringement, the provision lists the control of a relevant product market. In addition, Article 36(3) of the Act defines what constitutes a dominant position, which is determined as a 20% share of the relevant market or more. Regular references to the relevant market concept under Brazilian law may partly be due to the fact that the current Competition Act dates from 2011 and is thus relatively young. Brazilian antitrust authority CADE (Conselho Administrativo de Defesa Econômica or Administrative Council for Economic Defense) has issued a number of soft law instruments that are directly relevant to market definition, such as its 1999 Resolution or its Horizontal Merger Guidelines of 2016. These Guidelines acknowledge alternative options for analyzing mergers that do not exclusively focus on market definition – for instance where multi-sided markets are concerned.
Under EU competition law, market definition is relied upon under Article 101 TFEU (anti-competitive agreements), Article 102 TFEU (abuse of a dominant position), and merger control (Regulation 139/2004). In addition, the binding block exemption regulations issued by the European Commission (e.g., Vertical BER, R&D BER) and many soft law instruments (e.g., De Minimis Notice, Effect on Trade Notice) rely on market share thresholds that require the prior delineation of a relevant market. In its Market Definition Notice, the European Commission sets out how it intends to delineate a relevant antitrust market in its product and geographic dimensions. The delineation exercise is based on demand-side and supply-side interchangeability, something that also has the backing of the General Court and the Court of Justice. The Market Definition Notice operationalizes this interchangeability test through the SSNIP test (small but significant non-transitory increase in prices). The Commission is currently reviewing its Market Definition Notice against the background of digital markets, as it strives to better depict these for antitrust purposes.
U.S. antitrust law relies on market definition under § 1 of the Sherman Act (anti-competitive agreements), § 2 of the Sherman Act (monopolization and attempted monopolization) and § 7 of the Clayton Act (merger control). The relevant product market includes all products that are functionally interchangeable with each other, based on demand-side and supply-side substitutability. In their joint Horizontal Merger Guidelines of 2010, the Department of Justice and the Federal Trade Commission set out the hypothetical monopolist test that is nearly identical to the SSNIP test. The Horizontal Merger Guidelines attempt to de-emphasize the importance of market definition for merger control, stating that a merger review need not start with market definition. So far, however, the U.S. courts have not followed this view.
III. Market Definition for Digital Markets: The Decisional Practice of Brazil, the EU, and U.S.
Many digital services are offered for “free,” at least in the sense that users don’t have to hand over any money. Facebook doesn’t ask you for your bank details before you can see what your uncle has had for dinner, Google doesn’t require you to swipe your credit card in order to search for a restaurant for your date tonight, and Twitter doesn’t ask you to do a bank transfer every time you want to share a 280-character brilliant idea. These companies are not charities, but often generate revenue through multi-sided strategies: the users of the social network or the search engine may not be handing over cash, but they are paying attention to advertising or generating information for data-analytical products. The lack of a monetary exchange between users of these services and the undertakings offering them initially had far-reaching consequences for market definition. “Free” was a concept with an almost magical effect: it not only induced consumers to increased appreciation of a product (the so-called “zero-price effect”), but prompted competition authorities to believe that the undertakings could not be subjected to the provisions of competition law. The lack of price for certain (digital) services baffled authorities and courts, who initially showed a reluctance to intervene in the activities of the undertakings who supplied them, struggling to identify the “trade” relationship between the company and the users of its free service. A U.S. district court even went so far as to declare there was no market for search (Kinderstart v. Google). Luckily, this approach did not last. Competition authorities in all three jurisdictions now acknowledge that these companies do engage in economic activities – thus requiring the definition of a market. Such acknowledgement is evident from statements in both the European Commission and CADE’s Google decisions (Google Search (Shopping), Google Android, E-Commerce Media Group Informação e Tecnologia Ltda/Google Brasil).
Acknowledging the possibility of a market was evidently an important step, but there is still an important practical challenge to overcome: how to perform substitution analysis in the absence of monetary price. In all three jurisdictions, there have been attempts to adapt the SSNIP test. A U.S. court raised the possibility of imagining a price by considering whether consumers would switch if even “the most nominal of fees were charged” (Streamcast), but this is inappropriate since going from no price to any price at all would fundamentally change the way the consumer sees the product. Instead, real costs and consideration ought to be integrated in a revised quantitative test, such as the attention or data costs incurred by the users, or even the costs the other side incurs in the case of a two-sided platform (a possibility raised by FTC Staff in the U.S. (In the Matter of Google Inc.). The decisional practice on this remains sparse. Instead, quantitative tests are considered which focus on quality rather than price or costs, which is an option both the European Commission and CADE seem increasingly willing to explore. Quality-based substitution assessments have been mentioned in multiple cases (Facebook/WhatsApp, Microsoft/LinkedIn, Google Search (Shopping), E-Commerce Media Group Informação e Tecnologia Ltda/Google Brasil). The European Commission explicitly referred to a SSNDQ test (small but significant non-transitory decrease in quality) in Google Android, although there is still some lack of clarity on the correct parameters for such tests. The smaller number of investigations in the U.S. to date mean the willingness of U.S. antitrust authorities to adopt revised quantitative tests is not yet clear.
Where reimagined “SSNIP” tests are unavailable, authorities are likely to return to the more traditional qualitative approach. In the Google cases (Google Search (Shopping), E-Commerce Media Group Informação e Tecnologia Ltda/Google Brasil), the European Commission and CADE both considered the characteristics and functionalities of the services in their substitution analysis. The differentiation in the services involved may make it difficult to identify the most pertinent characteristics and functionalities, however, so that there is some contention on where to draw the line between different products and their substitutes. On the whole, it is clear that headway is being made. As more cases are brought in the EU, Brazil, and now the U.S., the variety of quantitative and qualitative tests which now exist will likely be further developed and perfected.
The challenge of zero-price services often arises in the context of multi-sided platforms. A multi-sided platform caters to distinct groups of customers, who are brought together on the platform in a way which internalizes the indirect network effects between them. The number of markets to define when a platform has more than one side is probably the most common question in digital market definition. The existence of two or more distinct customer groups raises the query whether the platform operates on multiple markets (one for each “side”) or on a single platform market, and in any case whether and how to incorporate the relationship between the different groups. Considerable strides have been made in decisional practice, taking steps towards but not quite achieving a definite approach to market definition for digital, multi-sided, platforms.
The literature made a distinction between “transaction” and “non-transaction” platforms, which has inspired the decisional practice on multi-sided platforms both inside and outside digital markets. In transaction platforms, there is an observable transaction, a one-to-one interaction, between different sides of the platform. This is not present on a non-transaction platform, where interaction does not occur as a one-to-one transaction. Most of the decisions involving multi-sided market definition have concerned not digital services but payment cards, which are generally considered to be “transaction” platforms. The American Express judgment in the U.S. has by now become rather famous in antitrust circles, for the relatively simple but contentious answer given to market definition. This Supreme Court judgment came after a string of discussions on market definition in the lower courts. While the District Court had defined multiple but related levels (issuance, acquirer services, and the network level), the Second Circuit had contended that both sides (issuance for cardholders and acquiring for merchants) needed to be “collapsed” into a single market. The Supreme Court agreed with the Second Circuit, emphasizing the existence of a transaction product jointly consumed by both sides. Though this single-market-approach did not attract unanimous consensus, it fitted neatly into the transaction/non-transaction distinction whose simplicity renders it so appealing.
This decisional practice may inspire market definition for digital platforms, but only to the extent that digital services fit neatly in the transaction/non-transaction distinction. Moreover, the distinction may overlook the more nuanced approach which is present in the EU, and in Brazil in particular. The EU and Brazil have had their fair share of payment card cases, long before American Express. These decisions are characterized by more complexity, both in the business models under investigation and in the markets defined. Although the idea of a “joint” transaction product was not dismissed out-of-hand by the European Commission, the Commission was not convinced of its existence on the facts (MasterCard). The EU approach can be summarized as a recognition of the relevance of multi-sidedness to the economic analysis, and a willingness to define both markets for the whole platform as well as multiple one-sided markets where this is useful for the analysis (Cartes bancaires, Budapest Bank). Such a multi-layered approach also exists in Brazil, where CADE’s practice is noteworthy for its definition, in the same case, of both markets for a distinct side and a single market for the whole payment system, focusing its analysis on the relevant market most suited to its assessment (Visa-Visanet, Itaú/Credicard, Elo).
There have been cases specific to digital services in all three jurisdictions, with actual decisions being rendered in the EU and Brazil. The FTC investigated Google, but decided to close the investigation. Although there was no ultimate decision, the Staff and Economist Memoranda did give some insight into the market definition contemplated, namely the definition of three distinct but “interdependent” markets corresponding to the different sides of the Google search platform. A multiple markets-approach was also adopted by the EU in cases involving Facebook, LinkedIn, and Google (Facebook/WhatsApp, Microsoft/LinkedIn, Google Search (Shopping)). The rationale for defining distinct markets for each side was never fully set out, though the Google Search (Shopping) decision did describe the existence of the distinct sides on one platform an advantageous “strategy” by the company, which could be interpreted as an acknowledgement that the multi-sidedness is a commercial choice rather than that there is a joint product for which the participation of both sides is indispensable. In Brazil, CADE’s own Google investigation included reflections by Commissioners and the Department for Economic Studies on the transaction/non-transaction distinction. They seemed to more plainly consider the non-essential nature of the presence of both sides, and thus there was no joint product. Yet even in this case there was no definite statement that several markets had to be defined because Google’s search platform was a non-transaction market, let alone an overarching determination that the transaction/non-transaction distinction is conclusive in multi-sided market definitions. To truly come up with a framework for multi-sided market definition in digital markets, we will likely have to wait for more cases, preferably some involving a digital transaction platform, so that a comparison is possible between the decisional approach to digital non-transaction and digital transaction platforms.
Digital ecosystems, as the overarching building blocks of multi-sided digital platforms, are increasingly becoming a market force to reckon with and are also increasingly being looked at from the perspective of competition law and competition economics. They typically consist of an ecosystem orchestrator that brings together multiple actors and multiple products while maintaining maximum interoperability within the ecosystem. While the experience antitrust enforcers have gathered when taking zero-price services and multi-sided markets into account will be helpful in developing a useful approach to digital ecosystems, these types of commercial structures may also benefit from insights into cluster markets and aftermarkets.
As the European Commission’s Special Advisers’ Report on “Competition policy for the digital era” observed, companies are “draw[ing] consumers into more or less comprehensive ecosystems” which may need to be analyzed separately and/or alongside markets for specific products. Indeed, many digital platforms attract consumers because they offer multiple products in the same place: the convenience of a smartphone, a mobile operating system, a search app, an email service, a browser, a streaming service and a cloud service all in one place. The products that can be found in a digital ecosystem – such as online search, a mobile smart operating system, an app store, a photo storage software, a phone book, a browser, a messaging service, a document storage service, etc. – may often be complementary, but sometimes it is only the underlying technology or business model that links these digital services. Also, while the examples above relate to the software dimension of digital ecosystems, a digital ecosystem usually encompasses both hardware and software.
When defining cluster markets – such as in banking or in supermarkets – antitrust courts and authorities grouped together products that customers would usually expect from the same provider, even though the products themselves were not interchangeable (and not all clusters were made up of exactly the same products). In the U.S., cases included Philadelphia National Bank (commercial banking) or Grinnell (alarm systems); in the EU, they included Lombard Club and Carrefour; in Brazil, they included Banco Santander and Banco Nossa Caixa. Importantly, the cluster’s economic significance goes beyond the individual products or services that it groups together (United States v. Phillipsburg National Bank and Trust) – just like in digital ecosystems. However, digital ecosystems appear to go even beyond what the cluster market concept attempted to capture, through the focus on interoperability within the ecosystem and the capturing of customers that occurs thanks to this.
For digital ecosystems, it can be helpful to think of multiple layers of competition: at the ecosystem level, at the platform level, and on individual market sides. To understand these layers, market definition for aftermarkets can be a useful analogy: The question then centres on whether or not there is competition at the superior market level (for instance, see Eastman Kodak in the U.S. or EFIM in the EU). If one finds that there is no competition at the superior level, then customers at the inferior level may experience capture and lock-in despite the (seeming) possibility of competition – and thus the appearance of a broader market.
While the case law has already started to explore an appropriate methodology for taking zero-price services into account and for delineating multi-sided platforms, this area will require further antitrust attention in the near future – and can benefit from a cross-jurisdictional analysis of evolving issues.
While market definition will remain a central tool for competition law, it can – and must – be adapted to the specific characteristics present in digital markets, as cases before antitrust authorities and courts in all three jurisdictions have already shown. Nevertheless, there still is considerable scope for more convergence on these questions. As these issues frequently are of a global nature, a global response in the shape of a more harmonized approach to market definition in digital markets is desirable. In this endeavor, it is crucial to embark on joint enforcement actions (wherever possible) and collaborative research in order to carve out possibilities for such convergence.
** Professor and Head of the Competition Law and Digitalization Group, Vienna University of Economics and Business; Professor of International Antitrust Law, University of Graz; email@example.com. She is the author of Hart Publishing’s Competition Law’s Innovation Factor: The Relevant Market in Dynamic Contexts in the EU and the US.
In accordance with the ASCOLA declaration of ethics, we declare no conflict of interest. The following is a summary of Magali Eben and Viktoria H.S.E. Robertson, ‘The Relevant Market Concept in Competition Law and Its Application to Digital Markets: A Comparative Analysis of the EU, US, and Brazil’ Graz Law Working Paper No 01-2021, available at https://ssrn.com/abstract=3762447.
Viktoria H.S.E. Robertson