Derek Holt and Felix Hammeke, AlixPartners LLP
This is an extract from the second edition of the E-Commerce Competition Enforcement Guide - published by Global Competition Review. The whole publication is available here.
What is a digital platform? What is a multisided market?
Platforms operate in two or multisided markets. These are markets ‘in which [a platform] sells different products to different groups of consumers, while recognising that the demand from one group of customers depends on the demand from the other group’. This interdependence of demand between the different sides of the market represents an indirect network externality (INE). This means that platforms need to attract one group of customers to attract the other, and that they need to keep both happy to thrive.
The platform business model is by no means new. Newspapers have long sold the attention of their readers to advertisers, and payment cards, which allow shoppers and merchants to complete transactions, have been in circulation for many decades. INEs play an important role in these platforms. In the case of advertising-supported platforms (attention platforms), INEs are only one sided: advertisers care about the number of readers, but readers do not care about the number of advertisers. In the case of ‘transaction markets’ or ‘matching markets’, INEs are positive for both sides: shoppers prefer cards that are accepted by many merchants and merchants like to accept cards that shoppers want to use.
Digital platforms share these features. The definition proposed by Harold Feld describes digital platforms as companies:
- that operate in two or multisided markets, where at least one side is open to the public (e.g., as content creators or consumers);
- whose services are accessed via the internet; and
- that, as a consequence, enjoy particular types of powerful network effects.
How do business models of digital platforms vary?
Digital platforms employ various operating models. Both Google and Facebook have traditionally been attention platforms. Consumers can use their search or social media services for free (i.e., they ‘pay’ with their attention and data). The platforms monetise their services on the other side of the market through targeted advertising services (i.e., selling users’ attention). In these cases, INEs are positive on the advertiser-side only and any potential transactions between users and advertisers tend to take place outside the platform.
Amazon and Booking.com are examples of transaction platforms. They match shopper or travellers with vendors or hotels, respectively, who both prefer platforms that are more popular on the other side of the market. They further facilitate the transaction on the platform through the development of rules and governance structures.
A separation along the attention versus transaction platform line is not always possible, particularly for large platforms that have evolved into ecosystems. YouTube (owned by Google’s parent, Alphabet) operates in a three-sided market (viewers, content creators and advertisers) with INEs between viewers and content creators that go each way, and single-sided INEs between viewers and advertisers. Facebook’s Marketplace matches people who want to get rid of things with those who need them and live nearby. This transaction model differs from Facebook’s original business model, which was a pure attention platform.
Why are there so many big digital platforms?
As mentioned, platforms are by no means new. Marketplaces in medieval towns already matched shoppers and merchants long before the founders of eBay or Amazon were born, while newspapers and television stations sold eyeballs to advertisers long before the advent of online social media. However, the growth of some digital platforms to become among the largest firms globally may raise new issues concerning the use of data and the potential for market power to be exploited. Some of the key factors that contributed to the swift development of large global digital platforms are the following:
- the internet reduced the cost of communication to almost zero, allowing companies to gain a global reach at limited cost and to maximise network effects;
- the replacement of hardware through software created substantial economies of scale – reducing the marginal cost of serving additional customers – and scope – allowing businesses to roll out complementary services at limited cost; and
- the ability to harvest and process large data sets using more and more sophisticated techniques allowed companies to get better by getting bigger.
These factors generated significant consumer benefits. Consumers can reach a global network of friends or merchants for free and receive recommendations about products that they would never have known about in a world without digital platforms. At the same time, these factors may have the potential to increase barriers to entry, as explored further below.
Business strategies and outcomes in digital platform markets
Digital platforms may adopt different models than other businesses due the factors listed above. For example, the presence of INEs may mean that companies maximise their profits by offering their services for free to one side of the market. Facebook does not charge for its social network services because the additional revenue would be outweighed by the loss in advertising revenues resulting from the loss in usage. Property portals may charge estate agents rather than property seekers if they consider the latter are more likely to switch to competitors if fees increase (i.e., their demand is more elastic). Platforms need to take these indirect effects into account when they decide how to price.
Stronger competition between platforms may not necessarily lead to lower prices on one side of the market. For example, Graeme Guthrie and Julian Wright (2007) show that competition between payment platforms can lead to higher rather than lower charges for merchants. However, the overall price of the service – the combined price paid by merchants and cardholders – can be expected to decrease as a result of competition.
The presence of INEs also has implications for companies’ growth trajectories. Marketplaces need to attract a critical mass of merchants to be of interest for shoppers, but they must also attract a critical mass of shoppers to attract merchants. This is the ‘chicken-and-egg’ problem. Loss-leading strategies may be adopted, and some platforms have achieved high valuations before earning significant (or any) profits.
These strategies can be further rationalised by the presence of economies of scale and scope, and the value of data. Platforms can enter a virtuous circle of growth once they exceed a critical mass. The cost of writing new code can be shared across a larger user base. The customer relationship allows for the roll-out of complementary services. The availability of large data harvested from the user base and ability to process it allows larger platforms to be better than the smaller competition. Rational firms, therefore, have strong incentives to sacrifice short-term profits for long-term gains.
While there are many examples of digital markets in which multiple operators may flourish, the factors noted above can in some circumstances lead to greater concentration than might be observed in other markets. Some of the potential competition concerns identified in the context of digital markets are discussed below.
What competition concerns may arise in digital markets?
According to the authors of the Furman Report, the presence of the features identified above mean that digital platform markets ‘show a tendency to tip towards a single winner’. In the Vestager Report, the experts link the features to high levels of concentration. They consider digital platforms may hold a strong incumbency advantage and argue that this ‘changes the principles of enforcement of competition policy’. In particular, the experts find that rivals may be unable to attract a critical mass and that competition ‘for’ the market could be ineffective. Competition could be further weakened by the conduct of platforms, such as the introduction of most favoured nation (MFN) clauses or other actions that limit the ability of consumers to switch or ‘multi-home’ (i.e., use different platforms for the same purpose).
A word of caution with regard to these findings seems appropriate. First, there is no certainty that markets tip towards a single supplier. Most platform markets count more than one competitor. Limited switching costs and the ability to multi-home at low cost to the consumer means that incumbents may need to continue to innovate and offer good service to maintain their customer base.
Further, platforms in one market may face competition from those in other markets. While Facebook does not compete with YouTube in the social media market, they both compete to attract the attention of consumers. Facebook cannot rely on the existence of network effects alone – a lack of engaging features would translate into a lack of user attention and thus lack of advertising revenue. The incentives to innovate may thus remain strong even for platforms that have reached a large scale.
A commonly expressed view is that the data collected by large platforms may yield market power. However, any assessment of the role of data in yielding market power should recognise the varying forms of data that may be collected, whether individual or aggregated; whether it can be replicated or otherwise collected by rivals; and how much data may be required to profitably enter a market. For example, if the value of additional data decreases when data becomes abundant, this may reduce the extent to which greater access to data for large user bases yields competitive advantage. Geoffrey Manne and Joshua Wright (2011) argue that search engines such as Google, but also Yahoo and Bing, already have more data than they can profitably use to refine their search results. This suggests access to data alone, therefore cannot explain the significant market shares enjoyed by Google.
Further, the authors of the Vestager Report express concerns about the way in which platforms manage competition on the platform. For example, e-commerce or booking platforms may sell preferential search results or ‘monopoly positions’ to vendors, leading to competitive distortions. Platforms could even leverage their market power to promote their own products and extend their digital ecosystems (self-preferencing). The authors find that these concerns may not only apply to dominant platforms with significant market shares. They consider that platforms with ‘intermediation power’ – those that have a set of unique customers that can only be reached through them – may not be sufficiently disciplined by competition in their role as regulators.
However, these concerns should not be generalised but instead considered in light of the facts of each case. The incentives of platforms such as Booking.com to sell monopoly positions may be undermined by potential consumer backlash. Even if only a small proportion of consumers compares prices across booking portals, a perception that choice or quality on one platform is reduced by such policies could lead to reputational damage and a loss of support from consumers on one side of the market. In this sense, digital platforms may be no different than traditional intermediators such as insurance brokers, travel agencies and supermarkets: competition with rivals on other platforms encourages policies that offer good service on the consumer side of the market.
Similarly, concerns about self-preferencing by platforms should not be overstated. Vertical integration, including the provision of complementary services by platform operators, can generate substantial consumer benefits. Further, one needs to be careful regarding the view that intermediation power may lead to competition concerns. It is likely that many platforms have intermediation power insofar as some consumers may tend to use a single platform to book their holidays or restaurants. This does not mean that those platforms truly have market power – consumers may well start to look elsewhere if they are unhappy about the platform’s self-promotion. In other words, the ability and incentive to consider other platforms may be important even in relation to consumers who currently tend to single-home.
Some of the recent reports examining digital platforms have expressed more wide-ranging concerns about the implications of the platforms business model for consumers. The Stigler Report highlights the incentives of platforms to develop addictive content resulting from the necessity to sell eyeballs to advertisers. The ACCC Report looks at wider societal implications such as media diversity. These concerns are too complex to be assessed in this chapter.
Promoting competition for the market – contestability
In response to the concerns about tipping and ineffective competition between platforms, the Vestager Report proposes changes to current competition law and enforcement. The authors advocate for a strengthening of merger control to prevent ‘killer acquisitions’; stricter intervention against MFNs and other practices that may prevent multi-homing; and obligations for data portability and interoperability.
Some of these proposals may have unintended consequences. The General Data Protection Regulation already introduced requirements regarding data portability – allowing users to transfer their data to other platforms and thus potentially lowering switching costs. The extension of these requirements to include data interoperability – giving competitors real-time access to standardised data through application program interfaces – could create more competition in complementary services. However, it could also dilute the incentives of platforms to gather the data in the first place. The value of a platform operator such as Google in offering free search, navigation and email services, may well be reduced if it is less able to monetise these services in the most efficient way.
Our colleagues already raised these concerns with regard to similar data sharing proposals made in the Furman Report. They point out that this approach suffers from ‘static bias’ – namely it attaches too much weight to the number of competitors ‘in the market’ at the expense of the drivers of dynamic competition ‘for the market’. Given the importance of the latter in terms of explaining economic growth over time, the authors argued that this bias may be detrimental for consumers.
Finally, businesses that advocate for greater inter-platform competition should be careful what they wish for. As discussed above, greater inter-platform competition does not necessarily lead to lower prices on all sides of the market. On the contrary, it may further increase prices charged to merchants as platforms need to extract more revenue to compete successfully on the consumer side of the market.
Safeguarding competition in the market
In response to the concerns about the regulatory role of platforms, the Vestager Report sets out various proposals to safeguard fair competition on the platform markets. The authors argue that a ‘dominant platform that sets up a marketplace must ensure a level playing field on this marketplace and must not use its rule-setting power to determine the outcome of the competition’. They draw on Article 101 of the Treaty on the Functioning of the European Union case law involving sports organisations that are active in both the setting of rules and the organisation of events to support their conclusions, and argue that similar rules should apply to dominant platforms under Article 102. The authors also find that a lack of transparency by platforms about their regulation – such as how they rank results – could reinforce competition concerns. They argue that dominant platforms could be required to be transparent about their market design under Article 102. The Platform-to-Business Regulation, which was adopted on 20 June 2019, already provides for such transparency obligations on the business side of the market.
Again, these proposals are not without risks. The way in which platforms regulate their marketplaces is an important – potentially the most important – parameter of competition. Platforms set these rules with the aim of maximising the benefits or their users. As highlighted by David Evans (2012), platforms may be much better placed to govern interactions than a public regulator, because they can monitor behaviour more closely and deal with violations more expeditiously. The ability of platforms to exclude agents that behave badly should thus not be undermined by competition authorities that are eager to protect the rights of small businesses.
While increased transparency may not sound controversial, caution is advised as to how any transparency obligations are implemented in practice. For example, an obligation on platforms to reveal their search algorithm – a key parameter of competition – could be detrimental to innovation. Even if platforms were just required to provide detailed descriptions of their search algorithms, vendors may use this information to game the system. In such a world, the ability of a vendor’s marketing team rather than the relevance and quality of its products could determine search results, to the detriment of consumers.
The Vestager Report makes further recommendations with regard to vertically integrated platforms (i.e., platforms that compete with the companies that use them as intermediators). While it acknowledges that self-preferencing by such platforms is not in general prohibited by Article 102 – unless the platform is an essential facility – it suggests that it could constitute an abuse below this threshold. To facilitate the application of these provisions, the authors argue for a shift in the burden of proof. Dominant vertically integrated platforms, which operate marketplaces and want to engage in self-preferencing, would need to demonstrate that there is no long-term exclusionary effect. The same recommendations are made with regard to platforms that provide privileged data access to their subsidiaries.
These proposals also aim to promote inter-platform competition, rather than just promoting fair competition on platforms. Professor Jean Tirole, in a keynote speech in 2019, stressed that entry typically occurs in neighbouring markets to established platforms, allowing for expansion into the market in question at a later point once the entrant has built a user base. Self-preferencing by established platforms in these neighbouring markets could prevent these entry strategies.
The authors of the Vestager Report are cautious about more radical proposals, such as structural remedies, noting that the cost–benefit trade-off is less clear than for traditional infrastructures such as rail or energy networks. However, some commentators, such as the Open Markets Institute, as well as various US politicians, have called for a break-up of large platforms, including Facebook and Amazon.
Even leaving structural options to one side, the idea that the burden of proof should fall to platforms to demonstrate pro-competitive effects of various business practices should be viewed cautiously lest it lead to a reduction in innovation. For example, preventing platforms from using their own data to develop complementary services could not only limit innovation, it could also muddle incentives to invest in the platform business model in the first place. Companies will not enter markets unless they can be sure that they will be able to monetise their investment. Amazon may not have invested in its marketplace to the same degree if it was not able to promote its own products. Taking account of these dynamic factors relating to innovation incentives are crucial to understand the likely overall consequences of a given business practice compared with the counterfactual.
The idea that vertical integration, or vertical agreements between complementary producers in a supply chain, often have pro-competitive effects is consistent with economic theory. These benefits would also arise in the context of digital platforms. In fact, Andrea Amelio and Bruno Jullien (2012) show that there may be additional benefits from vertical integration in two-sided markets. This is the case where platforms wish to charge negative prices on one side of the market and where this would be welfare enhancing, but where platforms are unable to do so as they cannot pay consumers for the use of their platform. Tying of free products to platform usage allows for de facto negative prices that could enhance consumer welfare. Independently of where the burden of proof may lie in relation to the assessment of exclusionary effects, it would be important to take account of these factors in any competition assessment.
How might regulation of digital platforms operate?
A need for ex ante regulation may be required where the market power of incumbents is immutable and where there is little prospect that markets will self-correct. While an assessment of the case for regulation is beyond the scope of this chapter, on the assumption that some governments appear to be moving in this direction, it is useful to consider what form regulation could take. While one option would be to look to sectors where economic regulation has been applied for decades, such as utility sectors on the basis that fixed costs are a feature in both cases, digital platforms differ in many ways from infrastructure networks. They have limited physical infrastructure that could be clearly defined as being part of a regulated entity, evolve constantly and continue to add new services. It is unclear whether Amazon’s own logistics service would be considered as a separate business or part of the marketplace, or whether Facebook’s Marketplace would be part of its social media business. A sensible delineation of platform and non-platform business parts may therefore be impossible. This makes regulation significantly harder to implement.
In addition to their complexity, platform business models are also diverse. This means that no easy one-size-fits-all solution is available. Any regulatory intervention would need to take these differences into account, which would be difficult, costly and unpredictable for operators.
As discussed, policymakers need to be aware of unintended consequences of regulation. Any regulation of marketplaces could lead to distortions of dynamic competition. If potential entrants knew that they would be subject to regulation and unable to self-preference, they may not try to compete for the development of alternative marketplaces, but content themselves with the provision of complementary services. Who would want to develop Facebook 2.0 in a regulated platform market?
The suggestion that regulation can and almost always has unintended consequences should be uncontroversial. Price regulation of natural monopolies such as utilities did limit their monopoly power. However, it has also led to concerns regarding quality and overinvestment. Digital platforms are no different in that respect – the preceding discussion highlights a range of potential unintended consequences.
This does not mean that regulation is bad per se – in some cases, the unintended consequences may be more than offset by the benefits of regulation. However, any regulatory proposal must critically assess the trade-off between market failure and regulatory failure. This must be based on a careful evidence-based case-by-case analysis. Given the stakes of the game – digital platforms are likely to transform markets for years to come – we believe that a proper analysis is particularly important in this context.
Digital platforms play an increasingly important role in social and economic life. The large scale and rapid growth of many of these platforms has given rise to a wide range of consumer benefits but has also generated concerns on the part of traditional media and retail businesses, businesses using those platforms to reach their consumers and policymakers. Digital markets, while often sharing economic features such as the presence of indirect network effects across multiple markets, operate a variety of business models. It will be crucial to take these features into account when assessing the likely effects of any competition or regulatory interventions, and to reduce the risk of adverse unintended consequences for consumers.
Subscribe here for related content, breaking news and market analysis from Global Competition Review.
Global Competition Review covers the most crucial developments in competition law and enforcement worldwide with breaking news commentary and analysis five days a week, as well as original in-depth reports, interviews and features on antitrust enforcement worldwide.