This guest post is by Patrick Todd, an England-qualified solicitor and author on competition law/policy in digital markets.
The above quote is not about Democrat-nominee hopeful Elizabeth Warren’s policy views on sport. It is in fact an analogy to her proposal of splitting Google, Amazon, Facebook and Apple (“GAFA”) apart from their respective ancillary lines of business, a solution to one of the current hot topics in antitrust law, namely the alleged practice of GAFA exploiting the popularity of their platforms to gain competitive advantages in neighboring markets. Can a “referee” favor its own “players” in the digital platform game? Can we blame the “referee” if one “player” knocks out another? Should the “referee” be forced to intervene to protect said “player”? The analogy reflects a growing concern that platform owners’ entry into adjacent markets that are, or theoretically could be, served by independent firms creates an irreconcilable misalignment between the interests of users, independent companies and platform owners. As Margrethe Vestager, European Competition Commissioner and Vice-President of the European Commission (“EC”), has said:
[O]ne of the biggest issues we face is with platform businesses that also compete in other markets, with companies that depend on the platform. That means that the very same business becomes both player and referee, competing with others that rely on the platform, but also setting the rules that govern that competition.
Whether and to what extent successful firms in digital markets can enter and compete in neighboring markets, utilizing their existing expertise, has matured into an existential question that plagues and polarizes the antitrust community. Perhaps the most famous and debated case is the EC’s 2017 decision in Google Shopping, where it concluded that Google’s preferential placement of its comparison shopping results in a special box at the top of its search pages constituted an abuse of a dominant position under Article 102 TFEU. The EC found that such prominent placement, coupled with the denial of access to the box for rival price comparison websites, had the effect of driving traffic to Google’s own shopping site, depriving Google’s rivals of user-traffic. Google is strongly contesting both the facts and theory underpinning this decision in its appeal, the hearing for which took place in February. Meanwhile, complaints in relation to Google’s similar treatment of its other ancillary services, such as vacation rentals, have followed suit. Similar allegations have been made against Apple (see e.g. here), Amazon (see e.g. here) and Facebook (see e.g. here) for the way they design their platforms and organize their search results.[KS1]
What links these cases, investigations and accusations is the doctrine of leverage, i.e. the practice of exploiting one’s market power in one market in order to extend that power to an adjacent market. Importantly, leveraging is not a standalone theory of harm in antitrust law: it is more appropriately regarded as a category of conduct where competitive effects are felt in a neighboring market (think tying, refusals to deal, margin squeeze, etc.). Examples of such conduct in the platform context could include platform owners: promoting their own adjacent products/services in search result pages; bundling, tying or pre-installing their adjacent products/services with platform software code; shutting off access to Application Programming Interfaces or data to third parties to decrease the relative interoperability of their rivals’ products/services; or generally reducing the compatibility of third-party products/services with the platform as a means of distribution.
This post examines various proposals that have been put forward to solve the alleged prevalence of anticompetitive leveraging in digital platform markets, namely:
- blocking platform owners from also owning adjacent products/services;
- prohibiting “favoring” or “self-preferencing” behavior (i.e. enforcing a non-discrimination standard); and
- reversing the burden of proof so that dominant platform firms bear the burden of showing that such conduct does not harm competition.
Each of these proposals would abrogate the “consumer welfare standard” baked into antitrust law, which permits exclusionary behavior as long as it constitutes “competition on the merits”, i.e. that the conduct ultimately benefits consumers. As Judge Frank Easterbrook has mercilessly held, “injuries to rivals are byproducts of vigorous competition, and the antitrust laws are not balm for rivals’ wounds.” Antitrust law maintains a distinction between pro- and anti-competitive leveraging because consumers frequently benefit from the conduct outlined above. Conversely, implementing any of the above proposals would decrease or negate entirely the ability of platform owners to show that such conduct benefits consumers.
This post then examines whether protecting competition in adjacent markets is important enough to sacrifice the consumer benefits that flow from pro-competitive leveraging. Empirical criteria that have been present in comparable instances of such intervention, such as bottleneck power over distribution, evidence of widespread harm to competition in neighboring markets, static product boundaries, and a lack or unimportance of integrative efficiencies, are not satisfied in the current context. Absent some proof that they are, the consumer welfare framework under antitrust law should prevail without recourse to more intrusive intervention.
Proposals to regulate the activities of digital platform owners in neighboring markets
1. Structural separation
Some scholars, such as Lina Khan, propose to implement “[s]tructural remedies and prophylactic bans [to] limit the ability of dominant platforms to enter certain distinct lines of business.” Senator Warren has echoed this proposal, calling for “large tech platforms to be designated as ‘Platform Utilities’ and broken apart from any participant on that platform.” Under this proposal, Amazon would be unable to act both as an online marketplace and a seller on its own marketplace, Google would be unable to act as both a search engine and a mapping provider, and Apple and Google would be unable to act as both producers of mobile operating systems and apps that run on those operating systems. Meanwhile, Facebook would be unable to operate both its core social media platform and separate services, such as dating, local buy-and-sell, and other businesses like Instagram and WhatsApp. Khan posits that such separation is the primary method of “prevent[ing] leveraging and eliminat[ing] a core conflict of interest currently embedded in the business model of dominant platforms.”
A rule that prohibits entry into neighboring markets will certainly catch all instances of harmful leveraging, but it will inevitably also condemn all instances of leveraging that are in fact beneficial to consumers (see below for examples). Moreover, structural separation would also condemn efficiencies stemming from vertical integration that do not depend on leveraging behavior, e.g. elimination of double marginalization. As Bruce Owen sums up, such intervention “is not necessary, and may well reduce welfare by deterring efficient investments,” in circumstances where “[e]mpirical evidence that vertical integration or vertical restraints are harmful is weak, compared to evidence that vertical integration is beneficial.”
2. Non-discrimination principles
Other scholars seek a prohibition on leveraging, i.e. a “non-discrimination” or “platform neutrality” standard whereby platform owners cannot treat third-party products/services differently to how they treat their own. Though framed as a regulatory regime operating in parallel to antitrust law, such regulation would have the effect of supplanting antitrust in favor of a standard that blocks all leveraging behavior, whether pro- or anti-competitive. For example, Apple and Google could still produce both apps and software platforms, but they would be unable to bundle them together, even if doing so improves the user experience (or benefits app developers).
This proposal also disregards the distinction between pro- and anti-competitive leveraging (albeit in a less intrusive manner than structural separation). It would, however, appear to maintain efficiencies stemming solely from vertical integration, as long as said benefits do not result in preferential treatment of the platform owner’s products/services.
3. Reversing the burden of proof
Though not regulatory in nature, it is also worth including the proposal in the EC’s expert report on “competition in the digital era” of recalibrating the legal analysis of leveraging conduct by “err[ing] on the side of disallowing potentially anti-competitive conducts, and impos[ing] on the incumbent the burden of proof for showing the pro-competitiveness of its conduct.” Under this proposal, once a plaintiff establishes that leveraging conduct exists (without having to establish that it satisfies pre-existing legal criteria), the defendant would bear the burden of showing that its conduct did not have long-run anti-competitive effects or that the conduct had an overriding efficiency rationale.
As Dolmans and Pesch point out, proving that conduct does not have a long-term impact on competition may be nigh on impossible, as it involves proving a negative. This proposal would therefore bring non-discrimination in by the back door and return antitrust law to form-based rules that neglect the actual effects of conduct on competition or consumers. Moreover, making it unduly difficult for dominant firms to show that their conduct is in fact pro-competitive, despite any exclusionary effects, would similarly collapse an effects-based model for leveraging conduct into blanket non-discrimination. The report’s authors admit as much, citing for their proposal a report by the French telecoms regulator which advocates “a principle of ‘net neutrality’ for smartphones, tablets and voice assistants” (i.e. a non-discrimination standard).
Consumer vs. small-business welfare: which should prevail in digital markets?
Each of the above proposals would, to varying degrees, dissolve the distinctions between pro- and anti-competitive leveraging and (in the case of structural separation) pro- and anti-competitive vertical integration, significantly curtailing the ability of firms to legitimately out-compete their rivals in neighboring markets. All leveraging would be presumptively harmful to competition and, by extension, consumers.
The question then becomes whether we should ignore the incentive to innovate and compete in platform markets and turn our societal interests to competition within platforms. It has long been the case that “the primary purpose of the antitrust laws is to protect interbrand competition.” However, in certain circumstances, it can be preferable to shift the focus from inter- to intra-brand competition (often through legislation). Take, for example, must-carry provisions imposed on cable operators in the US or net neutrality regulation (repealed in the US, but prevailing in the EU). In circumstances such as these, society willingly foregoes benefits of continued innovation and competition in the inter-brand market because it has concluded, for one reason or another, that bolstering competition in the intra-brand market is more important. This can entail tolerating counterfactually higher prices or reduced quality as a byproduct of protecting interests deemed to be more important, such as maintaining a pluralistic downstream market. In line with the above proposals, there is a growing belief that such an inversion of the goals of competition policy is exactly what is needed in digital markets. This section examines the empirical criteria that one would expect to be verified before shifting the focal point of competition policy from inter-platform competition to intra-platform competition.
1. Strategic bottleneck power over distribution
In past instances of “access regulation” or structural separation of vertically integrated firms, there have been concerns that the targeted firms had strategic “bottleneck” power over the distribution of some downstream product or service, i.e. the firms sat between a set of suppliers and consumers and, through the control of some vital input or method of distribution, controlled access between the two. Strategic bottleneck power was present in the must-carry provisions imposed on cable operators in the US, the non-discrimination principles enshrined in § 616 of the US Communications Act, and net neutrality regulation. The same applies to structural separation: when the District Court approved the consent decree structurally separating AT&T’s long-distance arm from its local operating companies, it was motivated by the fact that “the principal means by which AT&T has maintained monopoly power in telecommunications has been its control of the Operating Companies with their strategic bottleneck position.”
Do GAFA possess strategic bottleneck power? Take Google Search, for example. In the EC’s decision in Google Shopping, it found that referrals from Google Search accounted for a large proportion of traffic to rival comparison shopping websites and that traffic could not be effectively replaced by other sources. However, firms operating in neighboring markets have many more routes to consumers that flout discovery through a search engine. As John Temple Lang observes, they can access consumers through “direct navigation, specialized search services, social networks such as Facebook and Pinterest, partnerships with PC and mobile device markets, agreements with other publishers to refer traffic to each other, and so on.” Apple’s iOS and Google’s Android, on the other hand, compete against each other and thus neither firm, by definition, can possess the degree of strategic bottleneck power required to consider abandoning their respective incentives to innovate. As for Amazon: in 2019 it was estimated that Amazon accounted for 38% of all online sales in the US. This may seem like a staggering volume, but it in fact shows that distributors can – and do – bypass Amazon’s platform to reach consumers, with great success.
Insofar as GAFA possess strategic bottleneck power over particular categories of goods (i.e. in particular neighboring markets), this would not justify shifting the focus to intra-platform competition across all product categories. The market power element of traditional antitrust analyses serves to guard competition in these circumstances by carving a remedy around conduct that illegitimately hampers the ability of competitors in neighboring markets to compete.
2. Widespread harm in adjacent markets
To ban platform owners from leveraging anti- and pro-competitively, one would expect there to be cogent evidence of harm to competition across a multitude of adjacent markets that depend on the platforms for access to consumers. However, as Feng Zhu and Qihong Liu note, there is a dearth of empirical evidence on the effects of platform owners’ entry into complementary markets. Even studies that support the proposition that such entry dampens or skews innovation incentives of firms in adjacent markets conclude that the welfare effects are ambiguous, and that consumers may actually be better off (see e.g. here and here). Other studies show that third-party producers can benefit from platform entry into adjacent markets (see e.g. here and here). It is therefore clear that this criterion, which should also be a prerequisite to imposing blanket regulation to control the behavior of platform owners, has not been satisfied.
3. Discernible and static product boundaries
In prior cases of access regulation, the input that firms in neighboring markets have depended on to access consumers has been clearly distinguishable from their own products. Although antitrust literature commonly refers to “platforms” and “applications” as if these are perceptibly different products, the reality is much different: both platforms and their complementary applications are composed of individual components and, as Carl Shapiro notes, “the boundary between the ‘platform’ and services running on that platform can be fuzzy and can change over time.” Any attempt to freeze the definitional boundary of a platform would negate platform owners’ incentive to build upon and improve their platforms, to the detriment of consumers and (in the case of software platforms) app developers. If Apple were prevented from vertically integrating, what would iOS look like? Could it even have a voice-call function? Alternatively, under non-discrimination regulation, what would a new iOS device look like? Would it just be a blank screen where the user is then forced to choose between various alternatives?
The problem with proposing to separate platforms from adjacent products is that any platform component can theoretically be modularized and opened to competition from third-parties. Because integration of complementary components is an essential part of inter-platform competition, imposing the proposed interventions could destroy the very ecosystems that the competitors that critics seek to protect depend on, and prevent the next popular digital platform from emerging.
4. Lack or unimportance of integrative efficiencies
Critics may counter that efficiencies stemming from leveraging are unimportant or non-existent, or do not depend on conduct that has exclusionary effects, and thus nothing is lost by shifting antitrust’s focus to the protection of competitors in adjacent markets. However, any iPhone user will testify to the consumer benefits flowing from technically integrating multiple platform components and features into a single package (e.g. voice-assistant technology and mapping functionality). In a similar vein, the UK CMA, in approving Google’s acquisition of mapping software company Waze, was prompted in part by the fact that “[i]ntegration of a map application into the operating system creates opportunities for operating system developers to use their own or affiliated services (for example search engines and social networks) to improve the experience of users.” Integrating Product A (the platform) and Product B (a component or the software code of an ancillary product/service) can facilitate the creation some new functionality or feature in the form of Product C that users value and, crucially, could not achieve by combining Products A and B themselves (from one or multiple firms). Another potential consumer benefit flowing from leveraging is a reduction in consumer search costs i.e. providing users with the functionality or end results that they seek more quickly and efficiently. Even though anti-competitive concerns can theoretically arise, it remains the case that, empirically, integration of software code is predominantly motivated by efficiency justifications and occurs in both competitive and concentrated markets.
Much of the impetus to enact the above proposals stems from the perception that antitrust law in its current form does not act quickly enough to restore competition in the market. Indeed, it can take over a decade for the dust to settle in big ticket antitrust cases, by which time antitrust remedies may be too little too late in those cases where the authorities get it right. To the extent that authorities can think of innovative ways to enforce existing standards more quickly and accurately, this would be met with widespread enthusiasm (but may be idealistic).
However, introducing more intrusive measures to protect competition in neighboring markets, and undermining the consumer welfare standard that protects the ability of dominant firms to legitimately enter neighboring markets and compete on the merits, is not warranted. Intervention should remain targeted and evidence-based. If a complainant can adduce evidence that a platform owner is leveraging into a neighboring market and raising the complainant’s cost of doing business, and if the platform owner cannot show a pro-competitive justification for the behavior, antitrust law will intervene to restore competition under existing standards. For this, no regulatory intervention or other change to existing rules is necessary.
For a more detailed version of this post, see: Patrick F. Todd, Digital Platforms and the Leverage Problem, 98 Neb. L. Rev. 486 (2019).
Available at: https://digitalcommons.unl.edu/nlr/vol98/iss2/12.