Last updated November 23, 2022

Issue Spotlight: Self-Preferencing

The Issue

A growing chorus of critics have argued in recent years that so-called “Big Tech” platforms harm competition by favoring their own services over that of complementors with whom they compete on the platform. According to this line of argument, complementors are “at the mercy” of tech platforms.

The Debate Over Self-Preferencing

Critics Claim Self-Preferencing by Online Platforms Is Anticompetitive

Critics of platform self-preferencing claim that by discriminating in favor of their own content and against independent “edge providers,” tech platforms cause “the rewards for edge innovation [to be] dampened by runaway appropriation,” leading to “dismal” prospects “for independents in the internet economy—and edge innovation generally.”  Some have even claimed that “self-preferencing threatens the American Dream.” (See also here and here.)

The problem, however, is that the claims of presumptive harm from self-preferencing are based on neither sound economics nor sound evidence.

Against the Vertical Discrimination Presumption

The notion that self-preferencing by platforms is harmful to innovation is entirely speculative. Moreover, it is flatly contrary to a range of studies showing that the opposite is likely true. In reality, platform competition is more complicated than simple theories of vertical discrimination would have it, and there is certainly no basis for a presumption of harm.

Over the past several years, a growing number of critics have argued that big tech platforms harm competition by favoring their own content over that of their complementors. Over time, this “vertical discrimination presumption” has become the go-to argument for big tech’s staunchest critics seeking to level novel charges of anticompetitive conduct against these platforms. Indeed, judging by the grandiose claim made by one critic at a recent Senate hearing—“Digital platform self-preferencing threatens the American Dream”—the argument may be the very apotheosis of “populist antitrust.”

According to this line of argument, complementors are “at the mercy” of tech platforms. By discriminating in favor of their own content and against independent “edge providers,” tech platforms cause “the rewards for edge innovation [to be] dampened by runaway appropriation,” leading to “dismal” prospects “for independents in the internet economy—and edge innovation generally.

The problem, however, is that the claims of presumptive harm from vertical discrimination are based neither on sound economics nor evidence.

The Case For Self-Prefencing

Self-prefencing is a common business practice throughout the economy. Supermarkets and other retailers, for example, regularly sell — and preference — their own products alongside those of third-party businesses. Like offline retailers, platforms often prioritize their own products through, e.g., preferential product placement, by improving their products based on information gleaned from third-party sales, or by conditioning third-party use of their platform on the consumption of their services.

Platforms’ incentives are to maximize the value of their whole ecosystem, which includes both the core platform and the services they attach to it.

Platforms that grant preference to their own products frequently end up increasing the value of the market overall by growing the total share of users of a particular product, and those that give preference to inferior products end up hurting the attractiveness to users of the platform, weakening themselves to competition from rivals.

What Have the Intermediaries Ever Done for Us?

Executive Summary

Intermediaries may not be the consumer welfare hero we want, but more often than not, they are one that we need. Policymakers often assume that intermediaries and centralization serve as a cost to society, and that consumers are better off when provided with “more choice.” Concrete expression of this view can be found in regulatory initiatives that aim to turn “closed” platforms into “open” ones (see, in Europe, the Digital Markets Act; and in the United States, the Open App Markets Act and the American Innovation and Choice Online Act). Against this backdrop, we explain that, as with all economic goods, intermediation involves tradeoffs. Intermediaries emerge when it would otherwise be too difficult (or too costly) for groups of users to meet and interact. There is thus no guarantee that government-mandated disintermediation — such as that contemplated in the European DMA and the U.S. AICOA bill — will generate net benefits in a given case. The ongoing Epic v Apple proceedings are a good example of why it is important to respect the role of intermediaries in digital markets, and the unique benefits intermediation can bring to consumers. The upshot is that intermediaries are far more valuable than they are usually given credit for.

Read the full issue brief here.

Self-Preferencing: Building an Ecosystem

“Self-preferencing” is a normal part of how platforms operate, both to improve the value of their core product and to make money from it so that they have a reason to keep investing in it.

Background…

Digital platforms have been accused of unfairly favoring their own products over those of their competitors, most notably in the EU’s case against Google for displaying Google Shopping results in relevant Search result pages.

But

Platforms’ incentives are to maximise the value of their whole ecosystem, which includes both the core platform and the services they attach to it. Platforms that preference their own products frequently end up increasing the value of the market overall by growing the total share of users of a particular product, and those that preference inferior products end up hurting the attractiveness to users of their ‘core’ product, weakening themselves to competition from rivals.

Read the full tl;dr explainer here.

Vertical Integration: Economies of Scope

Vertical integration allows businesses to innovate more. It may be bad for certain competitors, but it is rarely bad for competition.

Background…

Vertical integration refers to businesses performing a number of different and important roles in the supply chain for a particular product — for example, a manufacturer that sells its products directly to the public in its own stores. While all businesses are vertically integrated to some extent, some worry that vertical integration in digital markets prevents smaller businesses from competing.

But

Empirical research has consistently found vertical integration to be good for consumers through a number of mechanisms that allow for reduced costs and better product quality.

Read the full tl;dr explainer here.

Self-preferencing Is a Product of Vertical Integration

Arguments against self-preferencing by platforms have grown out of a line of thought that vertical integration — that is, when a single firm integrates its operations by creating products or services at multiple levels of the same supply chain — is more harmful than previously understood.

When a company operates in a vertically related market, it inevitably “preferences” its own offerings over those of others elsewhere in the market. Amazon offers a platform for retail sales but also offers its own fulfillment services. Google provides search results that link to external sites but also offers its own direct answers to certain queries. Indeed, any company that does its own accounting internally instead of buying those services on the open market is engaged in vertical integration and is therefore engaged in “self-preferencing.”

The Fatal Economic Flaws of the Contemporary Campaign Against Vertical Integration

This paper proceeds as follows. First, we examine the academic calls for stronger presumptions against vertical mergers based on, among other things, the alleged substitutability of contract for merger as a means of vertical integration, and the alleged equivalence of harms that arise from vertical and horizontal mergers. We analyze these claims on their own terms before proceeding in the next part to survey the economic literature that undermines the foundation of these arguments. We then proceed to analyze the critical differences between horizontal and vertical mergers that makes conflation of these two distinct methods of business combination impossible to truly treat as analytically equivalent. Next, we discuss the mistake of substituting static analysis for a more thorough dynamic analysis, particularly in industries marked by fluid product cycles and flexible business models.

Vertical Integration Has Long Been Understood to Be Generally Procompetitive

For decades economists have virtually uniformly agreed that vertical integration — and the consequent self-preferencing of a firm’s own services — is an efficient (or, at worst, benign) organizational response to complications of doing business across firms at different stages of the supply chain.

Copious Empirical Evidence Supports the Vertical Integration and Self-Preferencing

The theory of vertical-discrimination harm is at odds not only with platform-specific empirical evidence…

…but it is also contrary to the longstanding evidence on the welfare effects of vertical restraints, more broadly.

Consumers Benefit from Platform Self-Preferencing

Consumers don’t want digital platforms to be dumb pipes, or to act like a telephone network or sewer system. The Internet has an abundance of information and options, as well as a host of malicious actors. Good digital platforms act as middlemen, sorting information in useful ways and taking on some of the risk that exists when, inevitably, we end up doing business with untrustworthy actors.

When users have the choice, they tend to prefer platforms that do quite a bit of “discrimination” — that is, favoring some sellers over others, or offering their own related products or services through the platform. Most people prefer Amazon to eBay because eBay is chaotic and riskier to use.

Platform Self-Preferencing: Benefits to Consumers and to Competition

Background…

Four months after a similar antitrust measure was advanced to the floor of the U.S. House, a bipartisan group of senators, led by Sens. Amy Klobuchar (D-Minn.) and Chuck Grassley (R-Iowa), announced that they will introduce legislation designed to drastically reduce the ability of digital platforms to favor their own goods and services. Dubbed the American Innovation and Choice Online Act, the bill presumes that the practice of “self-preferencing” is inherently harmful to competition.

But…

While the draft Senate bill does make certain improvements over its House counterpart, the legislation fundamentally misunderstands the nature of platform competition and the benefits that accrue to consumers from many self-preferencing practices. The bills also would delegate enormous power and discretion to antitrust regulators, who may use that power to achieve fundamentally political ends.

Read the full explainer here.

The complaints, by contrast, come from competitors. Is it fair, for example, for Amazon to use the data it gathers from its service to design new products if third-party merchants can’t access the same data? This seemingly intuitive complaint was the basis for the European Commission’s landmark case against Google.

The Real Reason Foundem Foundered

Summary

A pair of recent, long-form articles in the New York Times Magazine and Wired UK — the latest in a virtual journalistic cottage industry of such articles — chronicle the downfall of British price comparison site and stalwart Google provocateur, Foundem, and attribute its demise to anticompetitive behavior on the part of Google.

Unfortunately, the media’s hagiographies of Foundem and its founders, Shivaun and Adam Raff, approach the antitrust question as if it were imbued with the simple morality of a David vs. Goliath tale. The reality is far more complicated. In fact, these articles misunderstand and misstate the critical economic, business, and legal realities of Google Search, of Foundem’s claims of harm, and of the relationship between the two.

Was Foundem’s failure really the result of anticompetitive “gatekeeping” on Google’s part? Or could it simply be a pedestrian tale of yet another tech start-up that failed because its founders didn’t appreciate that a successful business is built on more than just a good idea?

While the import of the Foundem story has been misconstrued by journalists and EU regulators, it is useful in illuminating what may actually be the fundamental question regarding the antitrust fortunes of the platform economy:

What, if anything, does a successful platform “owe” to the companies that make themselves dependent upon it?

Making Sense of the Google Android Decision

The European Commission’s recent Google Android decision will go down as one of the most important competition proceedings of the past decade. Yet, in-depth reading of the 328-page decision leaves attentive readers with a bitter taste. The problem is simple: while the facts adduced by the Commission are arguably true, the normative implications it draws—and thus the bases for its action—are largely conjecture.

This paper argues that the Commission’s decision is undermined by unsubstantiated claims and non sequiturs, the upshot of which is that the Commission did not establish that Google had a “dominant position” in an accurately defined market, or that it infringed competition and harmed consumers. The paper analyzes the Commission’s reasoning on questions of market definition, barriers to entry, dominance, theories of harm, and the economic evidence adduced to support the decision.

Section I discusses the Commission’s market definition It argues that the Commission produced insufficient evidence to support its conclusion that Google’s products were in a different market than Apple’s alternatives.

Section II looks at the competitive constraints that Google faced. It finds that the Commission wrongly ignored the strong competitive pressure that rivals, particularly Apple, exerted on Google. As a result, it failed to adequately establish that Google was dominant – a precondition for competition liability under article 102 TFEU.

Section III focuses on Google’s purported infringements. It argues that Commission failed to convincingly establish that Google’s behavior prevented its rivals from effectively reaching users of Android smartphones. This is all the more troubling when one acknowledges that Google’s contested behavior essentially sought to transpose features of its rivals’ closed platforms within the more open Android ecosystem.

Section IV reviews the main economic arguments that underpin the Commission’s decision. It finds that the economic models cited by the Commission poorly matched the underlying fact patterns. Moreover, the Commission’s arguments on innovation harms were out of touch with the empirical literature on the topic.

In short, the Commission failed to adequately prove that Google infringed European competition law. Its decision thus sets a bad precedent for future competition intervention in the digital sphere.

But we cannot assume that something is bad for competition just because it is bad for certain competitors. A lot of unambiguously procompetitive behavior, like cutting prices, also tends to make life difficult for competitors. The same is true when a digital platform provides a service that is better than alternatives provided by the site’s third-party sellers.

Platform Self-Preferencing Can Be Good for Consumers and Even Competitors

Critics of big tech companies like Google and Amazon are increasingly focused on the supposed evils of “self-preferencing.” This refers to when digital platforms like Amazon Marketplace or Google Search, which connect competing services with potential customers or users, also offer (and sometimes prioritize) their own in-house products and services.

Read the full piece here.

Self-preferencing can even be good for competing services, including third-party merchants. In many cases, it expands the size of their potential customer base. For example, blockbuster video games released by Sony and Microsoft increase demand for games by other publishers, because they increase the total number of people who buy Playstations and Xboxes. This effect is clear on Amazon’s Marketplace, which has grown enormously for third-party merchants even as Amazon has increased the number of its own store-brand products on the site. By making the Amazon Marketplace more attractive, third-party sellers also benefit.

Claims that Self-Preferencing Causes Harm Are Weak

Considerable analysis has shown that specific claims of harm from self-preferencing by the largest digital platforms — Apple, Google, and Amazon — are misguided. Despite several antitrust actions (or threats of such action) rooted in self-preferencing claims in both the EU and the U.S., there is no actual evidence that such conduct has caused consumer harm.

Correcting Common Misperceptions About the State of Antitrust Law and Enforcement

On Friday, April 17, 2020, ICLE President and Founder, Geoffrey A. Manne, submitted written testimony to the U.S. House of Representatives Committee on the Judiciary, Subcommittee on Antitrust, Commercial, and Administrative Law. Mr. Manne contends that underlying much of the contemporary antitrust debate are two visions of how an economy should work. 

One vision, which tends to favor more intervention and regulation than the status quo, sees the economy and society as being constructed from above by laws and courts. In this view, suspect business behavior must be justified to be permitted, and . . . the optimal composition of markets can be known and can be designed by well-intentioned judges and legislators.

On the other hand, there is the view of individual and company behavior as emerging from each person’s actions within a framework of property rights and the rule of law. This view sees the economy as a messy discovery process, with business behavior often being experimental in nature. This second conception often sees government intervention as risky, because it assumes a level of knowledge about the dynamics of markets that is impossible to obtain.  

In Manne’s view,

Antitrust law and enforcement policy should, above all, continue to adhere to the error-cost framework, which informs antitrust decision-making by considering the relative costs of mistaken intervention compared with mistaken non-intervention. Specific cases should be addressed as they come, with an implicit understanding that, especially in digital markets, precious few generalizable presumptions can be inferred from the previous case. The overall stance should be one of restraint, reflecting the state of our knowledge. We may well be able to identify anticompetitive harm in certain cases, and when we do, we should enforce the current laws. But dramatic new statutes that undo decades of antitrust jurisprudence or reallocate burdens of proof with the stroke of a pen are unjustified.  

Manne goes on to address several of the most important and common misperceptions that seem to be fueling the current drive for new and invigorated antitrust laws. These misperceptions are that: 

  1. We can infer that antitrust enforcement is lax by looking at the number of cases enforcers bring;  
  2. Concentration is rising across the economy, and, as a result of this trend, competition is declining; 
  3. Digital markets must be uncompetitive because of the size of many large digital platforms; 
  4. Vertical integration by dominant digital platforms is presumptively harmful; 
  5. Digital platforms anticompetitively self-preference to the detriment of competition and consumers; 
  6. Dominant tech platforms engage in so-called “killer acquisitions” to stave off potential competitors before they grow too large; and 
  7. Access to user data confers a competitive advantage on incumbents and creates an important barrier to entry. 

 

See his full testimony, here.

Apple:

ICLE Brief for 9th Circuit in Epic Games v Apple

United States Court of Appeals
For the
Ninth Circuit

EPIC GAMES, INC.,
Plaintiff/Counter-Defendant, Appellant/Cross-Appellee,
v.
APPLE, INC.,
Defendant/Counter-Claimant, Appellee/Cross-Appellant

Appeal from a Decision of the United States District Court
for the Northern District of California,
No. 4:20-cv-05640-YGR ? Honorable Yvonne Gonzalez Rogers

BRIEF OF AMICI CURIAE INTERNATIONAL CENTER FOR LAW & ECONOMICS
AND SCHOLARS OF LAW AND ECONOMICS
IN SUPPORT OF APPELLEE/CROSS-APPELLANT

 

INTEREST OF AMICI CURIAE

 

The International Center for Law & Economics (“ICLE”) is a nonprofit, non- partisan global research and policy center aimed at building the intellectual foundations for sensible, economically grounded policy. ICLE promotes the use of law & economics methodologies to inform public policy debates and has longstanding expertise in the evaluation of antitrust law and policy.

Amici also include 26 scholars of antitrust, law, and economics at leading universities and research institutions around the world. Their names, titles, and academic affiliations are listed in Addendum A. All have longstanding expertise in, and copious research on, antitrust law and economics.

Amici have an interest in ensuring that antitrust promotes the public interest by remaining grounded in sensible legal rules informed by sound economic analysis. Amici believe that Epic’s arguments deviate from that standard and promote the private interests of slighted competitors at the expense of the public welfare.

INTRODUCTION

Epic challenges Apple’s prohibition of third-party app stores and in-app payments (“IAP”) systems from operating on its proprietary, iOS platform as a violation of the antitrust laws. But, as the district court concluded, Epic’s real concern is its own business interests in the face of Apple’s business model—in particular, the commission charged for the use of Apple’s IAP system. See Order at 1-ER22, Epic Games, Inc. v. Apple Inc., No. 4:20-CV-05640 (N.D. Cal. Sept. 10, 2021), ECF No. 812 (1-ER3–183). In essence, Epic is trying to recast its objection to Apple’s 30% commission for use of Apple’s optional IAP system as a harm to consumers and competition more broadly.

Epic takes issue with the district court’s proper consideration of Apple’s procompetitive justifications and its finding that those justifications outweigh any anticompetitive effects of Apple’s business model. But Epic’s case fails at step one of the rule of reason analysis. Indeed, Epic did not demonstrate that Apple’s app distribution and IAP practices caused the significant market-wide effects that the Supreme Court in Ohio v. Am. Express Co. (“Amex”) deemed necessary to show anticompetitive harm in cases involving two-sided transaction markets. 138 S. Ct. 2274, 2285–86 (2018). While the district court found that Epic demonstrated some anticompetitive effects, Epic’s arguments below focused only on the effects that Apple’s conduct had on certain app developers and failed to appropriately examine whether consumers were harmed overall. This is fatal. Without further evidence of the effect of Apple’s app distribution and IAP practices on consumers, no conclusions can be reached about the competitive effects of Apple’s conduct.

Nor can an appropriate examination of anticompetitive effects ignore output. It is critical to consider whether the challenged app distribution and IAP practices reduce output of market-wide app transactions. Yet Epic did not seriously challenge that output increased by every measure, and Epic’s Amici ignore output altogether.

Moreover, the district court examined the one-sided anticompetitive harms presented by Epic, but rightly found that Apple’s procompetitive justifications outweigh any purported anticompetitive effects in the market for mobile gaming transactions. The court recognized that the development and maintenance of a closed iOS system and Apple’s control over IAP confers enormous benefits on users and app developers.

Finally, Epic’s reliance on the theoretical existence of less restrictive alternatives (“LRA”) to Apple’s business model is misplaced. Forcing Apple to adopt the “open” platform that Epic champions would reduce interbrand competition, and improperly permit antitrust plaintiffs to commandeer the judiciary to modify routine business conduct any time a plaintiff’s attorney or district court can imagine a less restrictive version of a challenged practice, irrespective of whether the practice promotes consumer welfare. See NCAA v. Alston, 141 S. Ct. 2141, 2161 (2021) (“[C]ourts should not second-guess ‘degrees of reasonable necessity’ so that ‘the lawfulness of conduct turn[s] upon judgments of degrees of e?ciency.’”). Particularly in the context of two-sided platform businesses, such an approach would sacrifice interbrand, systems-level competition for the sake of a superficial increase in competition among a small subset of platform users.

Read the full brief here.

In Apple v Epic, 9th Circuit Should Remember that Antitrust Forbids Enhancing, not Exercising, Market Power

Every voluntary transaction between a buyer and seller involves the creation of surplus—the difference between the subjective value a buyer attaches to the thing and the seller’s cost of producing and selling the item. Price and other contract terms determine how that surplus is split between the buyer and seller.

Read the full piece here.

App Stores as Public Utilities?

App stores are at the forefront of policy debates surrounding digital markets. The gatekeeping position of Apple and Google in the App Store and Google Play Store, respectively, and related concerns about the companies’ rule-setting and dual role, have been the subject of market studies launched by the Australian Competition and Consumer Commission (ACCC), the Netherlands Authority for Consumers & Markets (ACM), the U.K. Competition and Markets Authority (CMA), the Japan Federal Trade Commission (JFTC), and the U.S. House of Representatives.

Read the full piece here.

Apple v Epic: The Value of Closed Systems

Background…

Apple Inc. and Epic Games are currently locked in a high-stakes antitrust dispute. Epic is challenging Apple’s rules that require apps to use Apple Pay for in-app purchases and that ban alternative app stores from iOS devices like iPhones and iPads. 

The suit is part Epic’s broader strategy, dubbed Project Liberty, to pay lower fees to online platforms. If it succeeds, Epic would be able to steer its users toward lower-priced payment processors. This would increase the competitive constraints that Apple faces when it sets platform fees.

But…

Epic’s proposals would allow large developers and rival payment processors to get the benefit of Apple’s investments in iOS and the App Store without paying for them. It could also undermine other online platforms that rely on commissions to earn a positive return on their upfront investments. This could shrink investments in platform creation and upkeep, hurting users and leading to worse platforms overall.

Read the full explainer here.

Google:

The Antitrust Assault on Ad Tech: A Law & Economics Critique

Executive Summary

For years, regulators and competition watchdogs have expressed concern about competition in the digital advertising business. They note that digital advertising appears to be dominated by a few dominant firms, such as Google, Facebook, and—to a lesser extent—Amazon. Some claim that this dominance allows these firms—and Google, in particular—to engage in anticompetitive conduct to extend their market power and to earn supercompetitive profit at the expense of advertisers, publishers, and consumers. This paper investigates the digital advertising market and assesses some of these claims. We conclude based on the information that is publicly available that many of the most significant claims made against Google’s advertising technology (ad tech) business are based on a misunderstanding of U.S. antitrust law, or of the details of the  ad tech market itself.

Digital advertising provides the economic underpinning for much of the Internet. Targeted digital advertising on independent websites is often facilitated by intermediaries that match advertisers and websites automatically, displaying ads to users for whom they are most relevant. This intermediation has advanced enormously over the past three decades. Some now allege, however, that the digital advertising market is monopolized by its largest participant: Google.

In particular, a lawsuit filed in December 2020 by the State of Texas and nine other U.S. states (later joined by five more states in March 2021) alleges anticompetitive conduct related to Google’s online display-advertising business. It has been reported that the U.S. Justice Department (DOJ) may bring a similar lawsuit before the end of 2022.

Moreover, a bipartisan group of U.S. senators introduced the Competition and Transparency in Digital Advertising Act in May 2022. If passed by Congress and signed into law, the bill would require some of the largest Internet firms—such as Google, Facebook, and Amazon—to break up their digital advertising businesses. A summary of the bill claims that Google is the “leading or dominant” firm in “every part” of the ad tech “stack” and that it uses this dominance to “extract monopoly rents” from advertisers and publishers.

Our paper focuses on the open-display digital advertising business. Display ads are designed to be visually engaging, combining text, images, and a hyperlink to a website. These are distinct from search ads, which are text ads displayed along with organic Internet search results. Most of today’s digital display advertising appears on heavily trafficked owned-and-operated sites such as TikTok, Instagram, YouTube, and Amazon, in which the company providing the advertising space is the same company that operates the platforms that place the ads. In contrast, open-display space is supplied by independent publishers—such as The New York Times (nytimes.com), MLB.com, or The Weather Channel (weather.com)—and is usually facilitated by intermediaries.

This paper begins with an overview of digital advertising. The market’s history is one of dynamic innovation, with many new developments arising to solve problems created by previous innovations, address new innovations, and respond to market developments. The market’s structure has changed dramatically as advertisers, publishers, and consumers have responded to new technologies. These changes and innovations all must balance the competing demands of advertisers, publishers, and consumers to maximize the total value of the advertising platform. Thus, any antitrust evaluation of digital advertising must consider whether the market’s structure and the conduct of its participants may be procompetitive responses to prior market changes, as well as the extent to which the overall market structure may mitigate superficially problematic elements within it.

Of particular importance, digital advertising intermediaries that are vertically integrated into some or all components of the digital advertising “stack” of services use the prices charged to each side of the market to optimize overall use of the platform. As a result, pricing in these markets operates differently than pricing in traditional markets: pricing on one side of the platform is often used to subsidize participation on another side of the market, increasing the value to all sides combined. Consequently, pricing (or other terms of exchange) on one side of the market may appear to diverge from the competitive level when viewed for that side alone. While one side of the market may pay superficially higher fees, this cost can be offset by the benefits from increased participation on the other side of the market. In this way, using subsidies to increase participation on another side of the market creates valuable network effects for the side of the market facing the higher fees.

In the second half of the paper, we address some of the specific allegations of anticompetitive structures and conduct made in the Texas Complaint and by critics of the digital advertising industry. We conclude that a flawed premise underlies many of these allegations. It is a version of the “big is bad” argument, in which conduct by dominant incumbent firms that makes competition more difficult for certain competitors is viewed as inherently anticompetitive—even if the conduct confers benefits on users. Under this approach, the largest firms are seen as acting anticompetitively if they do not share their innovations or reveal their business processes to competing firms. As a result, creating new and innovative products, lowering prices, reducing costs through vertical integration, and enhancing interoperability among existing products is miscast as anticompetitive conduct.

In contrast, we note that U.S. antitrust law is intended to foster innovation that creates benefits for consumers, including innovation by incumbents. The law does not proscribe efficiency-enhancing unilateral conduct on the grounds that it might also inconvenience competitors, or that there is some other arrangement that could be “even more” competitive.

Moreover, U.S. antitrust law does not second guess unilateral conduct simply because it may hinder rivals. Any such conduct would first have to be shown to be anticompetitive—that is, to harm consumers or competition, not merely certain competitors. In multisided markets, this means finding not simply that some firms on one side of the market are harmed, but that the combined net effect of challenged conduct across all sides of the market is harmful. The Texas Complaint, however, is built on the alleged harm of reduced revenue to publishers, without considering the corresponding benefit of lower prices to advertisers (and the consumers of advertised products and services).

Based on the information publicly available, we conclude that many of the most significant claims made against Google’s ad tech business are based on a misunderstanding of U.S. antitrust law, or of the details of the ad tech market itself. Although we cannot be sure how the Texas et al v. Google case will develop once its allegations are fleshed out into full arguments, many of its claims and assumptions appear wrongheaded. If the court rules in favor of these, the result will be to condemn procompetitive conduct and potentially to impose costly, inefficient remedies that function as a drag on innovation.

Legislators, too, who may be concerned about Google’s conduct and tempted to impose regulatory requirements on tech companies should bear in minds the risk of the “Nirvana fallacy,” in which real-life conduct is compared against a hypothetical “competition-maximizing” benchmark and anything that falls short is deemed worthy of intervention. That fanciful approach would pervert businesses’ incentives to innovate and compete and would make an unobtainable “perfect” that exists only in the minds of some economists and lawyers the enemy of a “good” that exists in the market.

Introduction

For years, regulators and competition watchdogs have expressed concern about competition in the digital advertising business. They note that digital advertising appears to be dominated by a few exceptionally large firms, such as Google, Facebook, and—to a lesser extent—Amazon. Some claim that this dominance allows these firms—and Google, in particular—to engage in anticompetitive conduct to extend their market power and to earn supercompetitive profits at the expense of advertisers, publishers, and consumers. This paper investigates the digital advertising market and assesses some of these claims. We conclude, based on the information that is publicly available, that many of the most significant claims made against Google’s advertising technology (ad tech) business are based on a misunderstanding of U.S. antitrust law, or of the details of the ad tech market itself.

Digital advertising provides the economic backbone for much of the Internet. By providing websites and apps a means to monetize their products without having to charge user fees, advertising enables access to entertaining and informative content without payment. Targeted advertising allows companies to inform potential customers of new products, giving new entrants a way to compete with popular incumbents, while effective targeting avoids wasting the time of those who aren’t likely to be interested. Advertising can endow products with new characteristics in customers’ minds and make consumers aware of product features they may not have known about.

Advertising on independent websites is often facilitated by intermediaries that match advertisers and websites automatically, targeting ads at the users to whom they are most relevant. This intermediation has advanced enormously over the past three decades. Some now allege, however, that the digital-advertising market is monopolized by its largest participant: Google. These allegations originate from various sources, including policy discussions, lawsuits, draft legislation, and academic reports.

In particular, a lawsuit filed in December 2020 by the State of Texas and nine other U.S. states (later joined by five more states in March 2021) alleges, among other things, anticompetitive conduct related to Google’s online display-advertising business.[1] This action is one of three currently pending lawsuits brought against Google by government antitrust enforcers in the United States; the other two relate to Google’s distribution agreements and search design.[2] It has been reported that the U.S. Department of Justice (DOJ) may bring a similar lawsuit before the end of 2022.[3] Along similar lines, the European Commission opened an investigation into Google’s display-advertising services in June 2021[4] and the German competition authority has published a report regarding its inquiry into non-search advertising.[5] Among other things, the European Commission is investigating whether Google “has made it harder for rival online advertising services to compete in the so-called ad tech stack.”[6]

These ongoing cases follow regulatory reports and hearings examining the market, including a year-long study by the United Kingdom’s Competition and Markets Authority (CMA). The CMA investigation of digital advertising (including search and social-media advertising) has thus far produced recommendations for a code of conduct and “pro-competitive interventions” into the market, as well as a new regulatory body to oversee these measures.[7] The Australian Competition and Consumer Commission is also conducting its own study of the digital advertising market,[8] and both houses of the U.S. Congress have held hearings on the market in recent years.[9] In October 2020, the Democratic majority staff of the U.S. House Judiciary Committee’s Subcommittee on Antitrust, Commercial, and Administrative Law issued a report that recommended, among other things, regulation for the display advertising market.[10]

The digital  advertising industry has also drawn the attention of legislators. In May 2022, a bipartisan group of U.S. senators introduced a bill that would require some of the largest Internet firms to break up their digital advertising businesses.[11] Dubbed the Competition and Transparency in Digital Advertising Act (CTDAA), the measure was introduced by Sen. Mike Lee (R-Utah) and co-sponsored by Sens. Amy Klobuchar (D-Minn.), Ted Cruz (R-Texas), and Richard Blumenthal (D-Conn.). Sen. Lee’s summary of the CTDAA identifies several allegations against the largest firms in the digital advertising business, with an emphasis on Google.[12] The summary claims that Google is the “leading or dominant” firm in “every part” of the “ad tech stack” and that it “exploits” conflicts of interest to “extract monopoly rents” from advertisers and publishers. Because of these monopoly rents, consumers are harmed in the form of higher prices for advertised goods and services and lower quality of online content, according to Lee’s press release.[13]

A 2020 paper published by the Omidyar Network—itself based on an interim CMA report produced during the authority’s then-pending investigation[14]—alleged anticompetitive practices within Google’s display advertising business and laid out a “roadmap” for a prospective antitrust case.[15] Other legal and economic consultants have also voiced concerns about Google’s role in the digital advertising industry.[16] These critiques were published before the Texas Complaint and provide more detail underlying the allegations and arguments described in the Texas Complaint. For that reason and because it may influence further lawsuits and regulatory interventions in the digital advertising market, including the DOJ’s, we also evaluate several of the Roadmap’s findings and conclusions.

This paper investigates the digital advertising market and assesses the aforementioned claims that it is uncompetitive. It explains some of the complex dynamics that underpin this market, thereby shedding light on the weaknesses and deficiencies in many of the arguments made about it, particularly those behind the Texas Complaint. This analysis is relevant to the entire Internet and to the wider economy, not just to Google and the display-advertising market. Many of the allegations made in the Texas Complaint would, if upheld by a court, have profound implications for antitrust law by establishing precedents that successful platforms effectively have a legal duty to act as essential facilities for their competitors; that efficiency-enhancing innovation by incumbent platforms is anticompetitive (particularly when it is not shared with competitors); and that courts or regulators can impose remedies that put these duties into effect without consideration of the harmful tradeoffs and unforeseen consequences that could themselves constrain competition and innovation. Such an approach would severely affect not only Google and the ad tech industry, but also businesses operating in unrelated markets and industries.

We begin with an overview and history of digital advertising. It is a history of dynamic innovation, with many new developments arising to solve problems created by previous innovations, address new innovations, and respond to market developments. These innovations must balance the competing demands of advertisers and publishers. The market’s structure has changed dramatically as advertisers, publishers, and consumers have responded to recent technologies. Because of this dynamism, we argue that it is a mistake to conclude that market structure and firm conduct at some specific point in time was ideal or better from a competition point of view, or that deviations from that paradigm represent a problem for competition enforcers to correct.

In the second half of the paper, we address some of the specific allegations of anticompetitive structures and conduct made in the Texas Complaint and by critics of the digital advertising industry. We conclude that a flawed premise underlies many of these allegations. Fundamentally, the allegations stem from an assertion that conduct engaged in by dominant, incumbent firms that makes competition more difficult for competitors is anticompetitive—even if the conduct confers benefits on users. This often amounts to a claim that the largest firms are acting anticompetitively by innovating and developing their business processes and products in ways that create benefits for one or more digital advertising constituents and for the advertising ecosystem more generally, including by creating new and innovative products, lowering prices, reducing costs through vertical integration, and enhancing interoperability between existing products.

I. Overview of the Digital Advertising Industry

For many people, digital ads are “just there.” They appear on one’s Facebook timeline, get inserted into one’s Twitter feed, or show up in the middle of an online news article. Unseen to most users is a complex stack of services that match advertisers with advertising space, using real-time auctions and other algorithms to deliver ads targeted to produce a user response, such as buying a product, supporting a cause, or visiting another website.

In this section we explain that digital advertising is just one small part of a much broader advertising and marketing industry. We provide a concise history of how the business evolved from simple banner ads to highly targeted display ads. Through this evolution, digital advertising has become a multisided market where intermediaries must balance the demands of advertisers, publishers, and users to maximize the total value of the advertising platform. Because of this balancing, it is a mistake for policymakers and regulators to focus only on a single set of users or a single segment of the stack of digital advertising services.

A.   Digital advertising is part of a broader advertising and marketing market

The Texas Complaint alleges that Google has market power in six distinct product markets, each of which the states claim to be nationwide in geographic scope. Four of these distinct product markets involve intermediation in the sale of “open” display ads on third-party websites and two involve intermediation in the sale of “in-app” display ads on mobile devices.[17] As we note in our earlier paper, it is likely that the states allege overly narrow product-market definitions.[18] In particular, structure and conduct viewed within a broader digital advertising market, overall advertising market, or marketing market indicates than no single firm has significant market power.

Digital advertising comprises about half of U.S. advertising revenues (Figure 1), while advertising itself accounts for about half of marketing activities. Marketing includes advertising, as well as events, sales promotion, direct marketing, telemarketing, product placement, and merchandising. Within digital advertising, advertisers have a broad set of options about where and how to run ads, including:

  • Search ads, in which the ad is displayed as a search-engine result (e.g., Google, Bing, DuckDuckGo);
  • Display ads on a site owned and operated by the firm that sells the ad space (e.g., Facebook, YouTube, Amazon Marketplace);
  • “Open” display ads on a third party’s site (e.g., The New York Times, Dallas Morning News, Runner’s World); or
  • “In-app” display ads served on mobile apps.

While total advertising spending in the United States has increased by about 15% since 2009, as a share of the economy spending has been relatively flat at slightly less than 1% of GDP.[19] As mentioned, about half of total U.S. advertising dollars go to digital channels, up from approximately 10% a decade ago. Approximately 30% of ad spending goes to TV and less than one-quarter goes to radio, newspapers, magazines, billboards, and other “offline” forms of media.[20]

Figure 1: U.S. Advertising Spending Over Time

Source: Benedict Evans, News by the Ton

It is well-understood that television broadcasters and cable networks compete with digital services.[21] And they do so on virtually all dimensions: for user attention, for labor, for content and other inputs—and for advertising. The same is true of competition for advertising among digital publications, newspapers, radio, magazines, video games, music streaming, and podcasts. The fact that offline and online advertising—to say nothing of marketing more broadly—employ distinct processes does not consign them to separate markets. Indeed, it is widely accepted that online advertising has drawn advertisers from offline markets, as previous technological innovations drew advertisers from other channels before them.[22] Moreover, while there is evidence that, in some cases, offline and online advertising may be complements as well as substitutes,[23] the distinction between these cases is becoming less and less meaningful as the revolution in measuring the effectiveness of advertising has changed how marketers approach different levels of what is known as the marketing “funnel.”[24] For example, economist David Evans’ review of the literature concludes that digital advertising is a segment of the broader advertising business in which different forms of advertising compete and complement each other:

Advertisers base decisions about the level and allocation of their budgets on formal or informal analyses of the rate of return on investment. For these ad campaigns, the different advertising methods can be substitutes to the extent they provide alternative ways of delivering messages to an audience, and complements to the extent they can reinforce each other.[25]

Economists Avi Goldfarb and Catherine Tucker demonstrate that display-advertising pricing is sensitive to the availability of offline alternatives.[26] Although advertising technology and both supplier and consumer preferences continue to evolve, the weight of evidence suggests a far more unified and integrated economically relevant market between offline and online advertising than their common semantic separation would suggest:

We believe our studies refute the hypothesis that online and offline advertising markets operate independently and suggest a default position of substitution. Online and offline advertising markets appear to be closely related. That said, it is important not to draw any firm conclusions based on historical behavior.[27]

In summary, there is evidence that open-display and in-app ads compete with search ads, while digital ads compete with offline advertising. Thus, courts and regulators should be skeptical of overly narrow market definitions focused on only small slices of a much larger relevant market for advertising.

B.   A simplified description of digital display advertising

The combination of software and processes that facilitate digital advertising transactions is known as the “ad tech stack.” The stack consists of several software components to match advertisers with publishers.

Advertiser ad servers are used by advertisers and media agencies to store ads, deliver them to publishers, track their activity, and assess their effectiveness (by, for example, tracking conversions). Demand-side platforms (DSPs) automate the purchase of advertising inventory by collecting bids in real-time auctions from multiple advertiser ad servers. DSP bids are based on the advertiser’s objectives, data about the final user, and data on impressions or conversions. Publisher ad servers manage publishers’ inventory and determine whether and where to serve a particular ad on a publisher’s site. Supply-side platforms (SSPs) automate the sale of publishers’ inventory, typically through real-time auctions involving multiple DSPs.

In general, the process of buying and selling digital ads through open-display auctions works as follows (Figure 2):

  1. When a user opens a webpage (or uses an app), the publisher’s ad server sends a bid request to SSPs for the advertising impressions available on that page for that user.
  2. The SSPs send bid requests to multiple DSPs for the ad impressions.
  3. DSPs evaluate the advertising opportunity based on user data and the objectives of their advertisers’ campaigns and send bids to the SSPs.
  4. SSPs rank the bids received based on price and other priorities set by the publisher. The SSPs send their winning bids to the publisher.
  5. The publisher ad server compares bids received from SSPs, along with any pre-existing direct deals between the publisher and specific advertisers and decides which ad to serve on the page.

Figure 2: A Simplified View of the Digital Advertising Stack

Source: OECD, CMA

While this process applies to most programmatic transactions, there are many variations. For example, there are diverse ways in which SSPs are contacted and asked to submit their bids. To the extent that a publisher and advertiser have a pre-existing, direct agreement, there are differences in how these arrangements are handled and integrated with deals arranged through automated platforms. The specific approach used to match ads with ad inventory will reflect a balance among different sides of a multisided market. One approach might increase the prices received by sellers (publishers) but expose buyers (advertisers) to increased risk of overpayment. Other methods might reduce risks to advertisers, but also reduce the prices received by publishers.

C.   A brief history of digital advertising

This history of the digital advertising market is a history of iterative innovation, with new developments and services arising to solve problems created by previous innovations and to respond to changing market conditions. At the heart of these innovations is an attempt to balance the competing demands of advertisers, publishers, and consumers. Given that this is a dynamic market, it would be mistaken to conclude that the market structure at some specific point was the “correct” one from a competition point of view. Moreover, it would be a mistake to conclude that deviations from some previous “ideal” world present a problem that can be corrected by disruptive regulation.

Digital advertising originally worked similarly to conventional print and broadcast advertising. Online publishers would negotiate with advertisers (or their ad agencies) to sell ad space on their websites, giving relevant advertisers information about their readership gathered through market research. All users would see the same ads. The resulting system was poorly targeted, inefficient, and carried high fixed costs, including the cost of things like market research and the transaction costs of publishers hiring sales teams and advertisers hiring ad agencies to do business with one another. Inevitably, these fixed costs meant that only larger publishers and larger advertisers could engage in the online market profitably.

In 1993, O’Reilly & Associates Inc. introduced its Global Network Navigator (GNN) magazine and other ad-supported online publications, which first rolled out clickable ads. O’Reilly is credited with the first attempt to create an “advertising medium” on the Internet.[28] The price of ads ranged from $500 for a one-page business profile to $5,000 for up to 25 pages about the company placing the ad.[29] A year later, Wired magazine’s digital affiliate HotWired ran what later became known as the web’s first banner ads. The ad—for AT&T’s “You Will” campaign—cost $30,000 for a three-month dedicated placement in a section of HotWired’s site.[30]

The first step toward automating this process came with the introduction of ad-server software on both the publisher and advertiser sides, which allowed each side to automate parts of the ad-placement process. Ad servers allowed publishers to automatically describe the type of content on their pages, which in turn allowed advertisers to place ads tailored to that content. An article about hiking could automatically indicate to a department store to place ads selling walking boots, for instance. It also allowed the publisher to sell to many advertisers without having to transact directly with any of them. Ad servers also allowed advertisers to browse and manage campaigns across a large, aggregated number of publishing sites, instead of having to interact with sales teams one by one. This process was, however, still negotiated directly, and often left publishers with unused “remnant” advertising space that they had not been able to sell.

To solve this problem, ad networks entered the market. These functioned as intermediaries between advertisers’ and publishers’ ad servers, aggregating unsold remnant ad space and allowing advertisers to buy that space en masse without having to deal directly with each publisher. Ad networks did not replace direct sales, but they allowed for residual space to be bought and sold more easily, increasing the amount of ad inventory available and lowering the fixed costs to use it. This, in turn, made the market feasible for smaller publishers, who would otherwise be unable to attract direct deals to sell ad space, and for advertisers to conduct large-scale ad campaigns across many publishers (including small ones).

In 1995, GNN was acquired by AOL.[31] That same year, marketing-communications agency Poppe Tyson spun off its Internet advertising division as DoubleClick, with the objective of “responding to advertisers’ need to be able to buy millions of impressions on the Internet without having to buy from hundreds of different sites.”[32] The company created “subnets” of publishers to target specific categories of consumers.[33]

Also in 1995, ad agency WebConnect, the first ad network, began to collaborate with its clients to identify the websites that their ideal consumers visit. WebConnect then placed ads on the websites where they were more likely to be seen by the audience most relevant to their clients. The company also produced a tool to prevent “ad fatigue,” which occurs when users are repeatedly shown the same ad.[34]

In 1998, GoTo.com, which was renamed Overture in 2001, launched its ad-supported search engine.[35] Search result rankings were based on an open-market bidding process. Advertisers on GoTo were informed of the amounts other advertisers were bidding for a click-through within the results for a given search term, and any advertiser could increase its bid to obtain a higher ranking, a process GoTo described as “pay for performance.”[36] One of GoTo’s key innovations was linking advertising pricing to click-throughs, rather than to page views.[37]

To drive home just how efficient these ads could be, [GoTo.com’s founder Bill Gross] came up with an audacious pricing scheme. Instead of paying for page-views—an old-media model that had come to dominate the Web—advertisers would pay only when people actually clicked on their ads. And their placement on the GoTo.com results page would be determined through an auction, so that more desirable keywords would command higher prices, while less common keywords could be had for as little as a penny per click. As a search engine, GoTo.com had nothing on Google. But as a way of making money on searches, it was ingenious.[38]

During the dot-com boom of the late 1990s, banner ads spread throughout the Internet, though growth was tempered by user complaints that the ads slowed page loading.[39] The dotcom bust wiped out many of the firms that were the biggest buyers of digital banner ads. In response, Wired predicted that digital advertising would undergo a “facelift.”[40]

Old media revenues will wither as mainstream advertisers storm the Net. Instead of stuffing junk mail into the mailbox outside your house, they’ll send it directly to your inbox. And companies get smarter, choosing sites that take better aim at their quarry, er, potential customers.

“It’s very much a targeted medium,” Robin Neilfield, co-founder of NetPlus Marketing. “You have to analyze the content on a site, you can’t just buy based on demographics.”[41]

Because ad networks were not comprehensive—they did not carry the entire inventory of the Internet—publishers began to use yield managers (later known as SSPs) to compare bids for their ad space and to decide which to accept. SSPs effectively allowed publishers to aggregate demand from a larger number of ad networks, which themselves aggregated demand from advertisers. This process allowed ad space to be more easily commoditized, with an SSP helping to identify an ad space’s relevance to potential advertisers.

As indirect sales became possible, ad exchanges emerged to sell ad space using real-time auctions. Ad space could be tagged according to characteristics like audience type, relevance to the advertiser, and/or prominence and quality of the ad, with bids gauged accordingly.

Finally, DSPs on the advertiser side allowed advertisers to engage with many ad networks and ad exchanges at one time. These also allowed advertisers to track campaigns and measure performance of different ads with different publishers, and to adjust their campaigns accordingly. Most ad exchanges now have DSP functionality built in.

In 2000, Google introduced a new self-service advertising product called AdWords (now Google Ads) that allowed businesses to purchase text ads on search-results pages. At the time it was reported that AdWords was designed to attract small-to-midsize advertisers with budgets of $5,000 or less.[42] AdWords differed from GoTo/Overture in a major way. GoTo/Overture placed ads within search results, with results ranked by bid. In contrast, AdWords placed ads separate from search results with pricing based on pageviews.[43] In this way, Google could display ads without compromising the relevance of search results. In 2002, it launched AdWords Select, its pay-per-click, auction-based search-advertising product.[44]

In 2003, Google acquired Applied Semantics, whose AdSense display advertising product allowed it to sell targeted ads on third-party websites.[45] With AdSense, the display-ad server was able to read text on a publisher’s site and serve relevant ads, considering factors like the user’s geographic location, age, demographics, and the search made.[46] AdSense was the forerunner of programmatic display advertising, the process of automating the buying and selling of ad inventory in real time through an automated bidding system. In 2005, Google introduced the Quality Score model, which considers an ad’s click-through rate, as well as the bid price, in placing ads.[47]

YouTube was launched in 2005 and acquired by Google the following year, when the company also introduced video ads. In 2007, Google acquired DoubleClick.[48]

Around 2015, “header bidding” began to roll out, with publishers Meredith Corp. and Townhall Media as two of largest early adopters.[49] Before header bidding, it was difficult for every demand-side partner to submit a bid for every ad request. As a result, publishers relied on approaches such as “ad waterfalls”[50] to try to get the most from each partner. Because of the way ad waterfalls are configured (based on historical, not real-time, data), publishers believed ad waterfalling led to winning bids that were below what some bidders might be willing to pay.[51] Client-side[52] header bidding was adopted as a way to increase real-time price competition among multiple SSPs, leading to higher returns for publishers and a more efficient allocation of ad space to advertisers.

Despite the widespread adoption of header bidding—as of the second quarter of 2021, about two-thirds of publishers were using it[53]—the technology has its own challenges. For example, the addition of extra code on the webpage, which client-side header bidding requires, can slow down the publisher’s website, driving away users.[54]

As an alternative to client-side header bidding, server-side header bidding was introduced. Prebid launched in 2015 as an independent and open-source option. Google released Open Bidding in April 2016 and Amazon introduced Transparent Ad Marketplace (“TAM”) at the end of 2016. In these alternatives, the auction among SSPs takes place in a remote server controlled by a third party (the provider of the server-side header-bidding solution) instead of in the user’s browser. This helps to improve site-load speed. On the other hand, this solution leads to less revenue for publishers and reduces the availability of data to advertisers and publishers.[55]

Over the past decade, the price of digital advertising has fallen steadily, while output has risen. U.S. digital-ad spending grew from $26 billion in 2010 to $189 billion in 2021, an average annual increase of 20%.[56] Over the same period, the Producer Price Index for Internet-advertising sales declined by an annual average of 4%.[57] The rise in spending in the face of falling prices indicates that the number of ads bought and sold increased by approximately 25% annually. The combination of increasing quantity, decreasing cost, and increasing total revenues are consistent with a growing and increasingly competitive market, rather than one of rising concentration and reduced competition.

D.   Digital advertising is a multisided market

The digital advertising market can be thought of as a complex multi-step and multisided market that involves three key parties—advertisers, publishers, and intermediaries—and is aimed at a fourth: consumers. In contrast, critics of the current structure of and conduct in the digital advertising industry have characterized it as a “straightforward and traditional” market in which publishers supply an inventory of ad space and advertisers are buyers of the ad space.[58] In this simplistic account of the market, for a given supply of inventory, publishers would seek to maximize the price received per ad, while advertisers would seek to minimize the price paid per ad. Targeting of ads would be based on the demographics of a publisher’s readership or the content of the publication, rather than the individual characteristics of each reader. In general, this is how the market initially operated before the introduction of clickable ads. But even this simple formulation is quite complex. Advertisers expect to maximize the return on their investment in advertising. Even at a low price, advertising expenditures would be wasted if that investment were not converted to increased sales of the advertiser’s product or service.

The invention of clickable ads with which users could interact changed the objective function of digital advertising. Publisher revenues and advertising costs became linked to individual consumers acting on an ad by, for example, clicking on it. Rather than paying or receiving a price-per-ad based on the size of a publication’s user base, advertising expenditures became a function of a price-per-click (or other action) and the number of clicks. This meant that the rewards for relevance—as well as the complexities of determining relevance—were greater because some viewers might be persuaded to act there and then.

In this multisided market, ad intermediaries must balance the interests of at least three constituencies: (1) advertisers creating ads and placing them; (2) publishers defining inventory and displaying ads; and (3) users consuming published content who view and act on ads. Intermediaries in these markets often benefit from network effects, through which the value of the platform to each user depends in part on the number and quality of other users on the platform.[59]

The quality and relevance of users is assessed by collecting information on the users as they browse the web. This information can include which ads they have viewed and clicked in the past, their geographical location, as well as their demographics, financial situation, and topics of interest. Broadly speaking, a larger network with diverse users provides more information and is better able to target ads to relevant users, benefiting advertisers, publishers, and consumers.

Network effects are not always positive, however, nor are they always captured by the platform that facilitates them.[60] While access to consumer data can help to improve the quality of the ads displayed—and increase the value of those ads to advertisers and publishers—claims that such access provides increasing returns to scale are not borne out by the burgeoning empirical literature on the topic. Summarizing these empirical findings, economist Catherine Tucker concludes that “empirically there is little evidence of economies of scale and scope in digital data in the instances where one would expect to find them.”[61]

Intermediaries in multisided markets often face difficult optimization problems caused by the interrelated demands of participants on different sides of the market, each group of whom benefits from the existence and size of the other, but whose interests conflict across many margins.[62] This highlights the key distinction between “straightforward and traditional” markets and multisided markets.

Ad tech intermediaries that are vertically integrated into some or all components of the ad tech stack use prices charged to each side of the market to optimize overall use of the platform. As a result, pricing in these markets operates differently than pricing in traditional markets. Pricing on one side of the platform is often used to subsidize participation on another side of the market, increasing the value to all sides combined. Consequently, pricing (or other terms of exchange) may appear to one side of the market to diverge from the competitive level when viewed for that side alone. While one side of the market may pay higher fees, this cost can be offset by the benefits from increased participation on the other side of the market. Thus, using subsidies to increase participation on another side of the market creates valuable network benefits for the side of the market facing the higher fees.

For example, among the criticisms of digital advertising business practices is the use of “second-price auctions” rather than “first-price auctions.”[63] First-price auctions are those most familiar to people: multiple bidders offer prices, and the highest bidder wins the auction and pays an amount equal to her winning bid. In a second-price auction, the highest bidder wins the auction but pays an amount based on the next-highest bid. In markets with many bidders possessing the same information, first-price auctions and second-price auctions would be expected to produce the same amount of revenue under the well-known auction-theory concept of revenue equivalence.[64]

The choice of auction approach reflects the tensions between different sides of the market in a multisided market. On the one hand, under certain circumstances, a first-price auction tends to increase the prices received by sellers (here, publishers), but exposes buyers (here, advertisers) to an increased risk of overpayment.[65] On the other hand, under certain conditions, a second-price auction reduces risks to advertisers, but also reduces the prices received by publishers.[66] It would be expected that an ad tech intermediary would balance these competing interests to maximize total revenues flowing through the ad tech stack, to maximize its profitability. In such a multisided market, it would be a mistake to focus only on one side of the market and ignore the effects that decisions such as this have on the other participants.

The extent to which ad tech intermediaries—in particular, vertically integrated services like Google’s—act to optimize the overall value of the platform is critical to understanding how these markets work. It also highlights how misleading it can be to assume that these processes can be analyzed as “straightforward and traditional” markets.

II. Antitrust Primer: Effective Competition Is not an Antitrust Offense

A flawed premise underlies much of the Texas Complaint, the Omidyar Network’s Roadmap report, and the CTDAA legislation. Fundamentally, most of the charges that each of these level against Google and Facebook’s ad tech businesses derive from an assertion that conduct engaged in by dominant incumbent firms that makes competition more difficult for competitors is anticompetitive—even if the conduct confers benefits on users. This often amounts to a claim that the largest firms are acting anticompetitively by innovating and developing their business processes and products in ways that create benefits for one or more digital advertising constituents and for the advertising ecosystem more generally. These include creating new and innovative products, lowering prices, reducing costs through vertical integration, and enhancing interoperability between existing products, among other things.

This approach entails an argument—made explicit in the Texas Complaint and the Omidyar Roadmap—that Google harms competition by creating obstacles for rivals without offsetting “incremental efficiencies.”67F[67] According to the report, this means that, even if Google’s practices produce benefits for such constituents as advertisers, publishers, or consumers, they could possibly be reimagined to create even more competition or achieve the same benefits in ways that better prop up rivals. According to the Roadmap, the practices should therefore be condemned as anticompetitive.68F[68]

But that is not how the law—or the economics—works. Such an approach converts manifestly beneficial aspects of Google’s ad tech business into anticompetitive defects, essentially arguing that successful competition creates barriers to entry that merit correction through antitrust enforcement. The CTDAA takes this argument a step further by imposing “best interests,” “best execution,” and “transparency” obligations on large firms and mandating divesture of parts of the largest firms to facilitate more entry by competitors. This approach turns U.S. antitrust law (and basic economics) on its head. As some of the most famous words of U.S. antitrust jurisprudence have it:

A market may, for example, be so limited that it is impossible to produce at all and meet the cost of production except by a plant large enough to supply the whole demand. Or there may be changes in taste or in cost which drive out all but one purveyor. A single producer may be the survivor out of a group of active competitors, merely by virtue of his superior skill, foresight and industry. In such cases a strong argument can be made that, although, the result may expose the public to the evils of monopoly, the Act does not mean to condemn the resultant of those very forces which it is its prime object to foster: finis opus coronat. The successful competitor, having been urged to compete, must not be turned upon when he wins.69F[69]

U.S. antitrust law is intended to foster innovation that creates benefits for consumers, including innovation by incumbents. The law does not proscribe efficiency-enhancing unilateral conduct on the grounds that it might also inconvenience competitors, or that there is some other arrangement that could be “even more” competitive. Under U.S. antitrust law, firms are “under no duty to help [competitors] survive or expand.”70F[70]

To be sure, the arguments are couched in terms of anticompetitive effect rather than being described merely as commercial disagreements over the distribution of profits. But these effects are simply inferred, based on assumptions that Google and Facebook’s vertically integrated business models entail an inherent ability and incentive to harm rivals. For example, Google is alleged to be able to surreptitiously derive benefits from display advertisers by “leveraging” its search-advertising capabilities,71F[71] or by “withholding YouTube inventory,”72F[72] rather than altruistically opening it up to rival ad networks, or by using its access to data to improve its products without sharing that data with competitors.

All these charges may be true, but none is inherently anticompetitive. Under U.S. law, companies are not obligated to deal with rivals and certainly are not obligated to do so on rivals’ preferred terms.73F[73] In the Texas Complaint, for example, the court, citing Charych v. Siriusware, noted, “[D]efendants were under no obligation to develop an interface that was compatible with plaintiffs’ product.”74F[74] As long ago as 1919, the U.S. Supreme Court held that “[i]n the absence of any purpose to create or maintain a monopoly, the [Sherman Act] does not restrict the long recognized right of [a] trader or manufacturer engaged in an entirely private business, freely to exercise his own independent discretion as to parties with whom he will deal.”75F[75] More recently (in 2004) the Court held:

Firms may acquire monopoly power by establishing an infrastructure that renders them uniquely suited to serve their customers. Compelling such firms to share the source of their advantage is in some tension with the underlying purpose of antitrust law, since it may lessen the incentive for the monopolist, the rival, or both to invest in those economically beneficial facilities.76F[76]

Moreover, U.S. antitrust law does not second guess unilateral conduct simply because it may hinder rivals. Any such conduct would first have to be shown to be anticompetitive—that is, to harm consumers or competition, not merely certain competitors.77F[77] In two-sided markets, this means finding not simply that some firms on one side of the market were harmed, but rather that the combined net effect of challenged conduct across all sides of the market was harmful.78F[78]

In the platform context, understanding whether there is harm to competition at all requires an assessment of the effects of conduct on all sides of the platform. “[N]o economic basis exists for establishing a presumption that ‘harm’ on one side of a two-sided platform is sufficient to demonstrate that market output has been restricted, or that consumer welfare has otherwise been harmed.” In fact, “[s]eparating the two markets allows legitimate competitive activities in the market for general purposes to be penalized no matter how output-enhancing such activities may be.”79F[79]

The Texas Complaint, however, is built on the alleged harm of reduced revenue to publishers, without considering the corresponding benefit of lower prices to advertisers, or the net effect on consumers.80F[80]

Beyond that, antitrust law does not condemn conduct on the basis that an enforcer (or a court) is able to identify or hypothesize alternative conduct that might provide similar benefits at lower cost. In alleging that there are ostensibly “better” ways that Google could have pursued its product design, pricing, and terms of dealing, both the Texas Complaint and Omidyar Roadmap do just that—assert that, had the firm only selected a different path, an alternative could have produced even more benefits or an even more competitive structure. This line of thinking seems to be one motivation for the CTDAA’s remedies.

The reason that the possibility of “better” theoretical arrangements cannot serve as the basis for antitrust intervention is that there are limits to what can be achieved through intervention, not least because of limitations on legislators’ and enforcers’ knowledge about the competitive dynamics of the markets they seek to regulate.81F[81] A practice’s departure from a theoretical competitive benchmark may be inextricably linked to the social benefits it generates. When this is the case, enforcement that requires the practice or product to change in order to adhere to a theoretical standard may end up undermining the benefits of the practice in the first place. That is particularly true in the context of the sort of “vertical foreclosure” arguments leveled against Google in the advertising space, in which it is alleged that the combination of different levels of the ad-supply chain by Google limits the ability of competitors to enter and compete effectively.82F[82] It is surely conceivable that the product improvements conferred by the combination of different functions into a single platform—e.g., greater efficiency, realization of network effects, more effective targeting—could be replicated by a different means that might also facilitate “even more competition.” But such an approach is fraught with the risk of serious and costly error.83F[83]

The alleged cure of tinkering with benefit-producing unilateral conduct by applying an “even more competition” benchmark is worse than the supposed disease. The adjudicator is likely to misapply such a benchmark, deterring the very conduct the law seeks to promote. As then-Judge Stephen Breyer explained in the context of above-cost low pricing (another “defect” that both the Texas Complaint and the Roadmap claim constitutes anticompetitive conduct by Google84F[84]), “the consequence of a mistake” is “to penalize a procompetitive price cut,” conduct that, from an antitrust perspective, is “the most desirable activity.”85F[85] That commentators or enforcers may be able to imagine alternative, theoretically more desirable, conduct is beside the point.

Similarly, subjecting the kinds of product-design decisions at issue in the Google case to refined balancing of benefits and harms would deter innovation. “To weigh the benefits of an improved product design against the resulting injuries to competitors is not just unwise, it is unadministrable. There are no criteria that courts can use to calculate the ‘right’ amount of innovation, which would maximize social gains and minimize competitive injury.”86F[86] Put simply, “no one can determine with any reasonable assurance whether one product is ‘superior’ to another.”87F[87]

For these reasons, a “product improvement by itself does not violate Section 2, even if it is performed by a monopolist and harms competitors as a result.”88F[88] “Any other conclusion would unjustifiably deter the development and introduction of those new technologies so essential to the continued progress of our economy.”89F[89] A benefit-creating product design, even if it hinders rivals, is “necessarily tolerated by the antitrust laws.”90F[90]

Nor does U.S. law condemn a firm’s decision not to share a product improvement with rivals on terms rivals might prefer, even when such sharing might lead to greater competition in the short term. “Compelling” innovators “to share the source of their advantage” with rivals, among other evils, “may lessen the incentive for the monopolist, or rival, or both” to invest in innovation.91F[91] Except in extremely limited circumstances, firms can decide the terms on which they offer their products and services.

Directly rejecting the Roadmap’s suggestion—and the CTDAA’s mandate—of compelling dealings on terms that might produce greater competition, the Supreme Court has decreed that the “Sherman Act . . . does not give judges carte blanche to insist that a monopolist alter its way of doing business whenever some other approach might yield greater competition.”92F[92] Firms are not obliged to go into new lines of business or abandon existing lines to throw lifelines to rivals.93F[93]

The law similarly encourages vertical integration, because it tends to foster innovation-enhancing synergies and lower prices by eliminating double marginalization.94F[94] As the Roadmap elsewhere admits, it is “not in itself uncommon” to see vertical integration result in “fewer and fewer companies,” even “in competitive markets.”95F[95] Thus, vertical integration by internal expansion—even by a monopolist—is presumptively lawful.96F[96] The Roadmap and the CTDAA, however, simply disregard this, instead presumptively condemning vertical integration that hinders rivals by creating efficiencies.97F[97] Again, this is simply not a defensible interpretation of U.S. antitrust law, nor should it be.

III. Allegations Against Google in Digital Display Advertising

Critics of the digital advertising industry—and Google’s role in it—have leveled numerous allegations. These include claims that Google “leverages” its ownership of YouTube to obtain and exert market power in the buying and selling of other digital-display ads. Some claim that Google anticompetitively uses cross-subsidies, charging supercompetitive prices at one end of the ad tech stack to subsidize supra-competitive prices at another end of the stack. It is also alleged that Google has superior information about consumers that it will not provide to competitors, giving Google an anticompetitive advantage. It is claimed that steep entry barriers—some allegedly erected by Google—inhibit entry and allow the company to achieve a supercompetitive “take rate” from its intermediation services. While the lawsuits may provide additional information and data to support these claims, we argue that, with the limited public information available to us, it is not clear that any of them constitute anticompetitive conduct.

A.   ‘Leveraging’ market power in video streaming into the digital open-display market

The Omidyar Roadmap argues that Google, by virtue of its vertical integration throughout the intermediary stack and into the supply side (as the owner of YouTube), has the incentive and ability to derive unwarranted benefits from its display advertising business. It alleges, for example, that, by offering a single interface for placing both search and display ads, “Google leverages its monopoly position in search to coerce advertisers into using Google’s display products.”98F[98] In support it cites the CMA as saying:

Google may also be able to leverage its market power in search into the open display market. Smaller advertisers often choose to single-home to minimize transaction costs. Advertisers that wish to single-home have a strong incentive to use Google Ads as they can use it to access Google search advertising and YouTube inventory as well as the open display market.99F[99]

An earlier version of the Texas Complaint echoed these claims:

Google’s practice of withholding YouTube video inventory from rival ad buying tools… effectively locks single-homing small advertisers into Google’s ad buying tool. In addition, other providers of ad buying tools indicate that it does not make economic sense to try to compete with Google Ads for small advertisers, because they cannot achieve sufficient scale with smaller advertisers who want to buy display, YouTube, and even search ads, through just one tool.100F[100]

And, similarly, the Roadmap also argues that most sources of demand for Google Ads purchase ad space through AdX because Google “designed its exchange in such a way that it operates more efficiently with requests from Google’s own ad server than it does when requests come from rival ad servers.”101F[101]

All these assertions describe efficiency-enhancing behavior as anticompetitive. The report does not allege that Google preferences its own ad exchange in ways that harm advertisers; rather, the company’s products simply work better together (which is not unusual when different software products must interact) and it is thus in advertisers’ best interests for Google to act this way.

U.S. law, rightly, does not consider efficiencies obtained from vertical integration in this way to be anticompetitive. Nor do efficiencies that rivals cannot beat qualify as “barriers to entry.” The alternative—requiring Google to refrain from using the cheapest and/or fastest option available, because doing so makes its product better than all competitors—would mean reduced innovation, higher overall costs, and no benefit to either advertisers or publishers.

Later, the Roadmap makes another similar allegation: that Google “leverages” its ownership of YouTube, and the fact that only Google’s DSP can place ads on YouTube, to give itself an anticompetitive advantage in open-display advertising because rival DSPs are inherently limited by being unable to place ads on YouTube. An earlier version of the Texas Complaint echoed this claim.102F[102]

The Roadmap characterizes this conduct as “a contractual way to deny interoperability,”103F[103] but there is no contractual restraint here. How Google distributes YouTube’s ad inventory is a unilateral distribution decision permitted under U.S. law. And Google’s policy is not unusual in any way: many other websites that carry video advertising—including Hulu, Instagram, and Twitter—self-distribute their own inventory and do not make it available for resale by third parties.104F[104] Google does not have a duty to maximize its competitors’ profits by allowing them to resell YouTube inventory.

Access to YouTube is also not essential to a DSP’s success. Before Google stopped third-party platforms from buying YouTube ad inventory, it reported that only a “small amount” of buying was being done through Google’s AdX, which allowed third-party platforms to bid. At the time, AdExchanger reported that “[b]y ’small amount,’ that reportedly means 5%.”105F[105] A competing DSP, TubeMogul, said that this decision was an “unfortunate development” but “immaterial, since less than 5% of total ad spend through our software in Q2 was directed to YouTube.”106F[106]

This is consistent with the fact that there are several successful DSP competitors that compete with Google, despite not having access to YouTube’s ad inventory. The Trade Desk went public for $1 billion in 2016, processed more than $6.2 billion in transactions in 2021, and had a market cap of more than $25 billion in the first week of August 2022. 107F[107] Other DSPs, like Amazon’s and Xandr (formerly AppNexus), both continue to compete with Google vigorously without access to YouTube inventory, as the Omidyar Roadmap admits in the case of AppNexus.108F[108]

The Roadmap further -alleges that Google’s owned-and-operated properties—including Search, YouTube, Shopping, Flights, and News—confer an anticompetitive advantage because “Google pays no ‘traffic acquisition costs’” for the advertising space on its own sites: “When Google places ads on YouTube, just as when it places ads on its own search results pages, Google pays no ‘traffic acquisition costs’ because it needn’t pay any publisher for access to the ‘eyeballs’ that will see or interact with the ads it helps place.”109F[109]

Google’s parent Alphabet reported that the company’s traffic-acquisition costs were approximately 20% of its revenues in 2021.110F[110] Over the past few years, 40-50% of Alphabet’s expenditures have been on “cost of revenues,”111F[111] and of these, roughly half have involved traffic-acquisition costs.112F[112] Alphabet defines traffic-acquisition costs as (a) “the amounts paid to our distribution partners who make available our search access points and services” and (b) amounts paid for ads displayed on Google Network Members properties. It identifies “distribution partners” as browser providers, mobile carriers, original equipment manufacturers, and software developers.

Contrary to the Roadmap’s insinuations, there is nothing to suggest that these expenditures become less burdensome as a company increases in scale. Indeed, the opposite may be true, if it is more costly to gain access to marginal users than inframarginal ones, consistent with Google’s traffic-acquisition costs increasing over the years as it has grown.113F[113] While Google does not have to pay itself for the use of its own display inventory, there is clearly an opportunity cost to displaying its own inventory rather that of another firm. The claim that the company faces no traffic-acquisition costs for these properties is inaccurate.

The Roadmap’s focus on traffic-acquisition costs also overlooks content-acquisition costs—the payments to content providers from whom Google licenses video and other content for distribution on ad-driven and subscription services such as YouTube and Google Play. While Google does not pay a publisher for access to “eyeballs” on its owned-and-operated properties, it pays substantial and increasing amounts for content on those properties that attract the “eyeballs.”114F[114] Alphabet CFO Ruth Porat indicates, for example, that YouTube pays content creators “a majority of our revenue.”115F[115] Leaving this expense out of the calculation is another example of the over-simplification that characterizes many of the claims that Google’s ad tech business is a simple (and simply anticompetitive) business.

B.   Excess pricing

Where Google’s critics diverge most significantly from the spirit of U.S. antitrust law is in their overriding concern for how advertising revenues are distributed among the recipients of advertisers’ payments: intermediaries (Google) and publishers. The Texas Complaint alleges that Google has a higher “take rate” than competing exchanges,116F[116] is able to increase its take rates without losing market share,117F[117] and “manipulates auctions to increase its take rate.”118F[118] This follows the Roadmap’s speculation—based on the CMA Interim Report—that Google may take a larger cut of advertising spending than its competitors.119F[119] And these allegations echo claims made in another report that Google introduces “hidden” fees that increase the overall cut it takes from ad auctions.120F[120]

First, it should be noted that the basis for these claims in the Roadmap are drawn from the CMA investigation’s interim report, published in December 2019. In the final report, after further investigation, the CMA abandoned this claim. The final report describes the CMA’s analysis of all the billions of Google Ad Manager open auctions related to U.K. web traffic during the period between March 8 and March 14, 2020. This, according to the CMA, allowed it to observe any possible “hidden” fees, as well. The CMA concludes:

Our analysis found that, in transactions where both Google Ads and Ad Manager (AdX) are used, Google’s overall take rate is approximately 30% of advertisers’ spend. This is broadly in line with (or slightly lower than) our aggregate market-wide fee estimate outlined above. We also calculated the margin between the winning bid and the second highest bid in AdX for Google and non-Google DSPs…. We found that Google’s average winning margin was similar to that of non-Google DSPs. Overall, this evidence does not indicate that Google is currently extracting significant hidden fees. As noted below, however, it retains the ability and incentive to do so.121F[121]

This is a crucial finding that severely undermines the allegations that Google extracts excessive or “hidden” fees. It also undermines the claim that there are “missing funds accruing to Google.” While these conclusions do not eliminate the possibility that the industrywide price could itself be above competitive levels (and it remains to be seen whether the plaintiffs states in the Texas case will produce different evidence), they do mean that the best evidence currently available calls into question the charge that Google exploits a lack of interoperability by prioritizing its own products or that it engages in opaque pricing to conceal hidden charges of which its customers are unaware.

More fundamentally, absent evidence of Google deceiving advertisers and publishers to extract above-competitive margins, claims that its prices are “too high” or its revenue sharing “too low” are at odds with established antitrust law. U.S. antitrust law does not attempt to derive “proper” prices and impose these obligations on companies to ensure a “fair” outcome. Absent anticompetitive defects in the process, even monopolists are free to charge monopoly prices. The alternative would be for some agency—a court or a regulator—to regulate pricing and second guess every business decision made by dominant firms.

C.   Cross-subsidies

At the same time, the Roadmap alleges that Google can “charge low prices at one end of the stack, to drive out competitors, while charging high prices at the other to counterbalance any losses.”122F[122] But even if true, this would not be anticompetitive. It is a widely understood feature of platforms that they can shift prices from one side of a multisided market to another to maximize the platform’s total value. For example, a marketplace may make sellers bear the burden of fraud or mis-selling to give assurance to customers, and grow the consumer side of the platform market, just as a ridesharing app may discount rides to attract customers to build a large enough base to induce drivers onto the app.

This is a normal part of platform economics, which has long recognized that offering one side a low, zero, or negative price can be efficient and procompetitive.123F[123] As the U.S. Supreme Court held in Ohio v. American Express, an integrated competitive-effects analysis should look at the overall effect on output, not the effect on one side of the market; the relevant market must include both sides of the platform or the market would not exist at all.124F[124] There is no reason to think that this kind of behavior would generally be classed as “predatory pricing” in the absence of other behavior, such as raising prices after driving out competitors.125F[125]

But neither the Texas Complaint nor the Roadmap allege that Google’s prices were predatory. On the contrary, their sole claim in this respect is that, after being acquired by Google, DoubleClick lowered its prices (by a factor of ten, according to the Roadmap126F[126]), which it then maintained at these lower levels. This price reduction is facially procompetitive, however. It is unusual, to say the least, to describe a price reduction, with no subsequent price rises, as anticompetitive. If less-efficient competitors were unable to compete with these lower prices, that is competition in action. The law does not preclude nonpredatory low prices, nor even predatory prices without recoupment.127F[127] Sustained price reductions are one of the primary goals of antitrust.

Moreover, the source of the Roadmap’s claim that these price reductions were done “to drive out competitors” was, notably, a company that was not actually driven out of business by these price reductions. The source was an ad server, Smart, which claimed that Google’s price reductions “made the provision of publisher ad server difficult to sustain as a standalone business. This was the main reason why Smart felt the need to expand into the provision of SSP services.”128F[128] A competitor of Google’s responding to price reductions by broadening its own offerings is, again, procompetitive, not anticompetitive.

D.   Data gathering and integration

The Texas Complaint and the Roadmap describe several pro-privacy measures Google has adopted or plans to adopt as being detrimental to its competitors, including the decision to disable third-party cookies (which allow digital advertising companies to track users across the web to serve them relevant targeted ads) on the Chrome browser.129F[129] The Complaint argues that this shift benefits Google to the detriment of other ad tech companies, because (it says) Google, but not its competitors, has other data sources it can use to target ads at users.130F[130] In the same vein, the Roadmap points to Google’s decision not to share with advertisers raw data that it compiles about users.131F[131]

The Complaint ignores regulatory causes of these changes altogether, and the Roadmap dismisses the suggestion that they may be driven by the European General Data Protection Regulation (GDPR), on the basis that “data sharing here in the U.S., where we have no privacy regulatory scheme akin to that which is in place in Europe” has also been curtailed.132F[132] Both forget data-privacy laws in U.S. states, such as the California Consumer Privacy Act (CCPA). And, even if that weren’t the case, the claim that GDPR would have no effect outside Europe ignores that companies may find it easier to comply with such laws by changing their practices globally, rather than on a country-by-country or state-by-state basis.133F[133] Many companies have done this: Microsoft, for example, announced in November 2019 that it would “honor California’s new privacy rights throughout the United States.”134F[134]

Both also ignore the possibility that these provisions may be a response to demand from users of Google Chrome. Google may have good reasons to maintain a reputation for protecting user privacy, particularly because of the wide range of services it provides where user privacy is often of paramount importance to many users: Search, Maps, Gmail, YouTube, and Chrome itself.

Apple and Mozilla, neither of which has a significant online display advertising arm (and thus, have no incentive to block cookies simply to disadvantage display advertisers, as the Complaint alleges Google has done) have taken similar steps to increase user privacy. These are direct competitors of Google Chrome’s, and when Apple made blocking third-party cookies the default in its Safari browser, it was reported by one major outlet as “beating Google by two years to the privacy feature.”135F[135] Indeed, one of the reasons that Google delayed its disabling of third-party cookies was reportedly to implement technologies to “make it easier for advertisers to target certain demographics without laser-sighting down to specific people, ensure that the infrastructure many sites use for logins don’t break, and help provide some level of anonymous tracking so advertisers can know if their ads actually converted into sales.”136F[136]

That Chrome’s competitors, neither of which has an incentive to hurt ad tech companies, have taken the same steps that the Complaint alleges Google is taking for anticompetitive reasons should be compelling evidence that Google, too, is responding to user demand and/or regulation. Under U.S. law, the fact that these are legitimate moves and benefit users interested in privacy—and, indeed, may be a response to competition in the browser market—undermines claims that Google has failed to maximize competition along other dimensions.

The Roadmap also presents a hypothetical circumstance that amounts to an allegation that Google “captures” data from ads served to publishers to “expropriate” publishers’ investments in content that attracts a particular audience:

Some publishers have invested in content that attracts and retains a specific type of consumer, for example, readers of the Wall Street Journal or Golf Digest; this in turn allows them to support their business by selling valuable ads to advertisers looking for exposure to those audiences. Google has two ways to expropriate that value. First, rather than serve an ad on the Wall Street Journal at a high price, it can track the user who visited the Wall Street Journal and wait until she visits a site that sells space at low prices, for example, a local recipe blogger. Google can then sell Wall Street Journal users to advertisers in a way that does not benefit the Wall Street Journal at all and costs advertisers much less. A second strategy used by Google is to take the data describing these differentiated audiences and use it to create an imitation portfolio of consumers that mimic the characteristics of the publisher’s audience. For example, Google could create an audience of consumers similar to the people who read Golf Digest. Then Google sells access to this group of consumers when they visit inexpensive websites. Advertisers are happy to buy these ads because the consumers likely belong to the specialized audience of interest but are available at a much lower price. In these ways the unique audience assembled by the publisher is copied and expropriated.137F[137]

It should be noted that the Roadmap does not conclude that Google engages in these practices, but merely describes strategies Google “can” undertake to “expropriate” publishers’ investments. The Roadmap concedes that advertisers would be “happy” under such hypotheticals, because they are buying effective ads at a “much lower price.” In the Roadmap’s example, the hypothetical recipe blogger is “happy” that it earned revenues from selling an impression and the advertiser is “happy” that it paid a lower price than it would have had the impression been sold to the Wall Street Journal. The Wall Street Journal may not be so “happy” that it did not serve that particular ad, but that display space did not sit empty; the hypothetical lost ad was replaced by another impression that was served. And it is neither Google’s nor antitrust law’s job to make specific publishers better off—nor to make publishers better off at the expense of advertisers—but to ensure that the market as a whole is competitive and acting in consumers’ interests.

These hypotheticals again highlight the tensions discussed above between the different sides of a multisided market. Actions that make advertisers “happy” may come at the expense of publishers’ advertising revenues and actions that increase publishers’ revenues may increase costs to advertisers. One of the goals of a multisided market intermediary such as Google is to balance these competing interests to maximize total revenues flowing through the ad tech stack.

The Roadmap concludes that, through its “entire family of products,” Google collects and analyzes substantial amounts of information about its users. It uses this information to maximize the “effectiveness,” “precision,” and “value” of the ads it intermediates.

First, Google offers an entire family of products—everything from Gmail and Google Maps to the Google Calendar, Google Chrome, Android mobile operating system and the search engine—that gather valuable personal data about its users. Second, the products across the ad stack further collect data on consumer activities that the company then integrates to maximize the effectiveness and precision of ad targeting and attribution and thereby the value of the ads.138F[138]

Rather than “expropriating” publishers’ data, it would be reasonable to conclude that Google is adding value to the data provided by publishers, advertisers, and consumers to better target ads. For example, the Wall Street Journal may not know that a consumer recently did a Google Search for “running shoes.” By adding valuable information from Search, the consumer might be served a relevant running shoe ad on the Wall Street Journal’s site. This benefits the publisher who is paid for serving a valuable impression, the advertiser who sells a pair of shoes, and the consumer who obtains useful information and purchases the product she was seeking.

E.   Accelerated Mobile Pages (AMP) and header bidding

The Texas Complaint, like the Roadmap, alleges that Google designed Accelerated Mobile Pages (AMP) “[t]o respond to the threat of header bidding… [by making it] essentially incompatible with JavaScript and header bidding. Google then used its power in the search market to effectively force publishers into using AMP.”139F[139] But this gets several key facts wrong. First, AMP is an open-source industry collaboration project and Google cannot unilaterally impose a design standard on it. 140F[140] Second, a version of header bidding can work with AMP.141F[141] Third, it is mistaken to assert that “non-AMP-formatted results often do not even show up on the first page of results, regardless of their relevance.”142F[142] AMP has been a prerequisite only for inclusion in the top news story carousel, while other listings are ranked by relevance and speed.

Importantly, the argument ignores the main benefit of AMP to publishers and users: faster load times for mobile users who may be on slow connections. One of header bidding’s chief downsides is that it increases page-load latency. It is obvious why an HTML framework built to maximize load times would not be compatible with header bidding. Because AMP confers undisputed benefits on users and publishers, Google and the other companies involved in the AMP project have no obligation to re-engineer AMP to be compatible with header bidding. Any conclusion otherwise would involve a court deciding that users should be forced to use a slower Internet so that websites can use header bidding.

F.   Alleged barriers to entry in the open digital display ad market

Claims about Google’s alleged market power in display advertising rest on assumptions that the company enjoys the benefits of significant barriers to entry throughout the ad tech stack, thus enabling it to extract supercompetitive rents without fear of competition: “With these barriers in place,” it is claimed, “entry seems nearly futile.”143F[143]

A key element in establishing a company’s durable market power—and thus, its ability to impose anticompetitive costs on its users—is the presence of entry barriers. Even a market with only a single company—a true monopoly—cannot act like a monopoly if entry into its market is easy; if it did profitably raise prices, new competitors would enter the market and undercut it.144F[144]

As the Roadmap concedes, “[m]arket power is not permanent, of course. It can be undercut by, among other things, new entrants that offer better quality or lower prices.”145F[145] This notion of “contestability” is a fundamental part of assessing the competitiveness of markets under U.S. antitrust law.146F[146] In the absence of barriers to entry, it is well-established that assumptions of future competitive harm from ongoing conduct cannot be sustained.147F[147] Thus, the Roadmap bases much of its brief against Google on the presence of barriers to entry, “which heighten[] the prospect that Google can engage in conduct that harms competition without restraint from new entrants or potential new entrants.”148F[148] On the strength of these asserted barriers, the Roadmap’s authors interpret ambiguous conduct as anticompetitive.

According to the Roadmap, “the CMA’s findings reveal a number of significant barriers to entry into the digital advertising market.”149F[149] But most of its assertions in this regard are flawed, either because the CMA did not, in fact, make “findings” in the ways it suggests, or else because it reaches incorrect conclusions that certain conduct constitutes a barrier.150F[150] The Texas Complaint’s assertions of similar barriers to entry are likewise problematic.151F[151]

1.     Consumer location information

Although the Texas Complaint does not discuss it as an explicit barrier to entry,152F[152] one of the Omidyar Roadmap’s key assertions about barriers to entry relates to Google’s access to user-location data. It asserts that:

The CMA concluded that Google has nearly insurmountable advantages in access to location data, due to the location information it receives from the Android operating system, Google search, and other applications…. An entrant into the ad tech stack requires information about the consumer to target an ad effectively. Because Google accounts for nearly the entirety of the mobile search sector in the UK—97%—and controls many of the known sources of location data, such an entrant faces a large barrier to entry.153F[153]

But the CMA does not, in fact, “conclude[] that Google has nearly insurmountable advantages in access to location data,” either in the CMA Interim Report to which the Roadmap refers, nor in the CMA Final Report. The CMA never makes any claim of “insurmountable advantage.” Indeed, it does not use the word “insurmountable” at all, except to note that “rival platforms did not suggest that accessing consumer data was an insurmountable barrier to entry.”154F[154] Rather, to support this claim, the Roadmap cites to a portion of the CMA Interim Report recounting a suggestion made by Microsoft regarding the “critical” value of location data in providing relevant advertising.155F[155]

Moreover, that portion of the CMA Interim Report, as well as the suggestion made by Microsoft, is about search advertising, not display advertising. While the CMA does not characterize this data in the way the Roadmap claims, it does allege that “Google has exclusive access to a large amount of user data that can be used for targeted advertising and for measuring advertising outcomes, collected through its consumer-facing services. Data collected on its search platform is particularly valuable for targeting purposes in open display as it reveals users’ purchasing intent.”156F[156]

While location data may also be valuable for display advertising in a comparable way, it is not clear that the GPS-level data that is so valuable in providing mobile-search ad listings is particularly useful for display advertising, which may be just as well-targeted by less granular city- or county-level location data.

Consider the Roadmap’s illustrative example:

A digital ad for a brick-and-mortar running store in Des Moines is of little use to a runner looking to test out new shoes in Omaha, and, if shown to the Omaha runner, is unlikely to prompt a click, much less a purchase, from the Des Moines store.157F[157]

This is certainly correct. But GPS or even cell-tower location data is not necessary to determine in which city a user is located. Publicly available databases of IP address locations can provide this information, and they are readily and often freely available to all competitors. It is difficult to imagine that display advertising uses location data at any greater level of granularity except in unusual circumstances; it simply would not be particularly useful or effective.158F[158]

Furthermore, to the extent that location data (like other consumer data) may be useful for display advertising, the most significant issue affecting its availability to advertisers is not Google’s presence in the ad tech stack; it is privacy regulations that limit the collection, use, and sharing of such data.159F[159] These privacy regulations, such as the GDPR, limit the ability of digital firms to sell or otherwise pass user data to third-party advertisers. What seems like unequal treatment, in this regard, is really a case of privacy regulation in action.

These laws may have the indirect effect of favoring larger digital conglomerates that can collect user data through one service and use it to target ads in another. In this sense, it could be true that Google has informational advantages over rivals, though in a critically different way than that alleged by the Roadmap. But it can hardly be considered anticompetitive if the source of such advantage is legal constraints on information sharing. Indeed, an empirical study by economists Avi Goldfarb and Catherine Tucker found that (pre-GDPR) privacy regulation in the EU “restricted advertisers’ ability to collect data on Web users in order to target ad campaigns. We ?nd that, on average, display advertising became far less effective at changing stated purchase intent after the EU laws were enacted, relative to display advertising in other countries.”160F[160] Along similar lines, Nils Wernerfelt and his co-authors show that access to data from different sources significantly improves ad targeting.161F[161] In turn, this may give a competitive advantage to firms that operate several successful web services and applications.

As Israeli competition law scholars Michal Gal and Oshrit Aviv found with respect to the GDPR, privacy regulations can function as a barrier to entry in several further ways.162F[162] These include creating new economies of scale associated with regulatory compliance, increased litigation risk, and uncertainty around interpretation of the rules. Because they serve to make reputation more central, they also can lead users to become more likely to entrust their data to incumbents but not to unknown, new entrants.163F[163]

2.     Attribution measurement

“Attribution” refers to the method by which advertisers can see which ad led a user to an action, such as visiting a website or making a purchase.164F[164] The Roadmap alleges that Google can design attribution to mislead advertisers by, for example, favoring search ads over display ads.165F[165] This would lead to more of the advertiser’s money going to Google instead of (in part) to a publisher, and (assuming, as the Roadmap implies, that this makes ad campaigns less effective166F[166]) can harm advertisers by misleading them into choosing a less-effective advertising channel.

The Roadmap provides no evidence this is taking place. What it describes is more a complaint about the nature of search advertising in general: that companies will sometimes end up paying for ads in lieu of identical organic search results for their pages. That is an argument to be had elsewhere, but there are clear reasons why it may be in an advertiser’s interest to advertise even in these situations. Paid search ads may give them greater control over how a link is displayed to a user (for example, with text the advertiser has chosen, rather than text that the search engine has retrieved) or guarantee a prominent listing for searches where the advertiser’s URL listing is not always guaranteed to be on top.

Apart from the broader objection to the nature of search advertising, the Roadmap’s authors also object to Google setting an attribution default in its DSP. But advertisers can choose from several different attribution models, not just the default one that the Roadmap objects to, which attributes to search ads the “last click.” Other options include “last non-direct click,” which “ignores direct traffic and attributes 100% of the conversion value to the last channel that the customer clicked through from before buying or converting”; “last Google Ads click”; “First interaction”; and others that give attribution weights according to where and when the customer saw or used the ads during their purchase or conversion “journey.”167F[167] These are precisely the kinds of models that the Roadmap’s authors implicitly believe are more appropriate for campaigns heavy on display advertising. Advertisers can also create their own custom models, and Google has published guides for advertisers to help them choose among the different models.168F[168]

The Roadmap’s objection is thus reduced to being about the choice by Google to use the “last click” attribution model as the default. But some model has to be the default, and “last click” is also the default on, for example, Microsoft Advertising.169F[169] Indeed, according to digital-ad-intermediary company, Outbrain (one of Google’s competitors), it is the most common attribution default across the industry.170F[170] For the Roadmap’s objection to carry any weight, a case based on this claim would need to demonstrate that it was unusual for Google to use “last click” as the default attribution. Even then, given the ease with which advertisers can change the attribution model, the charge would be thin.

3.     ‘Opaque’ pricing

Both the Texas Complaint and the Roadmap allege that Google’s “opaque pricing” constitutes a barrier to entry by impeding “advertisers from switching to a lower-cost for higher-quality” buying tool.171F[171] As the Roadmap puts it, a “new or potential new PAS or DSP cannot credibly claim to be able to undercut the Google products on price if the publishers and advertisers cannot tell how much Google actually is charging.”172F[172] The Texas Complaint further alleges that “Google compounds its exclusionary auction manipulations by purposefully keeping its auction mechanics, terms, and pricing, opaque and ‘nontransparent.’ This makes it nearly impossible for publishers and advertisers to discover Google’s misrepresentations, and even harder for rivals to neutralize or offset.”173F[173] Both the Texas Complaint and the Roadmap also suggest that competition is undermined when publishers and advertisers do not know the fee structure of the intermediaries they are using.

But it is not unusual for businesses’ costs and prices to be private to their competitors, and it is not a barrier to competition. Grocery stores do not need to know how much it cost a farmer to grow an orange or how much their rivals are paying for transportation, unless they are attempting to anticompetitively coordinate their prices; they just need to work to make their own costs as low as possible and to reduce their prices to consumers by as much as possible. Similarly efficient firms are perfectly able to offer competitive prices simply by making the best offer based on their own fundamentals; only less-efficient firms will struggle (as they should).

Along these lines, for competition to work effectively in display advertising, Google’s competitors do not need to know what Google is charging; they need to offer a price and product that is more attractive overall to potential customers than Google’s is. Similarly, a publisher does not need to know how much an advertiser bid to place an ad, nor does the advertiser need to know how much the publisher received to serve the ad. The advertiser’s competition concern is whether an effective ad can be served at a lower price from a different intermediary and the publisher’s competition concern is whether it can earn greater revenues from a different intermediary. One should not be surprised that Google does not reveal information on which competing intermediaries can free ride. Indeed, this is widely considered to be one of the hallmarks of vigorous competition.

Conclusion

As we have argued, many of the most significant claims made against Google’s ad tech products are based on a misunderstanding of U.S. antitrust law, or of the details of the ad tech market itself. Although we cannot be sure how the Texas, et al. v. Google case will develop once the allegations in the Complaint are fleshed out into full arguments, many of its initial claims and assumptions are wrongheaded. Based on the information currently available, if the court rules in favor of these, the result will be to condemn procompetitive conduct and potentially to impose costly, inefficient remedies that function as a drag on innovation.

Legislators, too, who may be concerned about Google’s conduct and tempted to impose regulatory requirements on it and other tech companies should bear in minds the risk of the Nirvana fallacy, in which real-life conduct is compared against a hypothetical “competition maximizing” benchmark, and anything that falls short is deemed problematic and in need of intervention.174F[174] That approach would pervert the incentives of businesses to innovate and compete, and would make an unobtainable “perfect” that exists only in the minds of some economists and lawyers the enemy of a “good” that exists in the market right now.

 

[1] Third Amended Complaint, Texas v. Google, 21-md-3010-PKC (S.D.N.Y. Jan 14, 2022) at 105 (hereinafter, “Texas Complaint”).

[2] See Complaint, United States v. Google LLC, No. 1:20-CV-03010 (D.D.C. Oct. 20, 2020); see also, Complaint, State of Colorado, et al. v. Google LLC, 1:20-CV-03715 (D.D.C. Dec. 17, 2020).

[3] DoJ Expected to File Antitrust Lawsuit Against Google in Weeks—Bloomberg News, U.S. News (Jul. 14, 2022), https://money.usnews.com/investing/news/articles/2022-07-14/doj-expected-to-file-antitrust-lawsuit-against-google-in-weeks-bloomberg-news.

[4] Antitrust: Commission Opens Investigation into Possible Anticompetitive Conduct by Google in the Online Advertising Technology Sector, European Commission (Jun. 22, 2021), https://ec.europa.eu/commission/presscorner/detail/en/ip_21_3143.

[5] Bundeskartellamt Publishes Report on Non-Search Online Advertising for Public Discussion, Bundeskartellamt (Aug. 29, 2022), https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2022/29_08_2022_SU_Online_Werbung.html?nn=3599398.

[6] Id.

[7] Online Platforms and Digital Advertising Market Study, U.K. Competition and Markets Authority (Jul. 1, 2020), https://www.gov.uk/cma-cases/online-platforms-and-digital-advertising-market-study (hereinafter “CMA Market Study”); Online Platforms and Digital Advertising, Market Study Final Report, U.K. Competition and Markets Authority (Jul. 1, 2020), https://assets.publishing.service.gov.uk/media/5fa557668fa8f5788db46efc/Final_report_Digital_ALT_TEXT.pdf (hereinafter “CMA Final Report”), at 21 & 37.

[8] See Josh Frydenberg, Competition and Consumer (Price Inquiry—Digital Advertising Services) Direction 2020 (Feb. 10, 2020); Ad Tech Inquiry Issues Paper 5, Australian Competition & Consumer Commission (Mar. 10, 2020); Digital Advertising Services Inquiry, Australian Competition & Consumer Commission (Feb. 26, 2021), https://www.accc.gov.au/focus-areas/inquiries-finalised/digital-advertising-services-inquiry/submissions-to-interim-report.

[9] Investigation of Competition in Digital Markets, Majority Staff Report and Recommendations, Subcommittee on Antitrust, Commercial and Administrative Law of the Committee on the Judiciary (Oct. 4, 2020), https://judiciary.house.gov/uploadedfiles/competition_in_digital_markets.pdf?utm_campaign=4493-519; Hearing on Stacking the Tech: Has Google Harmed Competition in Online Advertising?, Committee of the Judiciary, Subcommittee on Antitrust, Competition Policy and Consumer Rights (Sep. 15, 2020), https://www.judiciary.senate.gov/meetings/stacking-the-tech-has-google-harmed-competition-in-online-advertising.

[10] Investigation of Competition in Digital Markets, Majority Staff Report and Recommendations, id., at 20.

[11] Competition and Transparency in Digital Advertising Act, S.4258, 117th Congress (2021-2022), https://www.congress.gov/bill/117th-congress/senate-bill/4258/text.

[12] Lee Introduces Digital Advertising Act, Mike Lee US Senator for Utah (May 19, 2022), https://www.lee.senate.gov/2022/5/lee-introduces-digital-advertising-act.

[13] Support Mounts for Lee’s Digital Advertising Act, Mike Lee US Senator for Utah (May 27, 2022), https://www.lee.senate.gov/2022/5/support-mounts-for-lee-s-digital-advertising-act.

[14] Online Platforms and Digital Advertising Market Study Interim Report, U.K. Competition and Markets Authority (Dec. 18, 2019), https://assets.publishing.service.gov.uk/media/5ed0f75bd3bf7f4602e98330/Interim_report_—_web.pdf (hereinafter, “CMA Interim Report”).

[15] Fiona M. Scott Morton & David C. Dinielli, Roadmap for a Digital Advertising Monopolization Case Against Google, Omidyar Network (May 2020), https://omidyar.com/wp-content/uploads/2020/09/Roadmap-for-a-Case-Against-Google.pdf (hereinafter, “Roadmap” or “Omidyar Roadmap”). One of the Roadmap’s authors testified at a Senate hearing on the display-advertising market, and the report has been widely cited. See, e.g., Gilad Edelman, Here’s What an Antitrust Case Against Google Might Look Like, Wired (May 18, 2020), https://www.wired.com/story/antitrust-case-against-google-roadmap-paper.

[16] Damien Geradin & Dimitrios Katsifis, An EU Competition Law Analysis of Online Display Advertising in the Programmatic Age, 15 Eur. Comp. J. 55 (2019); Damien Geradin & Dimitrios Katsifis, “Trust Me, I’m Fair”: Analysing Google’s Latest Practices in Ad Tech from the Perspective of EU Competition Law, 16 Eur. Comp. J. 11 (2020); Damien Geradin & Dimitrios Katsifis, Online Platforms and Digital Advertising Market Study: Observations on CMA’s Interim Report, TILEC Discussion Paper No. DP2020-044 (Feb. 13, 2020), https://ssrn.com/abstract=3537864; Damien Geradin & Dimitrios Katsifis, Competition in Ad Tech: A Response to Google, TILEC Discussion Paper No. DP2020-038 (Jun. 3, 2020), https://ssrn.com/abstract=3617839.

[17] The markets alleged in the Texas Complaint involve (1) publisher ad servers, (2) ad exchanges, (3) ad-buying tools for large advertisers, (4) ad-buying tools for small advertisers, (5) in-app mediation tools, and (6) in-app networks. The complaint does not relate to other forms of advertising on the Internet, such as targeted text-based ads sold by search engines, video ads that run before or during video content, or shareable ads on social media platforms.

[18] This section is distilled from our much longer discussion of the broader market surrounding digital advertising. See Eric Fruits, Geoffrey A. Manne & Lazar Radic, Relevant Market in the Google AdTech Case, ICLE Issue Brief 2022-06-01 (2022), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4125569.

[19] Benedict Evans, News by the Ton: 75 Years of US Advertising (Jun. 15, 2020), https://www.ben-evans.com/benedictevans/2020/6/14/75-years-of-us-advertising; Benedict Evans, TV, Merchant Media and the Unbundling of Advertising (Mar. 18, 2022), https://www.ben-evans.com/benedictevans/2022/3/18/unbundling-advertising.

[20] See Fruits, Manne & Radic, supra note 18.

[21] Michael Schneider & Kate Aurthur, R.I.P. Cable TV: Why Hollywood Is Slowly Killing Its Biggest Moneymaker, Variety (Jul. 21, 2020), https://variety.com/2020/tv/news/cable-tv-decline-streaming-cord-cutting-1234710007 (“[B]asic cable feasted on a dual revenue stream of subscriber fees and advertising revenue. But that gravy train started going off the rails when the streaming services arrived.”).

[22] At the same time, as Benedict Evans notes, not all digital advertising is drawn from offline sources: “[I]f you talk to people at both Google and Facebook and in the agency world, you’ll hear that a lot of the money spent on Google and Facebook is money that was never spent on traditional advertising—it’s coming from SMEs [small and medium enterprises] and local businesses that might have spent in classified at most but probably wouldn’t have done even that.” Evans, News by the Ton, supra note 19 (emphasis in original).

[23] See Xi He, Rigoberto Lopez & Yizao Liu, Are Online and Offline Advertising Substitutes or Complements? Evidence from U.S. Food Industries, 15 J. Agricultural & Food Indus. Org. 1 (2017).

[24] David Bardey, Jorge Tovar & Nicolas Santos, Characterization of the Relevant Market in the Media Industry: Some New Evidence, Toulouse School of Economics Working Paper 16-719 (2016), https://www.tse-fr.eu/publications/characterization-relevant-market-media-industry-some-new-evidence (“The results show substitution and complementary patterns across certain media outlets. An increase in price for advertising in radio, for instance, leads to higher demand for newspapers and outdoors. Similarly, complementarity relationships between media outlets are observed, suggesting that advertising across the various media platforms is, overall, interwoven.”).

[25] David S. Evans, The Online Advertising Industry: Economics, Evolution, and Privacy, 23 J. Econ. Persp. 37, 49 (2009).

[26] Avi Goldfarb & Catherine Tucker, Search Engine Advertising: Channel Substitution When Pricing Ads to Context, 57 Management Sci. 458 (2011) (The authors find that the price of “ambulance chaser” lawyer ads was significantly more expensive in states prohibiting direct-mail solicitation by attorneys. This leads them to conclude that “online advertising substitutes for online advertising”).

[27] Avi Goldfarb & Catherine Tucker, Substitution Between Offline and Online Advertising Markets, 7 J. Competition L. & Econ. 37, 43 (2011).

[28] Daniel S. Levine, Ad-Supported Cyber-Magazines to Launch on Internet, Adweek (Sep. 10, 1993).

[29] Id.

[30] Brian Morrissey, How the Banner Ad Was Born, Digiday (Apr. 12, 2013), https://digiday.com/marketing/how-the-banner-ad-was-born.

[31] Chris Lapham, AOL and GNN Partner to Build Launch Pad, CMC Magazine (Jul. 1, 1995), https://www.december.com/cmc/mag/1995/jul/cutting.html.

[32] Kim Cleland, Poppe Creates Web Net, Advertising Age (Oct. 30, 1995).

[33] Id.

[34] The History of Online Advertising, OKO Ad Management (Jul. 19, 2019), https://oko.uk/blog/the-history-of-online-advertising.

[35] The company changed its name to Overture, which was acquired by Yahoo! in 2003.

[36] GoTo.com Announces First Round of Financing, Totaling More Than $ 6 Million, Led by Draper Fisher Jurvetson, Business Wire (May 19, 1998), https://www.internetnews.com/marketing/goto-com-raises-6-million-in-first-round-of-financing.

[37] Will Oremus, Google’s Big Break, Slate (Oct. 13, 2013), https://slate.com/business/2013/10/googles-big-break-how-bill-gross-goto-com-inspired-the-adwords-business-model.html.

[38] Id.

[39] Dean Schmid, The History of Display Advertising: Everything You Need to Know, DisruptorDaily.com (Aug. 14, 2017), https://www.disruptordaily.com/the-history-of-display-advertising-everything-you-need-to-know.

[40] Julia Scheeres, Death of Banner Ads Exaggerated, Wired (Jan. 26, 2001), https://www.wired.com/2001/01/death-of-banner-ads-exaggerated.

[41] Id.

[42] Breaking News, AdAge (Oct. 23, 2000).

[43] Mark Evans, Investors Leap off Overture Roller Coaster: Rival Google Elbows In, National Post (Feb. 21, 2002).

[44] Oremus, supra note 37.

[45] Google Grabs Applied Semantics, EuropeMedia (Apr. 25, 2003); Google Expands Advertising Monetization Program for Websites, Google Press Release (Jun. 18, 2003), http://googlepress.blogspot.com/2003/06/google-expands-advertising-monetization.html.

[46] Dean Schmid, The History of Display Advertising: Everything You Need to Know, DisruptorDaily (Aug. 14, 2017), https://www.disruptordaily.com/the-history-of-display-advertising-everything-you-need-to-know.

[47] Kate Walsh, Search Marketing: Understanding the Basics, B2B Marketing Magazine (March 2006).

[48] Louise Story & Miguel Helft, Google Buys DoubleClick for $3.1 Billion, The New York Times (Apr. 14, 2007), https://www.nytimes.com/2007/04/14/technology/14DoubleClick.html.

[49] Sarah Sluis, The Year Header Bidding Went Mainstream, AdExchanger (Dec. 27, 2016); Townhall Media Selects OpenX for Patent-Pending Header Bidding Solution, BusinessWire (Sep. 18, 2015), https://www.businesswire.com/news/home/20150918005110/en/Townhall-Media-Selects-OpenX-for-Patent-Pending-Header-Bidding-Solution.

[50] As the name suggests, ad waterfalls enable publishers to sell their inventory seriatim, beginning with premium, direct sales and flowing through the most historically profitable ad servers in succession to unload unsold inventory before offering its remnant inventory in the open display channel. See, e.g., Maciej Zawadzinski, What Is Waterfalling and How Does it Work?, Clearcode (Aug. 20, 2021), https://clearcode.cc/blog/what-is-waterfalling.

[51] See, e.g., Header Bidding, OKO Ad Management, https://oko.uk/topic/header-bidding (retrieved July 27, 2022).

[52] Client-side header bidding is so-named because it operates via a small piece of java script embedded in the header of a publisher’s website and executed within the user’s browser (i.e., client). See, e.g., Maciej Zawadzinski, What Is Header Bidding and How Does it Work?, Clearcode (Aug. 20, 2021), https://clearcode.cc/blog/what-is-header-bidding.

[53] Header Bidding Facts and Statistics 2021, Automatad (Jun. 27, 2021), https://headerbidding.co/header-bidding-statistics. Today, 70% of the top 10,000 U.S. publishers use header bidding. See Header Bidding (HBIX) Tracker, kevel (retrieved Nov. 1, 2022), https://www.kevel.com/hbix.

[54] See, e.g., CMA Final Report, supra note 7, at Appendix M, ¶ 33.

[55] See, e.g., Vishveshwar Jatain, Header Bidding Integrations: Client Vs. Server-Side, Explained, Blockthrough (Apr. 15, 2021), https://blockthrough.com/blog/header-bidding-integrations-client-vs-server-side-explained.

[56] IAB and PwC, IAB Internet Advertising Revenue Report, 2010 Full Year Results (Apr. 2011), available at https://www.iab.com/wp-content/uploads/2015/05/IAB_Full_year_2010_0413_Final.pdf; Megan Graham, Digital Ad Revenue Jumped 35% in the U.S. Last Year, Biggest Gain Since 2006, Wall Street Journal (Apr. 12, 2022), https://www.wsj.com/articles/digital-ad-revenue-jumped-35-in-the-u-s-last-year-biggest-gain-since-2006-11649759401.

[57] Producer Price Index by Commodity: Advertising Space and Time Sales: Internet Advertising Sales, Excluding Internet Advertising Sold by Print Publishers, U.S. Bureau of Labor Statistics, https://fred.stlouisfed.org/series/WPU365; Producer Price Index, December 2009—February 2021, U.S. Bureau of Labor Statistics, https://fred.stlouisfed.org/graph/?g=vtTd.

[58] Scott Morton & Dinielli, supra note 15, at 9.

[59] Importantly, however, network effects are not monolithic; nor do they increase forever. For different types of networks at different points in their growth, adding more users might not increase the value of the platform and could even reduce the platform’s benefits. See, e.g., D’Arcy Coolican & Li Jin, The Dynamics of Network Effects, Andreesen Horowitz (Dec. 13, 2018), https://a16z.com/2018/12/13/network-effects-dynamics-in-practice.

[60] See Stan J. Liebowitz & Stephen E. Margolis, Network Externality: An Uncommon Tragedy, 8 J. Econ. Persp. 133 (1994).

[61] Catherine Tucker, Digital Data, Platforms and the Usual [Antitrust] Suspects: Network Effects, Switching Costs, Essential Facility, 54 Rev. Indus. Org. 683, 686 (2019).

[62] See, e.g., David S. Evans, Economics of Vertical Restraints for Multi-Sided Platforms, University of Chicago Institute for Law & Economics Olin Research Paper No. 626 (Jan. 2, 2013), https://ssrn.com/abstract=2195778.

[63] See, e.g., Stylianos Despotakis, R. Ravi & Amin Sayedi, First-Price Auctions in Online Display Advertising, 58 J. Marketing Research 888 (2021). See also Display Advertising Switched to First-Price Auctions After Adoption of Header Bidding, New Study Finds, Tepper School of Business (Apr. 22, 2020), https://www.cmu.edu/tepper/news/stories/2020/april/display-advertisting-research-ravi.html.

[64]  Jonathan Levin, Auction Theory (Oct. 2004), available at https://web.stanford.edu/~jdlevin/Econ%20286/Auctions.pdf.

[65] Maciej Zawadzi?ski, How Do First-Price and Second-Price Auctions Work in Online Advertising?, Clearcode (Aug. 12, 2021), https://clearcode.cc/blog/first-price-second-price-auction.

[66] Id.

[67] Texas Complaint, supra note 1, at ¶ 351 (“Overall, the lack of transparency prevents more efficient competition that would drive greater innovation, increase the quality of intermediary services, increase output, and create downward pricing pressure on intermediary fees.”); Scott Morton & Dinielli, supra note 15, at 18 (“Based on the public facts known at the moment, however, it does not seem plausible that the incremental efficiencies created by the conduct described here could outweigh all the harms to competition resulting from this broad pattern of behaviors.”); Scott Morton & Dinielli, supra note 15, at 38 (“It also is true that Google has allowed some rivals to survive (although not necessarily to thrive). It is possible that Google adopted a strategy of incomplete foreclosure specifically so that it can paint an illusion of healthy competition when the reality is quite different. Indeed, to the extent Google has adopted ‘pro-competitive’ concessions, the narrative here demonstrates that they simply have not succeeded in addressing the harms or lowering the barriers to entry.”).

[68] Id. at 3 (“It is clear even to us as lay people that there are less anticompetitive ways of delivering effective digital advertising—and thereby preserving the substantial benefits from this technology—than those employed by Google.”).

[69] United States v. Aluminum Co. of America, 148 F.2d 416, 430 (2nd Cir. 1945) (Learned Hand, J.) (emphasis added).

[70] Cal. Computer Prods., Inc. v. Int’l Bus. Machine Corp., 613 F.2d 727, 744 (9th Cir. 1979) (“IBM, assuming it was a monopolist, had the right to redesign its products to make them more attractive to buyers whether by reason of lower manufacturing cost and price or improved performance. It was under no duty to help [its competitors] survive or expand.”).

[71] Scott Morton & Dinielli, supra note 15 at 18.

[72] Texas Second Amended Complaint at ¶ 113.

[73] See Verizon Commc’ns Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398, 408 (2004). The exception—”at or near the outer boundary of § 2 liability” (id. at 409)—is the extremely narrow case in which a monopolist (i) sacrifices profits, by (ii) terminating a prior course of dealing, (iii) for no purpose except to harm competition. See Novell v. Microsoft, 731 F.3d 1064, 1074-75 (10th Cir. 2013) (Gorsuch, J.) (holding that a refusal-to-deal claim requires terminating “a preexisting voluntary” course of dealing where the “monopolist decided to forsake short-term profits,” and “the monopolist’s conduct” is “irrational but for its anticompetitive effect”).

[74] Opinion and Order, Texas, et al. v. Google, 21-md-3010-PKC (S.D.N.Y, Sep. 13, 2022) (citing Charych v. Siriusware, Inc., 790 Fed. App’x 299, 302 (2nd Cir. 2019)).

[75] United States v. Colgate & Co., 250 U.S. 300, 307 (1919).

[76] Trinko, 540 U.S. at 407-08

[77] See Brunswick Corp. v. Pueblo Bowl-O-Mat, Inc., 429 U.S. 477, 488 (1977) (“The antitrust laws, however, were enacted for ‘the protection of competition not competitors.’”) (quoting Brown Shoe Co. v. United States, 370 U.S. 294, 320 (1962)).

[78] See Ohio v. American Express, 138 U.S. 2274, 2285 (2018) (“Due to indirect network effects, two-sided platforms cannot raise prices on one side without risking a feedback loop of declining demand…. Price increases on one side of the platform [] do not suggest anticompetitive effects without some evidence that they have increased the overall cost of the platform’s services.”).

[79] Geoffrey A. Manne, In Defence of the Supreme Court’s ‘Single Market’ Definition in Ohio v American Express, 7 J. Antitrust Enforcement 104, 111 (2019) (quoting Brief for Amici Curiae Antitrust Law & Economics Scholars in Support of Respondents at 19, Ohio v. American Express, 138 U.S. 2274 (2018) (No. 16-1454) and United States, et al. v. American Express, 838 F.3d 179, 198 (2nd Cir. 2016)).

[80] Among innumerable examples, see Texas Complaint, supra note 1, at ¶ 297 (“Google’s harm to the competitive process has harmed customers in this market, i.e., online publishers.”). Notably, the Texas Complaint does, in places, recognize that identifying the incidence of benefits and harms in multisided markets is complex—it just fails to carry its analysis to its logical conclusion. Thus, in ¶157 the Complaint notes that “[t]he higher advertising revenue publishers make from exchanges permits publishers to offer consumers better quality content and lower-priced or free access to their content.” (Emphasis added). Undoubtedly, this is true. But if it is correct, then it must also be correct that, at the same time, the correspondingly higher prices advertisers pay for advertising through exchanges limits their ability to provide marketing benefits directly to consumers and may increase the price to consumers of the advertised goods. It is an empirical question which effect is larger, but the mere possibility that one set of consumers could benefit from a different arrangement is insufficient on its own to identify harm when another set of consumers would be harmed by it.

[81] See Harold Demsetz, Information and Efficiency: Another Viewpoint, 12 J. L. & Econ. 1, 1-2 (1969) (“In practice, those who adopt the nirvana viewpoint seek to discover discrepancies between the ideal and the real and if discrepancies are found, they deduce that the real is inefficient…. The nirvana approach is… susceptible… to committing three logical fallacies—the grass is always greener fallacy, the fallacy of the free lunch, and the people could be different fallacy.”) (emphasis in original).

[82] See, generally, Thomas Nachbar, Less Restrictive Alternatives and the Ancillary Restraints Doctrine, Virginia Law and Economics Research Paper No. 2020-18 (2021) (forthcoming U. Seattle L. Rev.) at 57-8, available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3723807 (“The more general risk to tech markets comes from the intangible nature of the products and services they produce. Although many of the cases cited for less restrictive alternatives are horizontal cases, it is the vertical context (which normally receives more permissive antitrust review) in which less restrictive alternatives present the greatest likelihood of destabilizing current law because of the difficulty of specifying what is and is not less restrictive with regard to the intangible products produced by today’s ‘big tech’ economy. To the extent that less restrictive alternatives present problems of incrementalism, those problems will be exacerbated in the ‘big tech’ markets.”).

[83] See Geoffrey A. Manne, Error Costs in Digital Markets, Global Antitrust Institute Report on the Digital Economy (Joshua D. Wright & Douglas H. Ginsburg, eds., 2020) 33, 76, available at https://gaidigitalreport.com/wp-content/uploads/2020/11/Manne-Error-Costs-in-Digital-Markets.pdf (“The concern with error costs is especially high in dynamic markets in which it is difficult to discern the real competitive effects of a firm’s conduct from observation alone. And for several reasons, antitrust decision-making in the context of innovation tends much more readily toward distrust of novel behavior, thus exacerbating the risk and cost of over-enforcement.”).

[84] Among many other examples, see Texas Second Amended Complaint at ¶138 (“Then, through Dynamic Allocation, Google’s ad server passed inside information to Google’s exchange and permitted Google’s exchange to purchase valuable impressions at artificially depressed prices. Competing exchanges were deprived of the opportunity to compete for inventory and left with the low-value impressions passed over by Google’s exchange.”); Omidyar Roadmap, supra note 15, at 20 (“[A]fter purchasing DoubleClick, which became its publisher ad server, Google apparently lowered its prices to publishers by a factor of ten, at least according to one publisher’s account related to the CMA. Low prices for this service can force rivals to depart, thereby directly reducing competition.”).

[85] Barry Wright Corp. v. ITT Grinnell Corp., 724 F.2d. 227, 235 (1st Cir. 1983) (Breyer, C.J.).

[86] Allied Orthopedic Appliances, Inc. v. Tyco Health Care Grp. LP, 592 F.3d 991, 1000 (9th Cir. 2010).

[87] Berkey Photo, Inc. v. Eastman Kodak Co., 603 F.2d 263, 287 (2nd Cir. 1979). See also, Manne & Wright, Innovation and the Limits of Antitrust, 6 J. Comp. L. & Econ. 153–202 (March 2010), https://doi.org/10.1093/joclec/nhp032.

[88] Allied Orthopedic, 592 F.3d at 999-1000; see also, California Computers Prods. v. IBM, 613 F.2d 727, 744 (9th Cir. 1979); Foremost Pro Color, Inc. v. Eastman Kodak Co., 703 F.2d 534, 543-45 (9th Cir. 1983).

[89] Foremost Pro Color, 703 F.2d at 543.

[90] Allied Orthopedic, 592 F.3d at 1000.

[91] Verizon Commc’ns, Inc. v. Law Offices of Curtis V. Trinko, 540 U.S. 398, 400-41 (2004).

[92] Trinko, 540 U.S. at 415-16.

[93] Id.; see also New York Merc. Exch., Inc. v. Intercontinental Exch. Inc., 323 F.Supp.2d 559 (S.D.N.Y. 2004) (dismissing Section 2 claim and reiterating limited exceptions in which forced sharing is appropriate).

[94] See, e.g., Jack Walters & Sons Corp. v. Morton Bldg., Inc., 737 F.2d 698, 710 (7th Cir. 1984) (“We just said that vertical integration is not an improper objective. But this puts the matter too tepidly; vertical integration usually is procompetitive. If there are cost savings from bringing into the firm a function formerly performed outside it, the firm will be made a more effective competitor.”). There is a robust body of empirical research indicating that vertical integration is generally procompetitive or benign. For a summary of the leading meta-studies by DOJ and FTC economists and others, see Koren W. Wong-Ervin, Antitrust Analysis of Vertical Mergers: Recent Developments and Economic Teachings, The Antitrust Source (February 2019), https://www.americanbar.org/content/dam/aba/publishing/antitrust_source/2018-2019/atsource-february2019/feb19_wong_ervin_2_18f.pdf. See also, Francine Lafontaine & Margaret Slade, Vertical Integration and Firm Boundaries: The Evidence, 45 J. Econ. Lit. 677 (2007) (“In spite of the lack of unified theory, overall a fairly clear empirical picture emerges. The data appear to be telling us that efficiency considerations overwhelm anticompetitive motives in most contexts. Furthermore, even when we limit attention to natural monopolies or tight oligopolies, the evidence of anticompetitive harm is not strong.”). See also, generally, Geoffrey A. Manne, Kristian Stout & Eric Fruits, The Fatal Economic Flaws of the Contemporary Campaign Against Vertical Integration, 68 Kansas L. Rev. 923 (2020).

[95] Scott Morton & Dinielli, supra note 15, at 17.

[96] See, e.g., Port Dock & Stone Corp. v. Oldcastle Ne., Inc., 507 F.3d 117, 123-25 (2nd Cir. 2007) (affirming dismissal of a Section 2 claim and finding that even a monopolist’s “vertical expansion into another level of the same product market will ordinarily be for the purpose of increasing its efficiency, which is a prototypical valid business purpose”). Moreover, single-firm conduct that supposedly projects power into another market, even through anticompetitive means, does not violate Sherman Act Section 2 unless the practices threaten monopoly power in that distinct second market. Harming competition is not enough. See Trinko, 540 U.S. at 415 n.4 (citing Spectrum Sports, Inc. v. McQuillan, 506 U.S. 447, 459 (1993)).

[97] Thus, the Omidyar Roadmap condemns Google’s supposed integration of data “to maximize the effectiveness and precision of ad targeting and attribution and thereby the value of an ad,” Scott Morton & Dinielli, supra note 15, at 20, even though the conduct makes Google’s offering to advertisers more attractive.

[98] Id. at 18-19.

[99] CMA Interim Report, supra note 14, at ¶ 5.89.

[100] Texas Second Amended Complaint at ¶ 113.

[101] Scott Morton & Dinielli, supra note 15, at 22.

[102] Texas Second Amended Complaint at ¶¶ 284-91 (“Cutting off access to YouTube foreclosed competition in the ad buying tool markets and protected Google’s market power in these markets. Many DSPs stopped growing, many others went out of business, and the market overall has been closed to entry.”).

[103] Scott Morton & Dinielli, supra note 15, at 22.

[104] See, e.g., Ryan Joe, The Big Story: Call of the Peacock, AdExchanger (Jan. 22, 2020) at 31:05-31:26, https://www.adexchanger.com/podcast/the-big-story/the-big-story-call-of-the-peacock (indicating that NBC’s Peacock streaming service will have only direct sales when it launches); Kevin Weiss, What Is the Amazon Demand Side Platform (DSP)?, Amplio (July 2019), https://www.ampliodigital.com/blog/what-is-the-amazon-demand-side-platform-dsp (“Amazon DSP is the only way to access advertising inventory exclusively available on Amazon’s collection of owned online properties and devices like: Kindle; Fire TV; IMDb; Amazon Owned & Operated properties”); Tim Cross, Xandr Launches New Demand-Side Platform ‘Xandr Invest’, VideoAdNews (Jun. 10, 2019) https://videoadnews.com/2019/06/10/xandr-launches-new-demand-side-platform-xandr-invest (“Xandr [AT&T-Time Warner’s ad tech division] has announced it will be the exclusive source of inventory from Community, its recently announced video marketplace which includes content from various WarnerMedia brands as well as Vice, Hearst Magazines, Newsy, Philo, Tubi and XUMO.”).

[105] Neal Mohan, Focusing Investments to Improve Buying on YouTube, Google (Aug. 6, 2015) https://doubleclick-advertisers.googleblog.com/2015 (“To continue improving the YouTube advertising experience for as many of our clients as possible, we’ll be focusing our future development efforts on the formats and channels used by most of our partners. To enable that, as of the end of the year, we’ll no longer support the small amount of YouTube buying happening on the DoubleClick Ad Exchange.”); see also, Kelly Liyakasa, Google to Yank YouTube Inventory out of AdX by Year’s End, AdExchanger (Aug. 6, 2015), https://www.adexchanger.com/ad-exchange-news/google-to-yank-youtube-inventory-out-of-adx-by-years-end.

[106] Liyakasa, id.

[107] See Lara O’Reilly, Ad Tech Company The Trade Desk Goes Public at $28.75 Per Share—A Huge Pop on its $18 Price Target, Business Insider (Sep. 21, 2016), https://www.businessinsider.com/the-trade-desk-ipo-2016-9; Trey Titone, The Bill That Could Break Up Google and Shake Up Ad Tech, Ad Tech Explained (May 23, 2022), https://adtechexplained.com/competition-and-transparency-in-digital-advertising-act-ctda; Trade Desk Market Cap, YCharts, https://ycharts.com/companies/TTD/market_cap.

[108] Scott Morton & Dinielli, supra note 15, at 16 n.70 (identifying AppNexus as a “vigorous competitor to Google”).

[109] Id. at 2, 28-29.

[110] Annual Report (Form 10-K) for Year Ending December 31, 2021, Alphabet Inc. (Feb. 02, 2022), https://www.sec.gov/ix?doc=/Archives/edgar/data/0001652044/000165204422000019/goog-20211231.htm.

[111] Id.

[112] Id.

[113] Id.

[114] Id.

[115] Rachel Kaser, YouTube Claims to Share Billions in Ad Money with Creators, Unlike Instagram, The Next Web (Feb. 5, 2020), https://thenextweb.com/facebook/2020/02/05/youtube-claims-share-billions-ad-money-creators-unlike-instagram.

[116] Texas Complaint, supra note 1, at ¶¶ 61, 156, 253, 288.

[117] Id. ¶ 157.

[118] Id. ¶ 21.

[119] Scott Morton & Dinielli, supra note 15, at 14 (“The CMA estimates Google’s take rate, or price, at 40%, which it deems a supra-competitive price for the services provided by the Google-controlled players in ad tech stack. A recent study by the Incorporated Society of British Advertisers (ISBA) found that publishers received 51% of the price, while the amount they could track going to intermediaries was 34%. The study could not find where the remaining 15% of the price went. As we will describe below, Google has such a dominant position across all elements of the ecosystem, it seems likely that these missing funds are accruing to Google at least in part, which would support the CMA’s findings.”).

[120] Geradin & Katsifis, “Trust Me, I’m Fair,” supra note 16 (“[L]ack of competition across the ad tech chain enables Google to exploit advertisers and publishers by charging hidden fees for ad intermediation on top of its disclosed commission…. Unfortunately, we conclude that Google’s latest switch does nothing to increase auction transparency. Worse, it seems to strengthen Google’s ability to extract hidden margins from its customers, while undermining the competitive pressure exercised by header bidding.”).

[121] CMA Final Report, supra note 7, at 275 (emphasis added).

[122] Scott Morton & Dinielli, supra note 15, at 20.

[123] See, e.g., Jean-Charles Rochet & Jean Tirole, Platform Competition in Two-Sided Markets, 1 J. Eur. Econ. Ass’n 990 (2003); Bruno Jullien, Price Skewness and Competition in Multi-Sided Markets, IDEI Working Paper 504 (March 2008), available at https://core.ac.uk/download/pdf/6375977.pdf.

[124] See Joshua. D. Wright & John. M. Yun, Burdens and Balancing in Multisided Markets: The First Principles Approach of Ohio v. American Express, 54 Rev. Industrial Organization 717 (2019); Manne, In Defence of the Supreme Court’s ‘Single Market’ Definition in Ohio v American Express, supra note 79.

[125] As described here, true pricing is theoretically possible but difficult in practice: “To successfully engage in predatory pricing means taking enormous and rising losses that grow for the ‘predatory’ firm as customers switch to it from its competitor. And once the rival firm has exited the market, if the predatory firm raises prices above average cost (i.e., to recoup its losses), there is no guarantee that a new competitor will not enter the market selling at the previously competitive price. And the competing firm can either shut down temporarily or, in some cases, just buy up the ‘predatory’ firm’s discounted goods to resell later.” Sam Bowman, Buck’s “Third Way”: A Different Road to the Same Destination, Truth on the Market (Oct. 27, 2020), https://truthonthemarket.com/2020/10/27/bucks-third-way-a-different-road-to-the-same-destination.

[126] Scott Morton & Dinielli, supra note 15, at 20.

[127] See, e.g., Barry Wright, 724 F.2d at 234-35.

[128] CMA Interim Report, supra note 14, at Appendix H, ¶ 194.

[129] Texas Complaint, supra note 1, at ¶ 477.

[130] Id. at ¶¶ 473-476.

[131] Scott Morton & Dinielli, supra note 15, at 28.

[132] Id. at 28.

[133] As one study on the effects of GDPR (in this case, on app development) notes, “While 42.1 percent of EU-developed apps exit in the year following GDPR, the analogous ?gure averages between 37.7 and 50 percent in the other six countries, con?rming the di?culty in ?nding an untreated part of the world.” Rebecca Janßen, Reinhold Kesler, Michael E. Kummer, & Joel Waldfogel, GDPR and the Lost Generation of Innovative Apps, NBER Working Paper 30028 (May 2022) at 19-20, available at https://www.nber.org/papers/w30028.

[134] Julie Brill, Microsoft Will Honor California’s New Privacy Rights Throughout the United States, Microsoft Blog (Nov. 11, 2019), https://blogs.microsoft.com/on-the-issues/2019/11/11/microsoft-california-privacy-rights.

[135] Nick Statt, Apple Updates Safari’s Anti-Tracking with Full Third-Party Cookie Blocking, The Verge (Mar. 24, 2020), https://www.theverge.com/2020/3/24/21192830/apple-safari-intelligent-tracking-privacy-full-third-party-cookie-blocking.

[136] Dieter Bohn, Google to “Phase Out” Third-Party Cookies in Chrome, but not for Two Years, The Verge (Jan. 24, 2020), https://www.theverge.com/2020/1/14/21064698/google-third-party-cookies-chrome-two-years-privacy-safari-firefox.

[137] Scott Morton & Dinielli, supra note 15, at 30.

[138] Id. at 20.

[139] Texas Complaint, supra note 1, at ¶¶ 407-408. See also, Scott Morton & Dinielli, supra note 15, at 26.

[140] See David Besbris, Introducing the Accelerated Mobile Pages Project, for a Faster, Open Mobile Web, Google (Oct. 7, 2015), https://blog.google/products/search/introducing-accelerated-mobile-pages/.

[141] See Automated Team, Header Bidding on AMP—A Complete Guide, Automated (Jan. 10, 2020), https://headerbidding.co/header-bidding-amp.

[142] Scott Morton & Dinielli, supra note 15, at 26.

[143] Id. at 17.

[144] See William J. Baumol, Contestable Markets: An Uprising in the Theory of Industry Structure, 72 Am. Econ. Rev. 1, 14 (1982) (“In the limit, when entry and exit are completely free, efficient incumbent monopolists and oligopolists may in fact be able to prevent entry. But they can do so only by behaving virtuously, that is, by offering to consumers the benefits which competition would otherwise bring. For every deviation from good behavior instantly makes them vulnerable to hit-and-run entry.”).

[145] Scott Morton & Dinielli, supra note 15, at 15.

[146] See, generally, William J. Baumol, John C. Panzar, & Robert D. Willig, Contestable Markets and the Theory of Industry Structure (1982).

[147] United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001) (“Because a firm cannot possess monopoly power in a market unless that market is also protected by significant barriers to entry… it follows that a firm cannot threaten to achieve monopoly power in a market unless that market is, or will be, similarly protected.”).

[148] Scott Morton & Dinielli, supra note 15, at 15.

[149] Id.

[150] See CMA Final Report, supra note 7, at 252–55 for a discussion of barriers to entry.

[151] See Texas Complaint, supra note 1, at ¶ 127: In addition to these barriers, Google’s own anticompetitive conduct imposes additional barriers to entry and expansion. As addressed below in Section VII.A, from 2010 to present, Google has tied its ad server to its ad exchange, requiring publishers to use Google’s ad server in order to receive live, competitive bids from Google’s ad exchange. This tie effectively forces almost every large publisher to use Google’s ad server. And because it is difficult-to-impossible for a publisher to use multiple ad servers simultaneously, requiring publishers to use Google’s ad server effectively prohibits them from using a competitor’s ad server. Google’s anticompetitive conduct creates an unnatural and nearly insurmountable barrier to entry.

[152] An earlier version of the Texas Complaint did make assertions regarding Google’s abuse of monopoly power through the “use [of] its data advantages to trade on inside information” (Texas Second Amended Complaint at ¶ 311), by which the state plaintiffs may mean (or have meant) to encompass location data, among other things.

[153] Scott Morton & Dinielli, supra note 15, at 15.

[154] CMA Interim Report, supra note 14, at 189 (emphasis added).

[155] Id., at ¶ 3.71 (“Microsoft suggested that accessing at-scale location data from user devices is a critical input to providing relevant, localized results. It indicated its belief that Google has unique advantages in this area, due to the location data that it receives from the Android operating system and the location data it receives when users access Google Search or other apps like Google Maps/Waze.”).

[156] CMA Final Report, supra note 7, at ¶ 5.268.

[157] Scott Morton & Dinielli, supra note 15, at 15.

[158] To be sure, location data can be helpful in assessing the efficacy of advertising by, for example, enabling an advertiser to better evaluate whether an advertisement led users to go to the advertiser’s physical location. But this function hardly seems necessary to a well-functioning market, and other sources of such information (e.g., questionnaires) are available.

[159] Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the Protection of Natural Persons with Regard to the Processing of Personal Data and on the Free Movement of Such Data, and Repealing Directive 95/46/EC (General Data Protection Regulation), OJ L 119, 4.5, 2016; See, e.g., Bert Peeters, Processing of Location Data: Navigating the EU Data Protection Framework, CiTiP Blog (Feb. 4, 2021), https://www.law.kuleuven.be/citip/blog/processing-of-location-data-navigating-the-eu-data-protection-framework. (“The general understanding seems to be that, while European law does not qualify location data as a ‘special category’ of data under article 9 of the GDPR, location data should for all intents and purposes be treated with the utmost of care.”).

[160] Avi Goldfarb & Catherine Tucker, Privacy Regulation and Online Advertising, 57 Mgmt. Sci. 57, 57 (2011).

[161] Nils Wernerfelt, Anna Tuchman, Bradley Shapiro, & Robert Moakler, Estimating the Value of Offsite Data to Advertisers on Meta, University of Chicago, Becker Friedman Institute for Economics Working Paper No. 114 (August 22, 2022) at 1, available at https://ssrn.com/abstract=4198438 (“Taken together, our results suggest a substantial benefit of offsite data across a wide range of advertisers, an important input into policy in this space.”).

[162] Michal S. Gal & Oshrit Aviv, The Competitive Effects of the GDPR, 16 J. Competition & Econ. 349 (May 18, 2020). See also James Campbell, Avi Goldfarb, & Catherine Tucker, Privacy Regulation and Market Structure, 24 J. Econ. & Mgmt. Strategy 47, 68 (2015) (“[A] potential risk in privacy regulation is the entrenchment of the existing incumbent firms and a consequent reduction in the incentives to invest in quality. These incentives are stronger when firms have little consumer-facing price flexibility, as is the case in online media.”).

[163] Gal & Aviv, id. at 16.

[164] See, e.g., Cheok Lup, Explaining Marketing Attribution Models [Scenario Example], tinkerEdge (Nov. 12, 2015), https://www.tinkeredge.com/blog/web-analytics/explaining-marketing-attribution-models. (“On Day #1: User wants to purchase a coffee table for his new house, and perform a keyword search on Google. He clicks on one of the organic listings on Google Search Engine Result Page (SERP) to land onto Overstock.com. On Day # 2: He continues his search for his coffee table, and clicks on one of the PPC ads on Google SERP to land onto Overstock.com again. He subscribes to the email newsletter this time. On Day #3: He receives an eDM [electronic direct mail] from Overstock.com with a promotional offer of 30% discount sale, and clicks the “Buy Now” button from the eDM to enter the website. Unable to resist the discount offer, he decides to make a purchase of the furniture from the website.”). Attribution metrics determine which channel gets credit for the ultimate sale.

[165] Scott Morton & Dinielli, supra note 15, at 29

[166] “Furthermore, the default makes the advertiser believe that search ads are very effective relative to display ads, so the advertiser has no reason to change the default.” Id.

[167] See About the Default MCF Attribution Models: Learn How Each MCF Model Assigns Conversion Credit, Google Analytics Help (last visited Nov. 1, 2022), Attribution Models, Google, https://support.google.com/analytics/answer/1665189?hl=en. For the Google Analytics 4 version of these attribution models currently being implemented, see [GA4] About Attribution and Attribution Modeling, Google Analytics Help (last visited Nov. 1, 2022), https://support.google.com/analytics/answer/10596866. Google even created a guide called “Beyond Last Click Attribution” to help advertisers select the most appropriate model. See Beyond Last Click Attribution: Official Guide to Attribution Modeling in Google Ads, Google Ads Help (last visited Nov. 1, 2022), https://support.google.com/google-ads/answer/7003286.

[168] See Joan Arensman & Wilfred Yeung, Move Beyond Last Click Attribution in AdWords, Google Blog (May 10, 2016), https://adwords.googleblog.com/2016/05/move-beyond-last-click-attribution.html.

[169] How Does Conversion Tracking Work?, Microsoft Advertising (n.d.) https://help.ads.microsoft.com/#apex/ads/en/56710/2.

[170] Nir Elharar, How to Choose the Right Marketing Attribution Model for Your Content, Outbrain (Apr. 8, 2019), https://www.outbrain.com/blog/marketing-attribution-model-content.

[171] Texas Complaint, supra note 1, at ¶ 195.

[172] Scott Morton & Dinielli, supra note 15, at 17

[173] Texas Complaint, supra note 1, at ¶ 351.

[174] See Demsetz, supra note 81.

Antitrust Unchained: The EU’s Case Against Self-Preferencing

Abstract

Whether self-preferencing is inherently anticompetitive has emerged as perhaps the core question in competition policy for digital markets. Large online platforms who act as gatekeepers of their ecosystems and engage in dual-mode intermediation have been accused of taking advantage of these hybrid business models to grant preferential treatment to their own products and services. In Europe, courts and competition authorities have advanced new antitrust theories of harm that target such practices, as have various legislative initiatives around the world. In the aftermath of the European General Court’s decision in Google Shopping, however, it is important to weigh the risk that labeling self-preferencing as per se anticompetitive may merely allow antitrust enforcers to bypass the legal standards and evidentiary burdens typically required to prove anticompetitive behavior. This paper investigates whether and to what extent self-preferencing should be considered a new standalone offense under European competition law.

I. Introduction

In recent years, widespread concern has emerged that large digital platforms may misuse their market positions by giving preferential treatment to their own products and services. One fear is that, by engaging in self-preferencing, so-called “Big Tech” firms may be able not only to entrench their power in core markets, but also to extend it into associated markets.1F[1] Notably, by controlling ecosystems of integrated complementary products and services—which usually represent important gateways for business users to reach end users—dominant platforms may enjoy a strategic market status that allows them to exercise bottleneck power. As the argument goes, by acting as gatekeepers and regulators within their ecosystems, these platforms represent unavoidable trading partners and may pick winners and losers in the marketplace.

Moreover, digital platforms often serve a dual role, acting as both an intermediary and a trader operating on the platform. Hence, they may be tempted to influence results in their own favor (so-called “biased intermediation”). Indeed, once an intermediation platform is also active in complementors’ markets, it loses its status of neutrality and risks of discrimination against rivals may arise because of potential conflict of interests. Therefore, quoting a slogan delivered by U.S. Sen. Elizabeth Warren (D-Mass.) during the 2020 Democratic Party presidential primary campaign: “you get to be the umpire or you get to have a team in the game—but you don’t get to do both at the same time.”2F[2] European Commissioner for Competition Margrethe Vestager has used a similar sporting analogy—arguing that a platform cannot be both a player who competes against rivals in the downstream market and, at the same time, the upstream referee who determines the conditions of that competition.3F[3]

In short, self-preferencing may allow large digital platforms to adopt a leveraging strategy to pursue a twofold anticompetitive effect—that is, excluding or impeding rivals from competing with the platform (defensive leveraging) and extending their market power into associated markets (offensive leveraging). The latter scenario may take the form of envelopment, in which a platform attempts to both exclude rivals and facilitate its own entry into a target market by tying the core functionalities of its platform to the services offered in that market.4F[4]

The legislative initiatives that have been undertaken around the world posit that, to ensure a level playing field, digital gatekeepers must be prevented from engaging in various forms of self-preferencing. The European Union’s Digital Markets Act (DMA), for example, prohibits gatekeepers from: engaging in any form of self-preferencing in ranking services and products offered by the platform itself; using any non-publicly available data generated through activities by business users to compete with those users on the platform; preventing the removal of preinstalled applications; giving preferential access to hardware, operating-system, or software features to their own ancillary services; and refusing to grant “fair, reasonable, and non-discriminatory” (FRAND) access to app stores, search engines, and social-networking services.5F[5] The United Kingdom’s proposed regulatory regime for digital markets, which imagines the adoption of firm-specific codes of conduct for online platforms with “strategic market status,” includes self-preferencing as an example of exclusionary behavior that large digital platforms sometimes engage in when they exert control over an ecosystem.6F[6] The German Digitalization Act likewise includes a ban on platforms favoring their own offers when they mediate access to supply and sales markets, particularly in cases where they present their own offers in a more favorable manner, exclusively pre-install them on devices, or integrate them in any other way.7F[7]

The American Innovation and Choice Online Act (AICOA) would go even further. The bill would declare it unlawful to engage in conduct that would “unfairly preference the covered platform operator’s own products, services, or lines of business over those of another business user on the covered platform in a manner that would materially harm competition on the covered platform.”8F[8] Accordingly, for example, Google would be prevented from launching only Google Maps in response to a query for restaurants, or from placing Google services at the top of a search-results page unless it is accompanied by all possible rival services. Similarly, Amazon would be constrained from showcasing its branded products or favoring third-party products that use its fulfillment service, while Apple would be banned from supplying prominent app-search results for its own apps or even from preinstalling its own apps.9F[9]

These provisions and others like them would essentially treat digital platforms as common carriers, and therefore subject them to a neutrality regime and utilities-style regulation. In some markets, lawmakers have proposed even more stringent reforms designed to reduce digital platforms’ potential bottleneck and intermediation power, and to prevent conflicts of interest, such as requirements that intermediation and operating units be structurally separated, restrictions on lines of business, and imposed duties to deal.10F[10]

In addition to these legislative initiatives, self-preferencing has also emerged as a theory of harm before European courts and antitrust authorities. After all, much of the behavior prohibited explicitly in the DMA initially attracted attention as part of antitrust investigations. In particular, the ban against self-preferencing appears to have been informed by the European Commission’s decision in the Google Shopping case, in which Google was fined for having systematically demoted the results of competing comparison-shopping products on its search results pages, while having granted prominent placement to its own comparison-shopping service.11F[11] The fact that the decision came following a protracted seven-year investigation has been cited as evidence of the need for an ex ante prohibition of such practices, thus removing the annoying hurdles and burdens posed by standard antitrust analysis.

The European General Court recently upheld the Commission’s decision,12F[12] although it narrowed the original decision’s scope by focusing on the context in which the practice occurred. Rather than articulating a legal test for a new antitrust offense, the Court applied fact-specific criteria to examine the potential for discrimination by a search engine. This approach notably differs from defining self-preferencing as a standalone abuse, as has been supported by the European Commission and some national competition authorities (NCAs).13F[13]

The DMA, it should be noted, will not displace Europe antitrust rules;14F[14] rather, the law will be implemented alongside them. This heightens the potential for interpretative uncertainty regarding the degree to which self-preferencing will or ought to be treated, in practice, as an infringement of competition law. This paper therefore sets out to investigate whether, in the aftermath of the Google Shopping ruling, self-preferencing by digital platforms has peculiar features that justify its consideration as a new theory of harm.

Indeed, one of the primary challenges posed by treating self-preferencing as a competitive harm, from a competition-law perspective, is the lack of an obvious limiting principle.15F[15] Notably, recent European case law suggests that, rather than a standalone theory of harm, self-preferencing is a catch-all category that includes various practices already addressed by antitrust rules. The risk is that labeling self-preferencing as per se anticompetitive would merely provide antitrust authorities with the opportunity to elide the application of legal standards and evidentiary burdens traditionally required to prove anticompetitive behavior.

This paper calls for appreciation of the continuing wisdom of antitrust orthodoxy against the prevailing zeitgeist, arguing that many of the perceived limits of antitrust actually represent its virtues.16F[16] Indeed, the goal of competition law ought not be to satisfy urgent policy objectives. Rather, antitrust is about limiting principles, even where that means it is unpopular.17F[17]

The remainder of the paper is structured as follows. Section II provides an overview of the relevant traditional antitrust theories of harm and emerging case law to analyze whether and to what extent self-preferencing could be considered a new standalone offense in EU competition law. Section III investigates whether platform neutrality more generally belongs to the scope of competition law, according to its legal foundations and settled principles. Section IV concludes.

II. Self-Preferencing as a Standalone Offense

The debate over self-preferencing revolves around its novelty. Antitrust concerns are raised regarding the preferential treatment granted by a vertically integrated dominant firm to its own products and services because of the firm’s dual role as both host and competitor. This is of particular interest when such potential conflicts of interest may result in the leveraging of market power in adjacent lines of business in ways capable of producing exclusionary effects.

From this perspective, competitive risks associated with self-preferencing do not appear significantly different from those that emerge in any scenario of vertical integration. Vertical integration is, indeed, often procompetitive, specifically because it can be used to improve efficiency and reduce transaction costs. Furthermore, while there is some dispute as to whether a dominant firm is required to ensure a level playing field by treating rivals in the same way as it does its own businesses, competition law is already equipped with tools to forbid practices that pursue discriminatory leveraging strategies. The emergence of digital platforms does not, in and of itself, challenge antitrust enforcement. To investigate whether self-preferencing should be considered a new standalone offense, it is necessary to first analyze the scope of relevant antitrust prohibitions and to evaluate the peculiar features of self-preferencing, as illustrated by courts and antitrust authorities that have recently sanctioned this behavior.

A.   Traditional European Theories of Antitrust Harm

Although predatory pricing and loyalty rebates may sometimes lead a firm to favor its own downstream services, our attention will be devoted to those practices that appear closer to self-preferencing: namely, refusal to deal, tying, bundling and mixed bundling, margin squeezes, and discrimination. In particular, the last of these represents the most obviously relevant comparison, as the favorable treatment a platform grants to its own products and services entails discriminatory treatment of rivals.

Under European competition law’s non-discrimination provisions, preferential treatment may be investigated when a vertically integrated firm applies to rivals (primary line injury) or other partners (secondary line injury) more onerous conditions than it applies to its own downstream businesses.18F[18] The second-degree price discrimination is mainly addressed by Article 102(c) TFEU, which establishes the abusive character of applying dissimilar conditions to equivalent transactions with trading parties, thereby placing them at a competitive disadvantage. It has been noted that the provision may be considered a straightforward legal basis for a theory of self-preferencing, as shown by the case law that has predominantly applied the provision in settings where a vertically integrated dominant firm sought to advantage its downstream operations at the expense of rivals.19F[19]

In the aftermath of the MEO ruling and following the effects-based approach affirmed in Intel,20F[20] discrimination is not, in itself, problematic from the point of view of competition law.21F[21] As a consequence, not every disadvantage that affects some customers of a dominant firm will amount to an anticompetitive effect; competitive disadvantages cannot be presumed. Antitrust enforcers are instead required to consider all the circumstances of the relevant case, assessing whether there is a strategy to exclude from the downstream market a trading partner that is at least as efficient as its competitors.

Self-preferencing may also take the form of tying, bundling, or mixed bundling. In the first of these, a dominant player leverages its market position in the tying product, making the purchase of the latter subject to the acceptance of another (tied) product. Bundling refers to the way products are offered and priced. In the case of pure bundling, the products are only sold jointly in fixed proportions. In mixed bundling, the products are also made available separately, but the sum of prices when sold separately is higher than the bundled price.22F[22]

Any of these practices may lead to anticompetitive foreclosure in the tied market and, indirectly, in the tying market. The exclusion of as-efficient-competitors is key to triggering antitrust liability for competition foreclosure. Mixed bundling may be anticompetitive if the discount is so large that equally efficient competitors offering only some of the components cannot compete against the discounted bundle. With bundling, the greater the number of products on which the undertaking exerts market power, the stronger the likelihood of anticompetitive foreclosure. In the case of tying, if an insufficient number of customers would buy the tied product on its own to sustain competitors of the dominant undertaking in the tied market, the tying could lead to those customers facing higher prices. Finally, the risk of foreclosure in tying and bundling strategies is expected to be greater where the dominant player makes it last—e.g., through technical tying (i.e., designing a product in such a way that it only works properly with the tied product and not with alternatives offered by competitors).

As tying strategies can be implemented either through contractual terms or by technical means, antitrust authorities are increasingly prone to challenge platforms’ product-design decisions that favor their own products or services by limiting interoperability, thereby impeding compatibility with rival products or services.23F[23] In Microsoft, the European General Court argued that the ubiquity of a dominant player in the tying market is likely to foreclose competition in the tied market. The Court noted that the practice of bundling a specific piece of software to an operating system through pre-installation allows the tied product “to benefit from the ubiquity of that operating system … which cannot be counterbalanced by other methods of distributing media players.”24F[24]

Foreclosure also may arise when consumers obtain the tied product free of charge and are not prevented from obtaining rival services. In Google Android, the Commission fined Google for having engaged in a leveraging practice to preserve and strengthen its position in the search-engine market by requiring device manufacturers to preinstall Google Search and the Chrome browser as preconditions to license the Google Play app store. By locking down Android in the Google-controlled ecosystem, manufacturers wishing to pre-install Google apps were prevented from selling smart-mobile devices that run on versions of Android not approved by Google (so-called Android “forks”).25F[25] According to the Commission, pre-installation can create a status quo bias, which reduces the incentives for manufacturers to pre-install competing search and browser apps, as well as the incentives for users to download such apps. The therefore affects rivals’ ability to compete effectively with Google. Despite the fact that Android is mostly distributed as open-source software, the Commission rejected both of the justifications Google put forward: that leveraging practices reflected a legitimate appropriation strategy to preserve incentives to innovate in a regime of weak appropriability26F[26] and that fork restrictions fell under governance rules needed to protect multi-sided platforms from negative externalities (in this case, preventing software fragmentation and the potential diffusion of incompatible versions of the software).27F[27]

Taken to its extreme, self-preferencing can result in refusals to deal,28F[28] which explains why European policymakers have invoked the essential facilities doctrine to address such cases. The aim of the doctrine, which imposes on dominant firms a duty to deal with all who request access, is to prevent a firm with control over an essential asset from excluding rivals or from extending its monopoly into another stage of production. Because it requires sacrificing the dominant firm’s freedom of contract and right to property, however, it may weaken incentives to invest, innovate, and compete.

These refusal-to-deal infringements are, under European competition law, generally limited to “exceptional circumstances.” According to Magill, a refusal to deal may trigger an antitrust violation when: (i) access to the product or service is indispensable to a firm’s ability to do business in a market; (ii) the refusal is unjustified; (iii) the refusal excludes competition on a secondary market; and (iv), if intellectual property rights are involved, it prevents the emergence of a new product for which there is potential consumer demand.29F[29] The IMS30F[30] and Microsoft31F[31] judgments substantially dismantled the third and fourth requirements, respectively, by considering the secondary-market requirement met even if that market is just potential or hypothetical, and the new product requirement satisfied even when access to the facility is necessary for rivals to develop follow-on innovation (i.e., improved products with added value).

Nonetheless, pursuant to the interpretation provided in Bronner, the requirement that a requested facility be indispensable remains in place and represents the last bulwark against the dangers of uncontrolled application of the doctrine.32F[32] Indeed, access to an input is considered indispensable if there are no technical, legal, or even economic obstacles that would render it impossible (or even unreasonably difficult) to duplicate. To demonstrate the lack of realistic potential alternatives, a requesting firm must establish that it is not economically viable to create the resource on a scale comparable to that of the firm controlling the existing product or service.

Against this background, the recent Slovak Telekom judgment introduced a relevant novel claim that the conditions laid down in Bronner do not apply where the dominant undertaking does give access to its infrastructure but makes that access subject to unfair conditions.33F[33] In addition, the Court of Justice (CJEU) implied that enforcers do not have to prove indispensability when access to a facility has been granted as a result of a regulatory obligation.34F[34] The implications are particularly relevant to digital markets, as the regulatory framework established by the DMA requiring access to platforms designated as gatekeepers would exempt antitrust authorities from having to demonstrate the indispensability of those facilities.

Finally, self-preferencing may be construed as a “margin squeeze,” which EU competition law defines as a standalone abuse that undermines the condition of equality of opportunity between economic operators. The European Commission initially equated this practice to a constructive refusal to deal, noting that, instead of refusing to supply, a dominant undertaking charges a price for a product on the upstream market that would not allow even an equally efficient competitor to trade profitably in the downstream market on a lasting basis.35F[35] The Commission therefore introduced the so-called Telefonica exceptions to categorize a specific class of cases where Bronner’s requirements would not apply. These exceptions hold that an obligation to supply cannot have negative effects on the input owner’s and/or other operators’ incentives to invest and innovate upstream.36F[36]  The CJEU has, however, gradually moved toward rejecting the concept of an implicit refusal to grant access, holding that margin squeezes should be treated as a separate theory of harm, thereby introducing an even broader exception to Bronner.

Notably, while an essential facility was involved in Deutsche Telekom I, the owner of the facility had a regulatory obligation to share and rivals’ margins were negative.37F[37] Teliasonera found a margin squeeze in a situation where the input of the dominant undertaking was not indispensable, there was no regulatory obligation to supply, and rival firms’ margins were positive, but insufficient, as the rivals were forced to operate at artificially reduced levels of profitability.38F[38] Telefonica39F[39] and Slovak Telekom40F[40] upheld the approach of considering margin squeezes as an independent form of abuse to which the criteria established in Bronner are not applicable.

B.   European Case Law on Self-Preferencing

Against this background, doubts about the potential to identify self-preferencing as a standalone abuse under EU law emerge from the court analysis and antitrust decisions that have been issued to date sanctioning dominant platforms for preferential treatment granted of own products and services. Indeed, recent European case law would appear to question whether self-preferencing is sufficiently novel to constitute a standalone theory of harm, given that it has been readily addressed under existing theories of harm. With the exceptions of the Amsterdam Court of Appeal41F[41] and the Italian Competition Authority,42F[42] courts generally do not even use the term “self-preferencing,” opting instead to label the conduct “favoring.”

1.  UK Streetmap decision and European Commission Google Shopping decision

The birth of self-preferencing as a standalone theory of harm is usually associated with the European Commission’s investigation of Google for having positioned and displayed, in its general search-results pages, its own comparison-shopping service more favorably than rival comparison-shopping services.43F[43]

However, a similar issue was addressed a few months earlier by the High Court of England and Wales in the dispute between Streetmap and Google, which involved the interaction of competition between online search engines and online maps.44F[44] Indeed, Streetmap contended that Google abused its dominant position by prominently and preferentially displaying its own related online-map product. Streetmap contended that, by visually displaying a clickable image from Google Maps (and no other map) in response to certain geographic queries (Maps OneBox) at or near the very top of its search-engine results page (SERP), and consequently relegating Streetmap to a blue link lower down the page, Google abused its dominant position in the market for online search and online search advertising.

Given the evident similarity with the Google Shopping case, Justice Roth’s analysis is worthy of examination. While Streetmap framed Google’s practice in terms of bundling or tying, and referred extensively to the European Microsoft decision, the U.K. Court held that the complaint should have been appropriately characterized as an allegation of discrimination.45F[45] The user who sees the Maps OneBox is, indeed, under no obligation to click on it or to use Google Maps; she remains free to use any other online-mapping provider without penalty. In contrast to Microsoft, where obtaining a competing streaming-media player by downloading from the Internet was regarded by a significant number of users as more complicated than using the pre-installed Microsoft product, the Google SERP includes clickable links to other relevant online maps and users experience no difficulty in clicking on those blue links.

To establish whether Google’s conduct constituted anticompetitive foreclosure, the Court concluded that it was necessary to prove that the effects of that conduct appreciably affected competition, which cannot simply be assumed. Indeed, the Google’s introduction a Maps OneBox containing a thumbnail map was intended to improve the quality of the SERP, and hence must be evaluated as a technical efficiency46F[46]: “The unusual and challenging feature of this case is that conduct which was pro-competitive in the market in which the undertaking is dominant is alleged to be abusive on the grounds of an alleged anti-competitive effect in a distinct market in which it is not dominant.”47F[47] For this reason, evaluating alternative ways that Google might have made this procompetitive improvement without allegedly distorting competition in online maps played a significant role in the Court’s analysis.

If anticompetitive effects are proven, then the issue of objective justification must be considered. This requires a proportionality assessment, which is a matter of fact and degree. Hence, the question of alternatives cannot be considered only with respect to competitive impact: “Where the efficiency is a technical improvement, proportionality does not require adoption of an alternative that is much less efficient in terms of greatly increased cost, or which imposes an unreasonable burden.”48F[48]

Following this line of reasoning, Justice Roth found that the introduction of the Maps OneBox with no shortcut hyperlinks to Streetmap (and other online maps) did not, in itself, have an appreciable effect in steering customers away from Streetmap; it therefore was not reasonably likely to give rise to anticompetitive foreclosure.49F[49] Moreover, even if it was likely to have such an effect, Google’s conduct was objectively justified because the way that it implemented the technical efficiency—i.e., presenting a thumbnail map on the SERP—was not disproportionate.50F[50]

The European Commission reached a different conclusion in Google Shopping. There, the Commission found that, by promoting its own comparison-shopping service in its search results and demoting those of competitors, Google engaged in a strategy of leveraging the dominance of its flagship product (i.e., the search engine) in the adjacent market for comparison-shopping services.

According to the Commission’s findings, Google’s strategy rested on two related practices: ensuring prominent placement for its own comparison-shopping service and demoting rival comparison-shopping services in its search results. Notably, while competing comparison-shopping services could appear only as generic search results—potentially subject to demotion in search listings by Google’s algorithms—Google’s own comparison-shopping service was prominently positioned, displayed in rich format, and free from the risk of demotion to the second page of search results.51F[51]

The Commission concluded that the conduct fell outside the scope of competition on the merits, could extend Google’s dominant position in the national markets for general search services to the national markets for comparison-shopping services (offensive leveraging), would tend to protect Google’s dominance in the former (defensive leveraging).

Rather than recognizing that it was deploying a novel theory of harm, the Commission argued that Google’s conduct belonged to the well-known category of leveraging. Accordingly, there was no need to look for a new legal test, since “it is not novel to find that conduct consisting in the use of a dominant position on one market to extend that dominant position to one or more adjacent markets can constitute an abuse.”52F[52] The Commission therefore found that self-preferencing constitutes a “well-established, independent, form of abuse.”53F[53]

To support its line of reasoning, the Commission invoked disparate case law, including judgments involving either specific theories of harm (e.g., Tetra Pack,54F[54] Irish Sugar,55F[55] and Microsoft,56F[56] with regards to tying and predatory pricing, loyalty rebates, and refusal to deal, respectively) or that are outdated (e.g., Telemarketing57F[57]).58F[58] The Commission’s rationale in offering this selection of decisions is unclear. References to one case in which the essential facilities doctrine was applied (i.e., Microsoft) and to another ruling that has since been replaced by the elaboration of the essential facilities doctrine (i.e., Telemarketing) are even more surprising.

The Commission ultimately dismissed Google’s claim that its conduct could be considered abusive only if Bronner’s criteria were fulfilled: namely, if access to Google’s general search results pages were indispensable to being able to compete.59F[59] According to the Commission, the decision merely required Google to cease the conduct. Hence, the Bronner criteria were “irrelevant in a situation, such as that of the present case, where bringing to an end the infringement does not involve imposing a duty on the dominant undertaking to transfer an asset or enter into agreements with persons with whom it has not chosen to contract.”60F[60]

The case spurred debate over the legal test applied to require Google to grant equal treatment to rival comparison-shopping services and its own service. Among the questions raised by the case are whether the conduct fell more within exclusionary or discriminatory abuses and, if it was the former, whether tying or the essential facilities doctrine was the proper framework to assess such self-preferencing abuse.61F[61] For instance, the experts appointed by the Commission to provide suggestions for the design of a competition policy for digital markets considered self-preferencing a specific technique of leveraging, which is not abusive per se, but subject to an effects test.62F[62] Furthermore, quoting Microsoft, they argued that, according to well-established case law, the owner of an essential facility must not engage in self-preferencing. Nonetheless, they believed that self-preferencing by a vertically integrated dominant digital platform can be abusive, not only under the preconditions set out by the essential facilities doctrine, but also wherever it is likely to result in leveraging market power and is not justified by a pro-competitive rationale.

2.   Amsterdam Court of Appeal ruling in Funda

In May 2020, the Amsterdam Court of Appeal handed down a decision in litigation between VBO Makelaars and NVM, two associations of real-estate agents, brokers, and appraisers.63F[63] NVM owns Funda, the largest online real-estate platform in the Netherlands and which, according to VBO, granted NVM agents more prominent positions in the ranking of properties. Funda also applied higher tariffs to and granted only limited website functionality to VBO agents, who also did not have access to the underlying Funda database. VBO’s complaint charged NVM with anticompetitive discrimination.

Upholding the decision of the district court, the Court of Appeal found that Funda did not abuse its dominant position in favoring the listings of NVM members over those of rival agents. Assessing self-preferencing as discriminatory conduct under Article 102(c) TFEU, the Court cited MEO, arguing that VBO’s complaint failed to demonstrate that the discrimination distorted the company’s competitive position in ways that led to a competitive disadvantage on the downstream market for real-estate services.

In particular, the Court noted that several factors play relevant roles in consumers’ home-purchase decisions and that it is implausible that a buyer would automatically assume that the listing placed highest on a website would be the one that best meets their demands. To the contrary, the Court concluded that the market for homes differs significantly from markets for other consumer products. For example, buyers generally conduct an intensive search over a long period of time to consider all relevant offers. Therefore, the Court found, a lower website ranking would be of minor importance and would not necessarily lead to a competitive disadvantage.

Accordingly, the Court considered comparisons with Google Shopping to be unhelpful.64F[64] In assessing NVM’s preferential treatment in accordance with the principles the CJEU elaborated in MEO, however, the Dutch court did frame self-preferencing as a discriminatory abuse, thus anticipating the approach that the European General Court would ultimately endorse in Google Shopping.

3.   French Competition Authority Apple ATT and Google AdTech decisions

In June 2021, the French Competition Authority (AdlC) followed the European Commission’s lead in investigating practices implemented by Google in the online-advertising sector.65F[65]

Responding to referrals from news publishers who monetize their websites and mobile apps through advertising, the AdlC found that Google engaged in abusive practices to favor its own advertising intermediation technologies, granting preferential treatment to its proprietary technologies offered under the Google Ad Manager brand. Notably, in the Authority’s view, Google used its dominant publisher ad server (DoubleClick for Publishers, or DFP) both to favor its own programmatic advertising sales platform (AdX) and, separately, used AdX to favor DFP in the market for supply-side ad-intermediation platforms (SSPs). Regarding the first practice, the preferential treatment consisted of informing AdX of the prices offered by competing SSPs, thus allowing it to optimize the bidding process by varying the commissions received on impressions sold according to the intensity of competition. Regarding the second practice, Google imposed technical and contractual limitations on the use of the AdX platform through a third-party ad server. As a result, the modalities of interaction offered to third-party ad-server clients were inferior to the modalities of interaction between DFP and AdX, which penalized both third-party SSPs and publisher clients.

Similar concerns about the impact of Google’s conduct in ad-tech services have also been raised by the Australian Competition and Consumer Commission (ACCC). The ACCC concluded that Google’s vertical integration and dominance across the ad-tech supply chain and related services have allowed the company to engage in leveraging and self-preferencing conduct and that this, in turn, has likely interfered with the competitive process.66F[66]

According to the AdlC, the evidence showed that DFP’s favorable treatment of AdX had a foreclosure effect on competition among platforms selling ad space, significantly reducing the attractiveness of rival SSPs. In addition, according to the Authority, DFP’s preferential treatment strengthened Google’s dominant position, impairing the competitiveness of rival ad-server providers, and limiting their ability to compete on the merits. Therefore, as regards the latter, limitations on interoperability were deemed a practice that cannot be considered competition on the merits, as it would tend to impose on rivals a competitive disadvantage by applying to them less favorable technical conditions.67F[67]

By and large, the French decision did not provide insights on the theory of harm or type of abuse that this form of discrimination would constitute. Like the European Commission, the AdlC did not refer to self-preferencing explicitly, instead describing Google’s conduct as favoring. With regards to Google’s leveraging strategy, the AdlC cited Google Shopping and quoted the very same case law the Commission mentioned in that decision.68F[68] The only significant addition made by the Authority was a reference to Slovak Telekom, a margin-squeeze case that, as already mentioned, brought about a remarkable change in confining the application of Bronner’s indispensability condition to “pure” refusals to deal.69F[69]

However, the AdlC is also currently investigating Apple’s privacy policy, where self-preferencing is mentioned explicitly and appears to be framed differently.70F[70] Notably, in March 2021, the French Authority rejected the request for interim measures against Apple’s adoption of the App Tracking Transparency (ATT) framework for applications on iOS 14, which create new consent and notification requirements for app publishers. The ATT solicitation was considered part of Apple’s longstanding strategy to protect iOS users’ privacy and its implementation was not expected to impose unfair trading conditions, such as excessive or disproportionate restrictions on the activities of app developers. The AdlC nonetheless advised that, as part of its investigation into the merits of the case, it would examine how the user-consent collection processes differ between Apple’s own advertising services and third-party advertising services. Differences in those processes might result in a “form of discrimination (or self-preferencing)” if Apple applied, without justification, more binding rules on third-party operators than those it applies to itself for similar operations.71F[71]

The growing suspicion of self-preferencing has likewise prompted the German Competition Authority to initiate its own proceeding on Apple’s ATT policy,72F[72] while the U.K. Competition and Markets Authority (CMA) raised similar concerns in its market study on mobile ecosystems.73F[73]

4.   General Court ruling in Google Shopping

Given this background, the European General Court’s judgment in Google Shopping was much awaited.74F[74] For those who were looking for legal certainty from the judgment, however, those expectations have been not completely met.

What was new in the ruling was its broad interpretation of the general principle of equal treatment, which the Court affirmed obligates vertically integrated platforms to refrain from favoring their own services over rivals.75F[75] While this approach was in line with the Commission’s expansive reading of the special responsibility of dominant firms, however, the ruling framed self-preferencing as a discriminatory abuse.76F[76] Notably, the Court highlighted that the various judgments the Commission cited in its original ruling do not support the conclusion that any use of a dominant position on one market to extend that position to one or more adjacent markets constitutes a “well-established” form of abuse.77F[77] After all, “leveraging” is a generic term covering several practices that are potentially abusive, such as tied sales, margin squeezes, and loyalty rebates.78F[78]

The three rulings the Court cited involve, instead, practices found to be discriminatory abuses specifically because they place third parties at a competitive disadvantage. Two of three involve discriminatory conditions applied by public undertakings operating a commercial port79F[79] and an airport.80F[80] This may support a link with recent legislative initiatives categorizing digital platforms as common carriers and thus subject to the neutrality regime of public utilities-style regulation. Nonetheless, the Court clarified that prohibiting self-preferencing to enforce the policy goal of neutrality is appropriate only when a competitive harm is demonstrated. Indeed, rather than deeming self-preferencing to be per se abusive, the Court moved to its potential anticompetitive effects. This is in line with the effects-based approach affirmed in MEO,81F[81] as well as in other judgments that, although they involve different abusive conduct, entail similar discriminatory elements.82F[82] The Court, however, surprisingly did not even mention MEO.

The Court’s ruling focused on potential exclusionary effects associated with specific leveraging strategies, reflected in the Commission’s original finding of abuse on the basis of certain relevant criteria.83F[83] The Commission had noted that, due to network effects, the traffic that Google’s search engine generates represents a critical asset; that users are significantly influenced by favoring, as they typically concentrate on the first few search results and tend to assume that the most visible results are the most relevant; and that traffic directed from Google’s search-results pages accounts for a large portion of traffic to competing comparison-shopping services, which cannot be effectively replaced by other sources.84F[84]

The Court outlined four criteria that differentiated Google’s self-preferencing from competition on the merits, therefore warranting a finding of antitrust liability.

First, the Court highlighted the “universal vocation” and openness of a search engine as features of its core mission.85F[85] These features distinguish a search engine, which designed to index results that might contain any possible content, from other services referenced in the case law, which consist of tangible or intangible assets (press-distribution systems or intellectual property rights, respectively) whose value depends on a proprietor’s ability to retain their exclusive use.86F[86]

While not explicitly mentioned, the reference is clearly to the essential facilities doctrine case law. Unlike these services, “the rationale and value of a general search engine lie in its capacity to be open” to results from external sources and to display multiple and diverse sources on its general results pages.87F[87] Moreover, the legal obligation of equal treatment that ensues from net-neutrality regulations88F[88] for Internet access providers on the upstream market cannot be disregarded when analyzing the practices of an operator like Google on the downstream market.89F[89]

Second, because Google holds a “superdominant” (or “ultra-dominant”) position on the market for general search services and acts as a “gateway” to the Internet, it is under a stronger obligation not to allow its behavior to impair genuine, undistorted competition on the related market for specialized comparison-shopping search services.90F[90]

Third, the market for general search services is characterized by very high barriers to entry.91F[91]

Fourth, in light of prior considerations (i.e., the mission of a search engine, Google’s dominance, and the presence of very high barriers to entry), the Court found that Google’s conduct is “abnormal.”92F[92] Indeed, for a search engine to limit the scope of its results to its own services entails an element of risk and is “not necessarily rational.” This is especially the case in a situation where, because of barriers to entry and the search engine’s own dominance, it is significantly unlikely that there would be market entry within a sufficiently short period of time in response to the limitations placed on Internet users’ choices.93F[93]

In this scenario, in the Court’s view, Google’s promotion of its own specialized results over third-party results contradicts the basic economic model of a search engine and hence involves a certain form of abnormality.94F[94] The suspicion is strengthened by Google’s “change of conduct.”95F[95] While it initially provided general search services and displayed all the results of specialized search services in the same way and according to the same criteria, once the firm had entered the market for specialized comparison-shopping search services—and after having experienced the failure of its dedicated comparison-shopping website (Froogle)—Google changed its practices and comparison-shopping services were no longer all treated equally.96F[96]

These four criteria suggest that the Court saw Google’s search engine as an essential facility. The Court, indeed, noted that, by envisaging equal treatment for any comparison-shopping services on Google’s general results pages, the Commission’s decision was seeking to provide competitors with access to Google’s general results pages. This was presented as particularly important to competing comparison-shopping services and something that was not effectively replaceable, as it accounted for such a large proportion of traffic to their websites.97F[97] Moreover, the Court acknowledged that the Commission considered Google’s traffic to be indispensable to competing comparison-shopping services.98F[98] As a consequence, the analysis would have required an assessment of preferential treatment pursuant to the conditions set out in Bronne, as Google itself had requested, rather than relying on the case law applicable to abusive leveraging, as the Commission did in its decision.

But despite characterized the features of Google’s general results page as “akin to those of an essential facility,”99F[99] the Court upheld the Commission’s decision not to apply Bronner’s indispensability requirement. In doing so, it drew a line between express refusals to supply and exclusionary practices that do not lie “principally” in a refusal, as such.100F[100] Indeed, “the present case is not concerned merely with a unilateral refusal by Google to supply a service to competing undertakings that is necessary in order to compete on a neighboring market, which would be contrary to Article 102 TFEU and would therefore justify the application of the ‘essential facilities’ doctrine.”101F[101]

Therefore, the Court shared the Commission’s viewpoint that Google’s self-preferencing was a standalone form of leveraging abuse, involving positive acts of discrimination in the treatment of the search results for comparison-shopping services.102F[102]

The Court’s ruling has been generally welcomed for two reasons. By affirming self-preferencing as an independent abuse, the judgment provides legal support to the policy goal of imposing a neutrality regime over large digital platforms, which has informed all the regulatory interventions promoted in different jurisdictions. At the same time, the Court advances a clearer legal qualification of the conduct in question. Indeed, while the Commission’s approach appeared unprecedented—because it revolved around the notion of favoring as a specific form of leveraging—the Court opted for the more defined legal framework of discrimination. The outcome should help to restrain the scope of application for self-preferencing prohibitions in comparison to other traditional practices that, although belonging to the general category of leveraging and including elements of discrimination, reflect specific theories of harm and are assessed according to their respective legal tests.103F[103] By and large, the Court confirms that there is no well-established case law that would forbid any extension of a dominant position in adjacent markets, in contrast with the Commission’s stance.

Nonetheless, the ruling raises new doubts. Notably, the definition of the conduct that would be covered remains unclear. While adopting the general principle of equal treatment as a legal basis to prohibit self-preferencing may allow intrusions into platforms’ design choices,104F[104] the listed criteria appear to define a narrow framework, ultimately calling into question the broad application of self-preferencing as a standalone abuse.

The Court underscored the relevance of the “particular context” in which favoring occurred.105F[105] Namely, the emphasis was on the role played by search engines on the Internet, including their “universal vocation” and “open” business models. This is strengthened by analogies to net neutrality, the characteristics “akin to those of an essential facility,” and the “superdominant position” that made Google a “gateway” to the Internet. Furthermore, the peculiar features of search engines (notably, their openness) are also deemed relevant in assessing the “abnormality” of Google’s behavior—which, in the Court’s evaluation, is indeed at odds with the basic economic model of its search engine. The legal framework is completed with the detection of opportunistic behavior by Google, which changed its strategy once it entered the adjacent market of comparison-shopping services.

It is left far from clear whether Google Shopping is even sanctioning the favoring practice as such. Indeed, the Court describes the anticompetitive strategy in question as formed by a combination of two practices—namely the promotion of Google’s own services and the demotion of its rivals’ services. Therefore, the conduct is not necessarily abusive even if it consists “solely in the special display and positioning” of the platform’s own products and services.106F[106] The practice has instead been judged illegal because it included the relegation of competing services in Google’s general results pages by means of adjustment algorithms. That, in conjunction with Google’s promotion of its own results, there was a simultaneous demotion of results from competing comparison services is considered a “constituent element” of the conduct and moreover plays a “major role” in the exclusionary effect identified.107F[107]

In summary, rather than articulating a legal test for a new antitrust offense, the criteria pointed out in the judgment for considering the preferential treatment abusive appear extremely fact-sensitive: both Google-specific and search engine-specific. Therefore, it is difficult to see how, according to these criteria, a self-preferencing prohibition may be applied to different forms of preferential treatment, digital services, and business models.108F[108]

5.   Italian Competition Authority Amazon Logistics decision

A few weeks after the General Court’s ruling, the Italian Competition Authority (AGCM) handed down a decision that significantly departed from the legal framework elaborated in Google Shopping, thus confirming that the precise contours of self-preferencing abuses under Article 102 TFEU remained anything but clear.109F[109]

In late November 2021, the AGCM issued a mammoth fine against Amazon for granting preferential treatment to third-party sellers who use the company’s own logistics and delivery services (i.e., Fulfilment by Amazon, or FBA). Amazon was accused of having leveraged its dominance in the market for intermediation services on marketplaces to favor the adoption of its own FBA by sellers active on Amazon.it, as well as to strengthen its own dominant position. Under AGCM’s view, this strategy ultimately harmed both competing logistics operators, by putting them at a competitive disadvantage, and competing marketplaces, by creating incentives for sellers to single-home.

Indeed, although third-party sellers are free to manage the logistics associated with their operations on the platform themselves or outsource them to an independent operator (Merchant Fulfilment Network, or MFN), Amazon was deemed to be artificially pushing them to use its own logistics service, thus deterring them from multi-homing.110F[110] Notably, the Authority found that Amazon “tied” the use of FBA to access to a set of exclusive benefits essential for gaining visibility and increasing sales on the marketplace.111F[111]

Among those benefits, the most relevant is the Prime label, which allows sellers to participate in special events promoted by Amazon (e.g., Black Friday, Cyber Monday, Prime Day) and benefit from fast and free shipping. Furthermore, Prime increases the likelihood of sellers’ offers being selected as featured offers displayed in the Buy Box. This is of the utmost importance to sellers, as the Buy Box prominently displays just a single seller’s offer for a given product on Amazon’s marketplaces and generates the vast majority of all sales for that product.

Quoting from several of Amazon CEO Jeff Bezos’ letters to shareholders, the AGCM noted the company believes FBA is the “glue” that links Marketplace and Prime112F[112]: “Thanks to FBA, Marketplace and Prime are no longer two things … Their economics and customer experience are now happily and deeply intertwined.”113F[113] Furthermore, FBA is a “game changer” for sellers because it makes their items eligible for Prime benefits, which drives their sales.114F[114] Pursuant to its leveraging strategy, Amazon prevented third-party sellers from associating the Prime label with offers not managed by FBA. In addition, the AGCM noted that third-party sellers using FBA are not subject to the performance-evaluation metrics that Amazon applies in monitoring non-FBA sellers’ performance. Such metrics can ultimately lead to the suspension of non-compliant sellers’ accounts on Amazon.it. All these benefits derived from the use of FBA were considered, to various extents, to be “crucial” to success on the marketplace.115F[115]

It is worth noting that the European Commission has also launched an investigation of Amazon for facts identical to those already addressed in the Italian inquiry, with the relevant market defined as the European Economic Area, except for Italy.116F[116]

The alleged unequal treatment of non-FBA sellers has also been investigated by the Austrian Federal Competition Authority (BWB).117F[117] Although concerned about potential discrimination against sellers who organize their deliveries independently, the Authority conceded that  a better ranking could have also resulted from the better service offered under FBA, compared with the independent organization of deliveries. Hence, the BWB remained open to the possibility that the appearance of preferential treatment for FBA Marketplace sellers was objectively justified.118F[118] The Austrian Authority concluded that a comprehensive and transparent legal framework was the best way to counter problematic business practices and accepted Amazon’s modifications to its terms and conditions.119F[119]

The link between Amazon Marketplace and FBA was also scrutinized as part of the investigation conducted by the U.S. House Judiciary Committee’s Antitrust Subcommittee into the state of competition in digital markets. The subcommittee’s final report found that many third-party sellers have no choice but to purchase fulfillment services from Amazon to maintain a favorable search-result position.120F[120] The report characterized Amazon’s strategy as tying.121F[121]

For the sake of our analysis, the Italian Amazon decision is especially remarkable because of how it contrasts with Google Shopping. As already mentioned, with the exception of the Amsterdam Court of Appeal, it represents the only decision in which the term self-preferencing is used.122F[122] Self-preferencing is here defined as an unequal and unjustified preferential treatment granted by a dominant player to its own services in pursuing a leveraging strategy, hence falling outside the scope of competition on the merits.123F[123] Therefore, rather than reflecting the criteria set out by the General Court, the Italian decision is clearly inspired by the Commission’s approach in Google Shopping. Indeed, in line with the idea of describing self-preferencing as a new form of leveraging abuse, Amazon’s practice is characterized as a form of tying.124F[124]

This definition of self-preferencing is convenient for enforcers, in that it would allow them to bypass the legal standards otherwise required to prove unlawful tying. Indeed, tying requires a form of coercion, such that customers do not have the choice to obtain the tying product without the tied product. In the Amazon case, by contrast, there is neither a contractual obligation nor technical integration between marketplace services and logistics services. Business users are free to run the logistics by themselves or to outsource it to an independent operator without losing the ability to operate on the Amazon e-commerce platform.

Apart from the legal qualification of the conduct in question (which may be more properly characterized as bundling), finding an abuse in a tying case also requires proof of potential foreclosure effects against equally efficient rivals. Looking at the effects on logistics operators, according to the AGCM’s view, vertical integration between the marketplace and logistics constitutes Amazon’s main competitive advantage, which is unmatchable even by equally efficient rivals.125F[125] Indeed, FBA is an integrated logistics service designed to represent “a one-stop shop solution” for storage, shipping, and customer service within a “closed and complete ecosystem” in which Amazon plays multiple roles.126F[126] While Amazon recently started offering a multi-channel logistics service, few retailers have adopted it due to its high operating costs.127F[127] Moreover, part of the AGCM’s decision concerning complaints raised by Amazon’s major e-commerce rival eBay—which reported that a large portion of its market share had been absorbed by Amazon—ultimately recognized that Amazon’s superiority stemmed from its popularity with users and retailers, especially in the critical areas of trust, shipping, and returns.128F[128]

In short, the thing that has made the playing field uneven has been Amazon’s creation of a successful ecosystem, which provides the company with competitive advantages that cannot be replicated either by pure online marketplaces or pure logistics providers.129F[129] A prohibition on self-preferencing may therefore functionally reflect a bias against ecosystems, which require massive and uncertain investment to create, and which provide significant benefits to both business users and final customers.

In summary, the AGCM endorsed an expansive view of the scope of anticompetitive self-preferencing that was at odds with the legal qualifications and narrow criteria set out by the General Court in Google Shopping, and that lacked the context the General Court had laid out to assess the circumstances in which preferential treatment constitutes discriminatory abuse. Notably, an online marketplace does not share many relevant features with an Internet search engine. Indeed, Amazon’s business model is “a closed and complete ecosystem.”130F[130] Moreover, unlike in the Google Shopping case, Amazon is not accused of changing its conduct in response to its market position. The only elements of the criteria defined in Google Shopping that Amazon could be argued to meet are operating in a market with high barriers to entry and being a vertically integrated firm with a super/hyper dominance in the upstream market.131F[131]

Although the General Court’s ruling is mentioned a few times,132F[132] these appear to be last-minute references included merely to note that the Commission’s decision had been upheld by the General Court.133F[133] Since the Italian Amazon decision was delivered just a few weeks after the Google Shopping ruling, it is possible that the AGCM simply did not have time to adjust its line of reasoning to comport with the Court’s qualifications and criteria.

C.   Preferential access to non-public data

A broad interpretation of self-preferencing could find that it covers the preferential provision of data and information, which could similarly be prohibited as abusive. Following this line of reasoning, the European Commission sent a statement of objections to Amazon in November 2020 informing the company of the Commission’s preliminary view that its practice of systematically using non-public business data from independent retailers who sell on its online marketplace infringes antitrust rules, on grounds that Amazon uses that data to benefit its own retail business that directly competes with those retailers.134F[134]

More recently, the UK CMA has also opened an investigation into how Amazon collects and uses third-party seller data, including whether Amazon gains an unfair advantage in business decisions made by its retail arm.135F[135] Similar concerns were raised by staff to the U.S. House Judiciary Antitrust Subcommittee, whose final report argued that Amazon’s asymmetric access to and use of third-party seller data constitutes unfair treatment of those third-party sellers.136F[136] The ACCC likewise warned that Apple and Google both have the ability and incentive to use their positions as app-marketplace operators to monitor downstream competitors.137F[137] For instance, the ACCC found that Apple’s Developer Agreement allows the company to develop, acquire, license, market, promote, and distribute products and software that perform functions the same or similar to any of the products, software or technologies provided by app developers that use the App Store. By contrast, Apple requires that app developers follow obligations to avoid being copycats.

Similar allegations of unfair use of user data have been levied against Facebook by the European Commission and the U.K. CMA, which charge that the company uses data gathered from advertisers in order to compete with them in markets in which Facebook is active, such as classified ads.138F[138] Finally, one focus of the European Commission’s investigation of Apple’s App Store rule requiring developers to use Apple’s in-app purchase mechanism for the distribution of paid apps and/or paid digital content is the potential that competing developers may be disintermediated from important customer data, while Apple can obtain valuable data about the activities and offers of its competitors.139F[139]

These investigations have inspired the bans on so-called “sherlocking” (i.e., the use of business users’ data to compete against them) included both in the DMA140F[140] and the proposed American Innovation and Competition Online Act,141F[141] as well as calls for structural separation and line-of-business restrictions.142F[142]

Amazon has faced accusations that it takes advantage of its dual role and hybrid business model in serving both as a marketplace-service provider and a retailer on the same marketplace, in competition with independent sellers. The charge is that the company can leverage its access to non-public third-party sellers’ data—such as the number of units ordered and shipped, the revenues sellers earn on the marketplace, the number of visits to sellers’ offers, data related to shipping and to sellers’ past performance, and consumer product claims—to identify and replicate popular and profitable products from among the hundreds of millions of listings on its marketplace.

Notably, according to the European Commission’s preliminary findings, such granular and real-time business data feed into the algorithms of Amazon’s retail business, allowing them to calibrate retail offers and strategic business decisions to the detriment of the other marketplace sellers. Thus, it is argued that the appropriation and the use of third-party sellers’ data enables Amazon to avoid the normal risks of retail competition, such as those associated with investing in a new product or choosing a specific price level, and to leverage its dominance in the market for the provision of marketplace services in France and Germany (i.e., the biggest markets for Amazon in the EU).143F[143]

The U.S. House Antitrust Subcommittee similarly charged that Amazon is able to use marketplace data from third-party merchants to create competing private-label products or to source products directly from manufacturers in order to free ride off sellers’ efforts.144F[144] The impact assessment study supporting the DMA confirmed this suspicion, noting that the launch of Amazon Basics (i.e., the most successful private label brand on Amazon’s marketplace) has negatively impacted the sales on Amazon of third-party products in identified attractive product segments.145F[145]

Leveraging this information exclusively, without sharing it with third-party sellers, is considered a form of self-preferencing because Amazon is in position to use data from its marketplace to gain a competitive advantage in market research and to identify new business opportunities without incurring any financial risk.146F[146] Furthermore, by using information from its Amazon fulfilment program, Amazon can also determine where products offered by third-party merchants are being manufactured and by whom. Since Amazon Basics products are sold in large volumes, Amazon can approach the manufacturers of goods for third-party merchants, buy these items in larger quantities, and sell them for a lower price than the competition on its own platform.147F[147]

This line of reasoning aligns with the core concerns about self-preferencing, such as conflicts of interest and the competitive advantages that a platform’s dual role may yield. But to invoke antitrust in cases where a platform performing a dual role gains a competitive advantage would require demonstrating proof of competitive harm, which isn’t apparent in this case. Indeed, while the impact on innovation appears uncertain,148F[148] Amazon’s practice likely benefits consumers by permitting close price comparisons, increasing output, and forcing sellers to reduce their prices.149F[149] Such effects are even more relevant when it comes to sellers with market power, as the introduction of products and services in competition with third parties would reduce double marginalization.

Moreover, the relevance of non-public third-party merchants’ data in facilitating copying by Amazon is questionable. Indeed, as noted, Amazon’s public product reviews supply a great deal of information and any competitor can obtain an item for the purposes of reverse engineering.150F[150] Conversely, if the products in question are protected by intellectual-property rights, Amazon could be found guilty of infringement. Finally, it is unclear whether and how this form of self-preferencing would meet the legal qualification and criteria set out by the General Court in Google Shopping.

Nonetheless, the European Commission is currently evaluating the commitments offered by Amazon, which has proposed to refrain from using non-public data relating to, or derived from, the activities of independent sellers on its marketplace for its retail businesses that compete with those sellers.151F[151] The relevant data would cover both individual and aggregate data (e.g., sales terms, revenues, shipments, inventory-related information, consumer-visit data, or seller performance on the platform). Amazon commits not to use such data for the purposes of selling branded goods, as well as in its private-label products.

III. Imposing platform neutrality under antitrust law

Because preferential treatment may result from a wide range of practices, self-preferencing potentially covers different types of behavior that are subject to different legal standards and that may include exploitative elements.152F[152] Prohibitions on self-preferencing as per se anticompetitive would therefore grant antitrust enforcers significant leeway to bypass the legal standards ordinarily required to prove traditional anticompetitive harms. As a result, such prohibitions would provide antitrust authorities with a powerful tool to intervene in digital markets. This issue is particularly sensitive in Europe where the DMA entrusts the European Commission with the sole power to apply the new regulation but does not displace national competition law. Hence, national competition authorities will remain in charge of the enforcement of national and European antitrust rules.

Besides its potential as an enforcement shortcut, self-preferencing prohibitions may function to impose a neutrality regime on online gatekeepers. The aim would be to ensure a level playing field that currently appears uneven because of the bottleneck and intermediation power exerted by large online platforms. Such rules also could neutralize conflicts of interests raised by platforms’ dual-mode intermediation. The dual roles that some platforms perform fuel the never-ending debate over vertical integration and the related concern that, by giving preferential treatment to its own products and services, an integrated provider may leverage its dominance from one market to related markets. Indeed, the circumstances that may give rise to conflicts of interests and the circumstances that can give rise to leveraging strategies can be similar.153F[153] From this perspective, self-preferencing is a byproduct of the emergence of ecosystems. By integrating complementary products and services, a platform that controls and operates at all levels of the value chain may have an incentive to favor its own offers.154F[154]

But as antitrust authorities generally recognize, self-preferencing conduct is “often benign.”155F[155] Furthermore, since the value that the ecosystem generates depends on the activities of independent complementors, that value is not completely under the control of the platform owner.156F[156] Firms operating on the platform and competing with the platform owner may be disadvantaged by a variety of legal, technological, and informational measures implicated by self-preferencing, but there also may be legitimate justifications for such conduct that would need to be carefully considered in each instance.157F[157] Platforms implement different business models and are driven by different incentives, which in turn affects their strategies.

Against this backdrop, an outright ban on self-preferencing could undermine the very existence of ecosystems by challenging their design and monetization strategies.158F[158] Given that preferential treatment can take many different forms and have very different effects, the different business models adopted by platforms should be subject to case-by-case and effects-based assessment.159F[159] This is also consistent with the industrial-organization literature, which has found mixed evidence on the impact of duality on welfare, thereby supporting the insight that absolute neutrality is not desirable and interventions should be product- and platform-specific.160F[160]

Finally, and more importantly, antitrust law does not impose a general duty to ensure a level playing field by sharing competitive advantages with rivals. Indeed, a competitive advantage cannot be automatically equated with anticompetitive effects.161F[161] Within this framework, the relevance of the general principle of equal treatment that has been invoked by the General Court in Google Shopping to frame self-preferencing as a discriminatory abuse should be regarded with significant skepticism.

This is even more evident in the aftermath of the recent CJEU ruling in Servizio Elettrico Nazionale, which confirmed that the effects-based approach to the assessment of abusive practices remains core to European competition law.162F[162] Notably, the CJEU definitively stated that competition law is not intended to protect the competitive structure of the market, but rather to protect consumer welfare, which represents the goal of antitrust intervention.163F[163]

Accordingly, as illustrated in Intel, not every exclusionary effect is necessarily detrimental to competition.164F[164] Competition on the merits may, by definition, lead some competitors— those that are less efficient and thus less attractive to consumers from the standpoint of price, choice, quality or innovation—to become marginalized or to depart from the market.165F[165] In particular, given that exclusionary effects do not necessarily undermine competition, a distinction must be drawn between a risk of foreclosure and a risk of anticompetitive foreclosure, since only the latter may be penalized under Article 102 TFEU.166F[166] If any conduct having an exclusionary effect were automatically classified as anticompetitive, such rules would become a means to protect less capable, less efficient undertakings and would in no way protect the more meritorious undertakings that stimulate a market’s competitiveness.167F[167]

By and large, these well-settled principles do not support the claim that antitrust rules are designed to ensure platform neutrality. As acknowledged by the General Court in Google Shopping, self-preferencing cannot be considered prima facie unlawful and therefore outside the scope of competition on the merits. Its assessment instead requires the demonstration of anticompetitive effects, taking account of the circumstances of the case and the relevant legal and economic context.168F[168] Toward this aim, a dominant platform remains free to demonstrate that its practice, albeit producing an exclusionary effect, is objectively justified on the basis of all the circumstances of the case, or that the effects are counterbalanced or outweighed by efficiency advantages that also benefit consumers, such as through lower prices, better quality, or a wider choice of new or improved goods and services.169F[169]

In order to assess the anticompetitive nature of a practice, it is necessary to examine whether the means used come within the scope of normal competition.170F[170] Anticompetitive effects do not amount to a mere competitive disadvantage, but require an impact on efficient firms’ ability and incentive to compete.171F[171] Servizio Elettrico Nazionale also clarified the meaning of competition on the merits, considering outside its scope conduct that is not based on obvious economic reasons or objective reasons.172F[172] It is therefore necessary to assess the ability of equally efficient competitors to imitate the conduct of the dominant undertaking. Exclusionary conduct by a dominant firm that can be replicated by equally efficient competitors does not represent the sort of conduct that would lead to anticompetitive foreclosure; it therefore comes within the scope of competition on the merits.173F[173] In order to assess whether a given practice comes within the scope of competition on the merits, the test of whether it would be impossible for an equally efficient rival to imitate the dominant firm’s conduct arises from the case law on both price-related (e.g., TeliaSonera and Post Danmark II) and non-price-related conduct (e.g., Bronner).174F[174]

Moving away from the goal of ensuring a level playing field, recent European case law on self-preferencing centers instead on the competitive advantages that platforms enjoy due to their dual role. A competitive advantage, however, need not amount to anticompetitive foreclosure. Foreclosure not only needs to be proved, but also assessed against potential advantages for consumers, in terms of price, quality, and choice of new goods and services. It is even less clear how NCAs’ expansive approach toward self-preferencing as a standalone abuse fit within this legal framework. Both the AdlC’s decision in Google AdTech and the AGCM’s decision in Amazon Logistics appear inconsistent both with the legal qualification and criteria defined by the General Court, and with the CJEU principles recalled in Servizio Elettrico Nazionale. Similar doubts are raised by the investigations into the preferential access to and use of non-public business data. Moreover, in these cases, the benefits for consumers appear particularly significant as, for instance in Amazon Marketplace, the conduct under investigation led to an immediate output increase and price reduction: in short, more competition.

IV. Conclusion

In her opinion Post Danmark II opinion, Advocate General Juliane Kokott warned that, in enforcing antitrust rules, the CJEU “should not allow itself to be influenced so much by current thinking (‘Zeitgeist’) or ephemeral trends, but should have regard rather to the legal foundations on which the prohibition of abuse of a dominant position rests in EU law.”175F[175] Accordingly, this paper has addressed the prevailing zeitgeist in digital markets, analyzing the markets’ proclaimed peculiar features and the potential scope of application to evaluate whether it should be considered a novel standalone antitrust prohibition.

Indeed, common-carrier antitrust is on the rise. Following the 2017 decision in Google Shopping, the European Commission and some NCAs have advanced a new theory of harm pointing to the competitive disadvantage suffered by rivals. This therefore constitutes a de facto ban on any preferential treatment granted by dominant platforms to their own products and services. Such a strong stance in antitrust enforcement relies on the premise that the special responsibility that an incumbent dominant player bears implies that they must ensure a level playing field.

It remains the case, however, that European case law questions both the goal of relying on antitrust rules to impose a neutrality regime on dominant platforms and the very existence of self-preferencing as an autonomous abuse. Competition law does not impose a general duty to share competition advantages with rivals and does not protect the structure of the market; hence, not every exclusionary effect automatically undermines competition. Self-preferencing is not, in itself, unlawful and platform neutrality as such is outside the scope of competition law.

In contrast with the European Commission and some NCAs, the European General Court in Google Shopping not only framed self-preferencing as a discriminatory abuse but also highlighted some criteria to assess its potential exclusionary effects and considered it outside the scope of competition on the merits. Such criteria are particularly fact-sensitive, and therefore at odds with its wide application as a standalone abuse.

In summary, against the sirens of a fascinating, popular, and convenient new label, the limiting principles of competition law remind us that it cannot be weaponized to ensure a specific market outcome. Therefore, in the aftermath of the Google Shopping ruling, doubts about the characteristics and boundaries of self-preferencing remain on the table, and we still do not have a legal test that distinguishes such purported new antitrust offenses from other practices aimed at pursuing leveraging strategies and already addressed by antitrust rules.

[1] Jacques Cre?mer, Yves-Alexandre de Montjoye, and Heike Schweitzer, Competition Policy for the Digital Era, European Commission (2019) 7, available at https://ec.europa.eu/competition/publications/reports/kd0419345enn.pdf; Unlocking Digital Competition, UK Digital Competition Expert Panel (2019) 58, available at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/785547/unlocking_digital_competition_furman_review_web.pdf.

[2] Elizabeth Warren, Here’s How We Can Break up Big Tech, Medium (2019) available at https://medium.com/@teamwarren/heres-how-we-can-break-up-big-tech-9ad9e0da324c.

[3] Margrethe Vestager, Statement Before the U.S. House of Representatives, Subcommittee on Antitrust, Commercial, and Administrative Law (2020) 2, available at https://www.euractiv.com/wp-content/uploads/sites/2/2020/07/Statement-EVP-Vestager-House-SubCommittee-30-July.pdf. See also id., Technology with Purpose (2020), https://ec.europa.eu/commission/commissioners/2019-2024/vestager/announcements/technology-purpose_en.

[4] Thomas Eisenmann, Geoffrey Parker, and Marshall Van Alstyne, Platform Envelopment, 32 Strateg. Manag. J. 1270 (2011).

[5] Regulation (EU) on Contestable and Fair Markets in the Digital Sector (Digital Markets Act), Article 6(1), (3), (5), (7), and (12), https://www.europarl.europa.eu/doceo/document/TA-9-2022-0270_EN.html.

[6] Impact Assessment – A New Pro-Competition Regime for Digital Markets, U.K. Government (2021) para. 21, https://www.gov.uk/government/consultations/a-new-pro-competition-regime-for-digital-markets.

[7] GWB Digitalization Act (Jan. 18, 2021), Section 19a, https://www.bundesrat.de/SharedDocs/beratungsvorgaenge/2021/0001-0100/0038-21.html.

[8] S.2992 – American Innovation and Choice Online Act, 117th Congress (2021-2022) Section 3(a)(1), available at https://www.congress.gov/117/bills/s2992/BILLS-117s2992rs.pdf.

[9] Richard J. Gilbert, The American Innovation and Choice Online Act: Lessons from the 1950 Celler-Kevaufer Amendment, Concurrentialiste (2022), https://leconcurrentialiste.com/gilbert-innovation-choice-act/?mc_cid=8bdf17d95a&mc_eid=34922555f0; Randal Picker, The House’s Recent Spate of Antitrust Bills Would Change Big Tech as We Know It, Promarket (2021), https://promarket.org/2021/06/29/house-antitrust-bills-big-tech-apple-preinstallation.

[10] Investigation of Competition in Digital Markets’, Majority Staff Reports and Recommendations, U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law (2020), 380, https://judiciary.house.gov/uploadedfiles/competition_in_digital_markets.pdf?utm_campaign=4493-519. See also Elettra Bietti, Self-Regulating Platforms and Antitrust Justice, Tex. Law Rev. (forthcoming); Nikolas Guggenberg, Essential Platforms, 24 STLR 237 (2021); Rory Van Loo, In Defense of Breakups: Administering a “Radical” Remedy, 105 Cornell L. Rev. 1955 (2020); Lina M. Khan, The Separation of Platforms and Commerce, 119 Columbia Law Rev. 973 (2019); K. Sabeel Rahman, Regulating Informational Infrastructure Internet Platforms as the New Public Utilities, 2 GLTR 234 (2018).

[11] Case AT.39740, Google Search (Shopping), European Commission (Jun. 27, 2017).

[12] Case T-612/17, Google LLC and Alphabet Inc. v. European Commission, European General Court (Nov. 10, 2021), EU:T:2021:763.

[13] See, e.g., Decision No. 29925, FBA Amazon, Autorità Garante della Concorrenza e del Mercato (Nov. 30, 2021), https://www.agcm.it/dettaglio?db=41256297003874BD&uid=801201274D8FDD40C12587AA0056B614&view=&title=A528-FBA%20AMAZON&fs=Abuso%20di%20posizione%20dominante. Previously, see Decision 21-D-11, Google, Autorité de la Concurrence (Jun. 7 2021) , https://www.autoritedelaconcurrence.fr/en/decision/regarding-practices-implemented-online-advertising-sector.

[14] Giuseppe Colangelo, The Digital Markets Act and EU Antitrust Enforcement: Double & Triple jeopardy, ICLE White Paper (2022), https://laweconcenter.org/resource/the-digital-markets-act-and-eu-antitrust-enforcement-double-triple-jeopardy.

[15] Pablo Iba?n?ez Colomo, Self-Preferencing: Yet Another Epithet in Need of Limiting Principles, 43 World Competition 417 (2020).

[16] Nicolas Petit, A Theory of Antitrust Limits, 28 Geo. Mason L. Rev. 1399 (2021).

[17] Herbert Hovenkamp, Selling Antitrust, Hastings L.J. (forthcoming).

[18] See Opinion of Advocate General Wahl, 20 December 2017, Case C?525/16, MEO — Serviços de Comunicações e Multimédia SA v. Autoridade da Concorrência, EU:C:2017:1020, paras. 76-77, arguing that a distinction must be drawn between undertakings that are vertically integrated (and have an interest in displacing competitors on the downstream market) and those that have no such interest. In the case of vertically integrated undertakings, the application by a dominant undertaking of discriminatory prices on the downstream or upstream market is, in reality, similar to first-degree price discrimination, which indirectly affects the undertaking’s competitors. See also Inge Graef, Differentiated Treatment in Platform-to-Business Relations: EU Competition Law and Economic Dependence, 38 YEL 448, 452-453 (2019), distinguishing among pure self-preferencing (whereby a vertically integrated platform treats its affiliated services more favorably than non-affiliated services), pure secondary line differentiation (whereby a non-vertically integrated platform engages in differentiated treatment among unaffiliated services in a market in which it is not active itself), and a hybrid category in which either a vertically integrated or a non-vertically integrated platform engages in differentiated treatment among unaffiliated services in an effort to favor its own business.

[19] Nicolas Petit, Theories of Self-Preferencing Under Article 102 TFEU: A Reply to Bo Vesterdorf, 1(3) CLPD 4 (2015).

[20] Case C-413/14 P, Intel v. Commission, Court of Justice of the European Union (Sept. 6, 2017), EU:C:2017:632.

[21] Id., C?525/16, MEO v. Autoridade da Concorrência (Apr. 19, 2018),  EU:C:2018:270. See also Wahl, supra note 18, para. 61.

[22] Guidance on the Commission’s Enforcement Priorities in Applying Article 82 of the EC Treaty to Abusive Exclusionary Conduct by Dominant Undertakings, European Commission (2009) OJ C 45/7, paras. 47-62.

[23] See Autorité de la Concurrence, supra note 13, para. 410, binding limits on interoperability with third-party services servers cannot be considered competition on the merits. See also Herbert Hovenkamp, Antitrust and the Design of Production, 103 Cornell L. Rev. 1155 (2018); Pablo Iba?n?ez Colomo, Product Design and Business Models in EU Antitrust Law, SSRN (2021), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3925396.

[24] Case T-201/04, Microsoft v. Commission, European General Court (Sept. 17, 2007), para. 1036, EU:T:2007:289.

[25] Case AT.40099, Google Android, European Commission, (Jul. 18, 2018), confirmed by Case T-604/18, Google v. Commission, European General Court (Sept. 14, 2022) EU:T:2022:541.

[26] Dirk Auer, Appropriability and the European Commission’s Android Investigation, 23 CJEL 647 (2017).

[27] Christopher S. Yoo, Open Source, Modular Platforms, and the Challenge of Fragmentation, 1 Criterion Journal on Innovation 619 (2016).

[28] UK Government, supra note 6, para. 21.

[29] Joined Cases C-241/91 P and 242/91 P, RTE and ITP v. Commission, Court of Justice of the European Union (Apr. 6, 1995), EU:C:1995:98.

[30] Id., 29, Case C-418/01, IMS Health GmbH & Co. OHG v. NDC Health GmbH & Co. GH (Apr. 29, 2004), EU:C:2004:257.

[31] European General Court, supra note 24.

[32] Case C-7/97, Oscar Bronner GmbH & Co. KG v. Mediaprint Zeitungs- und Zeitschriftenverlag GmbH & Co. KG, Mediaprint Zeitungsvertriebsgesellschaft mbH & Co. KG and Mediaprint Anzeigengesellschaft mbH & Co. KG, Court of Justice of the European Union (Nov. 26, 1998), EU:C:1998:569.

[33] Id., Case C-165/19 P, Slovak Telekom a.s. v. Commission (Mar. 25, 2021) para. 50, EU:C:2021:239.

[34] Ibid., para. 57. In a similar vein, see Case T?814/17, Lietuvos gele?inkeliai AB v. Commission, European General Court (Nov. 18, 2020), para. 92, EU:T:2020:545.

[35] European Commission, supra note 22, para. 80.

[36] Ibid., para. 82; and Case COMP/38.784, Wanadoo Espan?a v. Telefo?nica, European Commission (Jul 4, 2007). This is likely to occur in two cases: where regulation compatible with EU law already imposes an obligation to supply on the dominant undertaking and it is clear, from the considerations underlying such regulation, that the necessary balancing of incentives has already been made by the public authority when imposing such obligation; or where the upstream market position of the dominant firm has been developed under the protection of special or exclusive rights or has been financed by state resources.

[37] Case C-280/08 P, Deutsche Telekom AG v. European Commission (Deutsche Telekom I), Court of Justice of the European Union (Oct. 14, 2010), EU:C:2010:603.

[38] Id., Case C-52/09, Konkurrensverket v. TeliaSonera Sverige AB (Feb. 17, 2011), EU:C:2011:83.

[39] Id., Case C?295/12 P, Telefónica SA and Telefónica de España SAU v. European Commission (Jul. 10, 2014), EU:C:2014:2062.

[40] Id., supra note 33.

[41] Case C/13/528337, VBO Makelaar v. Funda, Gerechtshof Amsterdam (May 16, 2020), para. 3.12.3.

[42] Autorità Garante della Concorrenza e del Mercato, supra note 13, paras. 236, 393, 436, 504, 708, 710, 716, and 901.

[43] European Commission, supra note 11.

[44] Streetmap.EU Ltd. v. Google and Others, [2016] EWHC 253 (ch).

[45] Ibid., paras. 51-54.

[46] Ibid., paras. 84 and 147.

[47] Ibid., para. 84.

[48] Ibid., para. 149.

[49] Ibid., para. 139.

[50] Ibid., para. 161.

[51] European Commission, supra note 11, para. 344.

[52] Ibid., para. 649.

[53] Ibid.

[54] Case C-333/94, Tetra Pak International SA v. Commission, Court of Justice of the European Union (Nov. 14, 1996), EU:C:1996:436.

[55] Case T-288/97, Irish Sugar plc v Commission, European General Court (Oct. 7, 1999), EU:T:1999:246.

[56] Id., supra note 24.

[57] Case C-311/84, Centre Belge D’Etudes de Marché – Télémarketing (CBEM) v. SA Compagnie Luxembourgeoise de Télédiffusion (CLT) and Information Publicité Benelux (IPB), Court of Justice of the European Union (Oct. 3, 1985), EU:C:1985:394.

[58] European Commission, supra note 11, para. 334.

[59] Ibid., para. 645.

[60] Ibid., para. 651.

[61] See, e.g., Jay Pil Choi and Doh-Shin Jeon, A Leverage Theory of Tying in Two-Sided Markets with Nonnegative Price Constraints, 13 Am Econ J Microecon 283 (2021); Edward Iacobucci and Francesco Ducci, The Google Search Case in Europe: Tying and the Single Monopoly Profit Theorem in Two?Sided Markets, 47 Eur. J. Law Econ. 15 (2019); Eduardo Aguilera Valdivia, The Scope of the ‘Special Responsibility’ upon Vertically Integrated Dominant Firms after the Google Shopping Case: Is There a Duty to Treat Rivals Equally and Refrain from Favouring Own Related Business?, 41 World Competition 43 (2018);  Pinar Akman, The Theory of Abuse in Google Search: A Positive and Normative Assessment under EU Competition Law, 2 J. Tech. L. & Pol’y 301 (2017); Ioannis Kokkoris, The Google Case in the EU: Is There a Case?, 62 Antitrust Bull. 313 (2017); John Temple Lang, Comparing Microsoft and Google: The Concept of Exclusionary Abuse, 39 World Competition 5 (2016); Renato Nazzini, Unequal Treatment by Online Platforms: A Structured Approach to the Abuse Test in Google, SSRN (2016), https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2815081; Bo Vesterdorf, Theories of Self-Preferencing and Duty to Deal – Two Sides of the Same Coin?, 1(1) CLPD 4 (2015); Petit, supra note 19.

[62] Cre?mer, de Montjoye, and Schweitzer, supra note 1.

[63] Gerechtshof Amsterdam, supra note 41.

[64] Ibid., para. 3.12.1.

[65] Autorité de la Concurrence, supra note 13.

[66] Digital Advertising Services Inquiry Final Report, Australian Competition and Consumer Commission (2021), available at https://www.accc.gov.au/system/files/Digital%20advertising%20services%20inquiry%20-%20final%20report.pdf.

[67] Autorité de la Concurrence, supra note 13, paras. 369 and 410.

[68] Ibid., paras. 366 and 369. Since Google did not hold a dominant position in the market for SSPs, the reference to Tetra Pak has also been used to argue that, under specific circumstances, behavior implemented in a non-dominated market that has effects on the dominated market may be considered abusive (see para. 367).

[69] CJEU, supra note 33.

[70] Decision 21-D-07, Apple, Autorité de la Concurrence, (Mar. 17, 2021) https://www.autoritedelaconcurrence.fr/en/decision/regarding-request-interim-measures-submitted-associations-interactive-advertising-bureau.

[71] Ibid., para. 163.

[72] Bundeskartellamt Reviews Apple’s Tracking Rules for Third-Party Apps (press release), Bundeskartellamt (2022) https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2022/14_06_2022_Apple.html.

[73] Mobile Ecosystems: Market Study Final Report, U.K. Competition and Markets Authority (2022), Chapter 6 and Appendix J, https://www.gov.uk/cma-cases/mobile-ecosystems-market-study.

[74] European General Court, supra note 12.

[75] Ibid., para. 155.

[76] More recently, see also Case 50972, Google Privacy Sandbox, U.K. Competition and Markets Authority (Feb. 11, 2022), https://www.gov.uk/cma-cases/investigation-into-googles-privacy-sandbox-browser-changes, which considered self-preferencing as a traditional discrimination abuse. In particular, the competitive risks the CMA highlighted involve Google’s self-preferencing its own ad inventory and ad-tech services by transferring key functionalities to Chrome. The Privacy Sandbox Project would offer Google the ability to affect digital-advertising-market outcomes through Chrome in a way that cannot be scrutinized by third parties. It could lead to conflicts of interests because Google operates as publisher and ad-tech provider simultaneously.

[77] General Court, supra note 12, para. 160.

[78] Ibid., para. 163.

[79] Case C-242/95, GT-Link A/S v. De Danske Statsbaner (DSB), Court of Justice of the European Union (Jul. 17, 1997), EU:C:1997:376.

[80] Id., Case C-82/01 P, Ae?roports de Paris v. Commission (Oct. 24, 2002), EU:C:2002:617.

[81] Id., supra note 21.

[82] See Lena Hornkohl, Article 102 TFEU, Equal Treatment and Discrimination after Google Shopping, 13 J. Eur. Compet. Law Pract. 99 (2022), mentioning Post Danmark I as an example of primary-line exclusionary discrimination in the predatory-pricing context and TeliaSonera (supra note 38) and Slovak Telekom (supra note 33) as examples of secondary-line exclusionary discrimination in the margin-squeeze context.

[83] General Court, supra note 12, paras. 166 and 175.

[84] Ibid., paras. 169-174.

[85] Ibid., paras. 176-177.

[86] Ibid.

[87] Ibid., para. 178.

[88] Regulation (EU) 2015/2120 laying down measures concerning open Internet access and amending Directive 2002/22/EC on universal service and users’ rights relating to electronic communications networks and services, and Regulation (EU) No 531/2012 on roaming on public mobile-communications networks within the European Union, (2015) OJ L 310/1.

[89] General Court, supra note 12, para. 180. Comparisons to net neutrality have also been made by the U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law, supra note 10, 382-383, recommending that Congress consider establishing nondiscrimination rules.

[90] General Court, supra note 12, paras. 180, 182 and 183.

[91] Ibid., paras. 178, 182, 183, and 237.

[92] Ibid., paras. 176 and 179.

[93] Ibid..

[94] Ibid., paras. 176 and 179.

[95] Ibid., para. 181.

[96] Ibid., paras. 182-184.

[97] Ibid., paras. 219-222.

[98] Ibid., paras. 227.

[99] Ibid., para. 224.

[100] Ibid., paras. 232 and 233.

[101] Ibid., para. 238.

[102] Ibid., para. 240.

[103] See Christian Ahlborn, Gerwin Van Gerven, and William Leslie, Bronner Revisited: Google Shopping and the Resurrection of Discrimination Under Article 102 TFEU, 13 J. Eur. Compet. Law Pract. 87 (2022); Friso Bostoen, The General Court’s Google Shopping Judgement Finetuning the Legal Qualification and Tests for Platform Abuse, 13 J. Eur. Compet. Law Pract. 75 (2022), and Hornkohl, supra note 82, arguing that the ruling has resurrected discriminatory abuses as potentially one of the most important tools for regulating the platform economy.

[104] See Elias Deutscher, Google Shopping and the Quest for a Legal Test for Self-preferencing Under Article 102 TFEU, 6 European Papers 1345, 1348 (2021), arguing that the judgment did not address the fundamental question of how far Article 102 TFEU can interfere with the design choices of dominant firms or prohibit them from granting favorable treatment to their own products or services.

[105] General Court, supra note 12, para. 196.

[106] Ibid., para. 187.

[107] Ibid., para. 245.

[108] See Bostoen, supra note 103, arguing that, at best, this form of favoritism may be applicable in other cases of prominent display and positioning in searches (e.g., the conduct of Amazon favoring its own retail offers); and Ahlborn, Van Gerven, and Leslie, supra note 103, noting that, in contrast to price discrimination, discriminatory access to an input can encompass a range of factors that are difficult to disentangle (e.g., greater interoperability or preferential access to core services).

[109] Autorità Garante della Concorrenza e del Mercato, supra note 13.

[110] Ibid., para. 702, describing Amazon’s conduct as an “abusive pressure.”

[111] Id., Italian Competition Authority: Amazon Fined over € 1,128 Billion for Abusing Its Dominant Position (2021), https://en.agcm.it/en/media/press-releases/2021/12/A528.

[112] Autorità Garante della Concorrenza e del Mercato, supra note 13, para. 254.

[113] Ibid..

[114] Ibid., paras. 253 and 737.

[115] Ibid., para. 698.

[116] See C(2020) 7692 Final, European Commission (Nov. 10, 2020). See also Margrethe Vestager, Statement by Executive Vice-President Vestager on Statement of Objections to Amazon for the Use of Non-Public Independent Seller Data and Second Investigation into Its E-Commerce Business Practices (2020), https://ec.europa.eu/commission/presscorner/detail/en/STATEMENT_20_2082. See also CMA Investigates Amazon over Suspected Anti-Competitive Practices, U.K. Competition and Markets Authority (2022), https://www.gov.uk/government/news/cma-investigates-amazon-over-suspected-anti-competitive-practices, opening an investigation into how Amazon sets criteria for allocation of suppliers to be the preferred in the Buy Box and how Amazon sets the eligibility criteria for selling under the Prime label. The European Commission is currently evaluating the commitments offered by Amazon (Commission Seeks Feedback on Commitments Offered by Amazon Concerning Marketplace Seller Data and Access to Buy Box and Prime, European Commission (2022) https://ec.europa.eu/commission/presscorner/detail/en/ip_22_4522). Amazon has committed to apply equal treatment to all sellers when selecting the winner of the Buy Box and to display a second competing offer to the Buy Box winner if there is a second offer that is sufficiently differentiated on price and/or delivery. Both offers will display the same descriptive information and provide for the same purchasing experience. Moreover, regarding Prime, Amazon has committed to set non-discriminatory conditions and qualifying criteria for marketplace sellers and offers, to allow Prime sellers to freely choose any carrier for their logistics and delivery services, and not to use any information obtained through Prime about the terms and performance of third-party carriers for its own logistics services.

[117] Bundeswettbewerbsbehörde (Jul. 17, 2019), available at https://www.bwb.gv.at/fileadmin/user_upload/Fallbericht_20190911_en.pdf.

[118] Ibid., para. 81.

[119] Ibid., para. 87.

[120] U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law, supra note 10, 287-291.

[121] Ibid., 287-288.

[122] Autorità Garante della Concorrenza e del Mercato, supra note 13, paras. 236, 504, 710, 716, and 901.

[123] Ibid., paras. 236, 504, 506, 716, 723, and 810.

[124] Ibid., paras. 505, 713, 726, 760, 826, 852, 857, and 874. See also Autorità Garante della Concorrenza e del Mercato, supra note 111.

[125] Id., supra note 13, para. 807.

[126] Ibid., paras. 127, 136, 188, 614, and 804.

[127] Ibid., paras. 834-835.

[128] Ibid., paras. 658-666, 679, and 682.

[129] Ibid., paras. 805-806.

[130] Ibid., para. 136.

[131] Ibid., paras. 506, 609, 610, 680, and 716.

[132] Ibid., paras. 610, 710, 716, and 723.

[133] Ibid., para. 710.

[134] Case AT.40462, Amazon Marketplace, European Commission (Nov. 10, 2020). See also Commission Sends Statement of Objections to Amazon for the Use of Non-Public Independent Seller Data and Opens Second Investigation into Its E-Commerce Business Practices, European Commission, (2020) https://ec.europa.eu/commission/presscorner/detail/en/ip_20_2077.

[135] U.K. Competition and Markets Authority, supra note 116.

[136] U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law, supra note 10, 275.

[137] App Marketplaces: Interim Report, Australian Competition and Consumer Commission, (2021), 134, available at https://www.accc.gov.au/system/files/Digital%20platform%20services%20inquiry%20-%20March%202021%20interim%20report.pdf.

[138] Case AT.40684, Facebook leveraging, European Commission (Jun. 4, 2021), https://ec.europa.eu/competition/elojade/isef/case_details.cfm?proc_code=1_AT_40684; CMA Investigates Facebook’s Use of Ad Data, U.K. Competition and Markets Authority (2021), https://www.gov.uk/government/news/cma-investigates-facebook-s-use-of-ad-data.

[139] Case AT.40716, Apple – App Store Practices, European Commission (Jun. 16, 2020).

[140] DMA, supra note 5, Article 6(1).

[141] AICOA, supra note 8, Section 3.

[142] See, e.g., U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law, supra note 10, 380. See also Simon P. Anderson and O?zlem Bedre-Defolie, Hybrid Platform Model, CEPR Discussion Paper No. 16243 (2021), available at https://cepr.org/active/publications/discussion_papers/dp.php?dpno=16243, arguing that the hybrid business model leads to higher platform fees for third-party sellers, less variety on the platform, and lower consumer welfare, compared to when the platform is a pure marketplace. On a different note, see Herbert Hovenkamp, Antitrust and Platform Monopoly, 130 Yale Law J. 1952 (2021), considering structural separation as the worst solution for the problems raised by Amazon’s strategy. See also Andrei Hagiu, Tat-How Teh, and Julian Wright, Should Platforms Be Allowed to Sell on Their Own Marketplaces?, 53 Rand J Econ 297 (2022), arguing that a structural remedy, such as an outright ban on the dual mode, would be detrimental to consumer surplus or total welfare, since the presence of the intermediary’s products constrains the pricing of the third-party sellers on its marketplace. The authors consider preferable policy interventions that target specific behaviors by the platform, such as a ban on product imitation and on self-preferencing.

[143] European Commission, supra note 134.

[144] U.S. House Judiciary Subcommittee on Antitrust, Commercial, and Administrative Law, supra note 10, 275.

[145] Digital Markets Act: Impact Assessment Support Study (Annexes), European Commission (2020), 301-308, https://op.europa.eu/en/publication-detail/-/publication/2a69fd2a-3e8a-11eb-b27b-01aa75ed71a1/language-en/format-PDF/source-search.

[146] Ibid., 304.

[147] Ibid..

[148] See Hagiu, Teh, and Wright, supra note 142, arguing that a ban on product imitation by a platform restores sellers’ incentive to innovate; Erik Madsen and Nikhil Vellodi, Insider Imitation, SSRN (2022) available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3832712, finding that a ban on a platform’s use of marketplace data can either stifle or stimulate innovation, depending on the nature of innovation. Namely, it stimulates innovation only for experimental product categories with significant upside demand potential. On the impact of data usage on consumer welfare, because of the tradeoff between static benefits from lower prices and dynamic costs from lower sellers’ incentives to investment, see Federico Etro, Product Selection in Online Marketplaces, 30 J Econ Manag Strategy 614 (2021).

[149] See Herbert Hovenkamp, The Looming Crisis in Antitrust Economics, 101 B.U. L. 489, 543 (2021), finding no evidence to suggest that the practice is so prone to abuse or so likely to harm consumers in other ways that it should be categorically condemned: “Rather, it is an act of partial vertical integration similar to other practices that the antitrust laws have confronted and allowed in the past. One close analogy is dual distribution, which occurs when a firm sells through both independent franchisees and its wholly owned stores. Such practices nearly always increase output, benefitting consumers and typically even independent competing firms.” On the different impact of Amazon’s practice on consumer welfare and third-party sellers’ welfare, see also Feng Zhu and Qihong Liu, Competing with Complementors: An Empirical Look at Amazon.com, 39 Strateg. Manag. J. 2618 (2018), finding that Amazon tends to enter into high-quality, popular products sold by third-party merchants and that such entry tends to lower prices and lead to the exit of third-party sellers; and Nan Chen and Hsin-Tien Tsai, Steering via Algorithmic Recommendations, (2021) https://www.tse-fr.eu/sites/default/files/TSE/documents/sem2021/tsai.pdf, arguing that Amazon tends to recommend products sold by Amazon Retail to consumers over products sold by third-party retailers, and that this steering is inconsistent with Amazon promoting consumer welfare.

[150] Hovenkamp, supra note 142, 2015-2016.

[151] European Commission, supra note 116.

[152] See Graef, supra note 18, distinguishing Google Shopping and Amazon Marketplace as pure self-preferencing cases (i.e., primary-line injuries whose key objective is to exclude rivals from the market) from the Italian Amazon Logistics case, which belongs instead to a hybrid category that includes a mix of exploitative and exclusionary elements.

[153] Australian Competition and Consumer Commission, supra note 66, 92.

[154] Colomo, supra note 15, 418.

[155] Updating Competition and Consumer Law for Digital Platform Services, Australian Competition and Consumer Commission (2022), 85, https://www.accc.gov.au/media-release/feedback-sought-on-potential-new-rules-for-large-digital-platforms.

[156] Tobias Kretschmer, Aija Leiponen, Melissa Schilling, and Gurneeta Vasudeva, Platform Ecosystems as Meta?Organizations: Implications for Platform Strategies, 43 Strateg. Manag. J. 405 (2022).

[157] Kevin J. Boudreau and Andrei Hagiu, Platforms Rules: Multi-Sided Platforms as Regulators, in (Annabelle Gawer, ed.) Platforms, Markets and Innovation, Cheltenham, Edward Elgar Publishing (2009), 163; David Evans, Governing Bad Behavior by Users of Multi-sided Platforms, 27 BTLJ 1201 (2012).

[158] See Hovenkamp, supra note 17, considering the pending U.S. self-preferencing legislation “an affront to both antitrust policy and intelligent regulatory policy.” See also Geoffrey A. Manne, Against the Vertical Discrimination Presumption, 2 Concurrences 1, 2 (2020), arguing that forcing platforms to allow complementors to compete on their own terms would affect platform incentives for innovation. Indeed, platforms have an incentive to optimize openness and mandating maximum openness is not necessarily optimal because it would disregard the trade-off faced by platforms. Consequently, any presumption of harm from vertical discrimination is not based on sound economics. In a similar vein, Jonathan M. Barnett, The Host’s Dilemma: Strategic Forfeiture in Platform Markets for Informational Goods, 124 Harv. L. Rev. 1861 (2011).

[159] See Australian Competition and Consumer Commission, supra note 155, 87, arguing that rules might need to be “specifically tailored to each digital platform service with a high level of precision, to target the specific conduct that causes anti-competitive harm.”

[160] See Patrice Bougette, Axel Gautier, and Fre?de?ric Marty, Business Models and Incentives: For an Effects-Based Approach of Self-Preferencing?, 13 J. Eur. Compet. Law Pract.136, 140 (2022). On the welfare effects of Amazon’s dual role and the welfare implications of proposed regulations, see Germa?n Gutie?rrez, The Welfare Consequences of Regulating Amazon, (2021) available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3965566, showing that interventions that eliminate either the Prime program or product variety are likely to decrease welfare. See also Guy Aridor and Duarte Gonçalves, Recommenders’ Originals: The Welfare Effects of the Dual Role of Platforms as Producers and Recommender Systems, 83 Int. J. Ind. Organ.102845 (2022), highlighting the importance of targeted restrictions on self-preferencing because of the ambiguity of the welfare implications of a policy remedy separating recommendation and production or imposing unbiased recommendations. However, with regards to app stores and device-funded gatekeepers, Jorge Padilla, Joe Perkins, and Salvatore Piccolo, Self-Preferencing in Markets with Vertically Integrated Gatekeeper Platforms, J. Ind. Econ. (forthcoming) find that consumer welfare would be increased by preventing the device seller from selling its own apps and associated services in competition with third-party apps. See also Morgane Cure, Matthias Hunold, Reinhold Kesler, Ulrich Laitenberger, and Thomas Larrieu, Vertical Integration of Platforms and Product Prominence, Quant. Mark. Econ. (forthcoming), studying the potential effects of self-preferencing in the online hotel-booking industry because of the integration between one of the major online travel agencies (Booking Holdings) and a meta-search platform (Kayak). According to their empirical findings, the horizontal ranking of sales channels for a given hotel indicate that sales channels of online travel agents by Booking Holdings are more often position leaders than price leaders and online travel agents affiliated to Booking Holdings have a higher probability than other online travel agents of being among the visible providers and of being the highlighted sales channel. Moreover, for the vertical ranking of hotels for a search request, hotels are ranked worse in the Kayak search results when an online travel agent of the Expedia Group is the cheapest sales channel.

[161] Colomo, supra note 15, 421; Id., Anticompetitive Effects in EU Competition Law, 17 J. Competition Law Econ. 309, 356 (2021).

[162] Case C-377/20, Servizio Elettrico Nazionale SpA v. Autorità Garante della Concorrenza e del Mercato, Court of Justice of the European Union (May 12, 2022), EU:C:2022:379.

[163] Ibid., para. 46.

[164] CJEU, supra note 20, paras. 133-134.

[165] CJEU, supra note 162, para. 73.

[166] Case C?377/20, Servizio Elettrico Nazionale SpA v. Autorità Garante della Concorrenza e del Mercato, Advocate General Rantos (Dec. 9, 2021), EU:C:2021:998, para. 43.

[167] Ibid., para. 45.

[168] See CJEU, TeliaSonera, supra note 38; Post Danmark I, supra note 82; Case C-23/14, Post Danmark A/S v. Konkurrencerådet (Post Danmark II), (Oct. 6, 2015), EU:C:2015:651; Intel, supra note 20; Case C-307/18, Generics (UK) and Others v. Competition and Markets Authority (Jan. 30, 2020), EU:C:2020:52; Case C-152/19 P, Deutsche Telekom v. Commission (Mar. 25, 2021) EU:C:2021:238.

[169] CJEU, supra note 162, paras. 84-85.

[170] Ibid., para. 75. See also Advocate General Rantos, supra note 166, paras. 48-50, arguing that demonstrating that a dominant undertaking used means other than those that come within the scope of normal competition is not a requirement that needs to be assessed separately from the restrictive effect of the conduct.

[171] See, e.g., CJEU, MEO, supra note 21, and Post Danmark II, supra note 168. See also Colomo, supra note 161.

[172] Advocate General Rantos, supra note 166, para. 62.

[173] Ibid., para. 68 and CJEU, supra note 162, paras. 77-79.

[174] CJEU, supra note 162, para. 79.

[175] Case C?23/14, Post Danmark A/S v. Konkurrencerådet, Advocate General Kokott (May 21, 2015), EU:C:2015:343, para. 4.

The FTC Did Not ‘Fumble the Future’ in Its Google Search Investigation

Politico has released a cache of confidential Federal Trade Commission (FTC) documents in connection with a series of articles on the commission’s antitrust probe into Google Search a decade ago. The headline of the first piece in the series argues the FTC “fumbled the future” by failing to follow through on staff recommendations to pursue antitrust intervention against the company. 

Read the full piece here.

The Real Reason Foundem Foundered

Summary

A pair of recent, long-form articles in the New York Times Magazine and Wired UK — the latest in a virtual journalistic cottage industry of such articles — chronicle the downfall of British price comparison site and stalwart Google provocateur, Foundem, and attribute its demise to anticompetitive behavior on the part of Google.

Unfortunately, the media’s hagiographies of Foundem and its founders, Shivaun and Adam Raff, approach the antitrust question as if it were imbued with the simple morality of a David vs. Goliath tale. The reality is far more complicated. In fact, these articles misunderstand and misstate the critical economic, business, and legal realities of Google Search, of Foundem’s claims of harm, and of the relationship between the two.

Was Foundem’s failure really the result of anticompetitive “gatekeeping” on Google’s part? Or could it simply be a pedestrian tale of yet another tech start-up that failed because its founders didn’t appreciate that a successful business is built on more than just a good idea?

While the import of the Foundem story has been misconstrued by journalists and EU regulators, it is useful in illuminating what may actually be the fundamental question regarding the antitrust fortunes of the platform economy:

What, if anything, does a successful platform “owe” to the companies that make themselves dependent upon it?

If Search Neutrality Is the Answer, What’s the Question?

Summary

In recent months a veritable legal and policy frenzy has erupted around Google generally, and more specifically concerning how its search activities should be regulated by government authorities throughout the world in the name of ensuring “search neutrality.”  Concerns with search engine bias have led to a menu of proposed regulatory reactions.  Although the debate has focused upon possible remedies to the “problem” presented by a range of Google’s business decisions, it has largely missed the predicate question of whether search engine bias is the product of market failure or otherwise generates significant economic or social harms meriting regulatory intervention in the first place.  “Search neutrality” by its very name presupposes that mandatory neutrality or some imposition of restrictions on search engine bias is desirable, but it is an open question whether advocates of search neutrality have demonstrated that there is a problem necessitating any of the various prescribed remedies. This paper attempts to answer that question, and we evaluate both the economic and non-economic costs and benefits of search bias, as well as the solutions proposed to remedy perceived costs. We demonstrate that search bias is the product of the competitive process and link the search bias debate to the economic and empirical literature on vertical integration and the generally-efficient and pro-competitive incentives for a vertically integrated firm to favor its own content. We conclude that neither an ex ante regulatory restriction on search engine bias nor the imposition of an antitrust duty to deal upon Google would benefit consumers. Moreover, in considering the proposed remedies, we find that by they substitute away from the traditional antitrust consumer welfare standard, and would impose costs exceeding any potential benefits.

Amazon:

The Amazon investigation and Europe’s “Big Tech” Crusade

The dust has barely settled on the European Commission’s record-breaking €4.3 Billion Google Android fine, but already the European Commission is gearing up for its next high-profile case. Last month, Margrethe Vestager dropped a competition bombshell: the European watchdog is looking into the behavior of Amazon. Should the Commission decide to move further with the investigation, Amazon will likely join other US tech firms such as Microsoft, Intel, Qualcomm and, of course, Google, who have all been on the receiving end of European competition enforcement.

Read the full piece here.

Amazon Italy’s Efficiency Offense

Early last month, the Italian competition authority issued a record 1.128 billion euro fine against Amazon for abuse of dominance under Article 102 of the Treaty on the Functioning of the European Union (TFEU). In its order, the Agenzia Garante della Concorrenza e del Mercato (AGCM) essentially argues that Amazon has combined its Amazon.it marketplace and Fulfillment by Amazon (FBA) services to exclude logistics rivals such as FedEx, DHL, UPS, and Poste Italiane.

Read the full piece.

The Real Story about Amazon, Counterfeit Listings, and Minimum Advertised Price (MAP) Policies

These days, lacking a coherent legal theory presents no challenge to the would-be antitrust crusader. In a previous post, we noted how Shaoul Sussman’s predatory pricing claims against Amazon lacked a serious legal foundation. Sussman has returned with a new post, trying to build out his fledgling theory, but fares little better under even casual scrutiny.

Read the full piece here.

The Dystopia of a World Without Self-Preferencing

Not only would banning self-preferencing not benefit consumers, but such a mandate would unleash any number of potential harms. In 2022, ICLE hosted a symposium at Truth on the Market in which we invited scholars to opine on what the future would look like following the passage of proposed legislation to ban self-preferencing. The resulting pieces offer some concrete insights into the costs of such a world. 

Selected entries from the Symposium

Legislative Proposals to Ban Self-Preferencing Are Misguided

Several proposals and pieces of proposed legislation would impose bans on self-preferencing. Previously, competition authorities have judged this type of conduct under the consumer welfare standard: does it hurt consumers in the long run, or does it help them? This test does seek to evaluate whether the conduct deprives consumers of choice by foreclosing rivals, which could ultimately allow the platform to exploit its customers. But it doesn’t treat harm to competitors — in the form of reduced traffic and profits for Yelp, for example — as a problem in and of itself.

“Non-discrimination” bills introduced in both the House and Senate in 2021 aim to change that, by creating an effective “per se” rule against self-preferencing based solely on claims of harm from competitors. Such proposals are costly and misguided.

AICOA Is Neither Urgently Needed Nor Good: A Response to Professors Scott Morton, Salop, and Dinielli

Earlier this month, Professors Fiona Scott Morton, Steve Salop, and David Dinielli penned a letter expressing their “strong support” for the proposed American Innovation and Choice Online Act (AICOA). In the letter, the professors address criticisms of AICOA and urge its approval, despite possible imperfections.

Read the full piece here.

Breaking Down the American Choice and Innovation Online Act

The American Choice and Innovation Online Act (previously called the Platform Anti-Monopoly Act), introduced earlier this summer by U.S. Rep. David Cicilline (D-R.I.), would significantly change the nature of digital platforms and, with them, the internet itself. Taken together, the bill’s provisions would turn platforms into passive intermediaries, undermining many of the features that make them valuable to consumers. This seems likely to remain the case even after potential revisions intended to minimize the bill’s unintended consequences.

Read the full piece here.

5 Thoughts on the Senate’s Proposed Platform Self-Preferencing Ban

A bipartisan group of senators unveiled legislation today that would dramatically curtail the ability of online platforms to “self-preference” their own services—for example, when Apple pre-installs its own Weather or Podcasts apps on the iPhone, giving it an advantage that independent apps don’t have. The measure accompanies a House bill that included similar provisions, with some changes.

Read the full piece here.

Innovation and innovative startups could be an unintended casualty

Bills such as the American Innovation and Choice Online Act, which would ban self-preferencing, may not directly restrict startup acquisitions, but the activities that they would restrict most certainly do dramatically affect the incentives that drive many startup acquisitions. If a platform is prohibited from engaging in cross-platform integration of acquired technologies, or if it can’t monetize its purchase by prioritizing its own technology, it may lose the motivation to make a purchase in the first place.

Technology Mergers and the Market for Corporate Control

Abstract

A growing number of policymakers and scholars are calling for tougher rules to curb corporate acquisitions. But these appeals are premature. There is currently little evidence to suggest that mergers systematically harm consumer welfare. More importantly, scholars fail to identify alternative institutional arrangements that would capture the anticompetitive mergers that evade prosecution without disproportionate false positives and administrative costs. Their proposals thus fail to meet the requirements of the error-cost framework.

Several high-profile academic articles and reports claim to have identified important gaps in current merger enforcement rules, particularly with respect to tech and pharma acquisitions involving nascent and potential competitors—so-called “killer acquisitions” and “kill zones.” As a result of these perceived deficiencies, scholars and enforcers have called for tougher rules, including the introduction of lower merger filing thresholds and substantive changes, such as the inversion of the burden of proof when authorities review mergers and acquisitions in the digital platform industry. Meanwhile, and seemingly in response to the increased political and advocacy pressures around the issue, U.S. antitrust enforcers have recently undertaken several enforcement actions directly targeting such acquisitions.

As this paper discusses, however, these proposals tend to overlook the important tradeoffs that would ensue from attempts to decrease the number of false positives under existing merger rules and thresholds. While merger enforcement ought to be mindful of these possible theories of harm, the theories and evidence are not nearly as robust as many proponents suggest. Most importantly, there is insufficient basis to conclude that the costs of permitting the behavior they identify is greater than the costs would be of increasing enforcement to prohibit it.

Our work draws from two key strands of economic literature that are routinely overlooked (or summarily dismissed) by critics of the status quo. For a start, as Frank Easterbrook argued in his pioneering work on The Limits of Antitrust, antitrust enforcement is anything but costless. In the case of merger enforcement, not only is it expensive for agencies to detect anticompetitive deals, but overbearing rules may deter beneficial merger activity that creates value for consumers. Indeed, not only are most mergers welfare-enhancing, but barriers to merger activity have been shown to significantly, and negatively, affect early company investment.

Second, critics mistake the nature of causality. Scholars routinely surmise that incumbents use mergers to shield themselves from competition. Acquisitions are thus seen as a means of eliminating competition. But this overlooks an important alternative: It is at least plausible that incumbents’ superior managerial or other capabilities make them the ideal purchasers for entrepreneurs and startup investors who are looking to sell. This dynamic is likely to be amplified where the acquirer and acquiree operate in overlapping lines of business. In other words, competitive advantage, and the ability to profitably acquire other firms, might be caused by business acumen rather than anticompetitive behavior.

Thus, significant and high-profile M&A activity involving would-be competitors may be the procompetitive byproduct of a well-managed business, rather than anticompetitive efforts to stifle competition. Critics systematically overlook this possibility. Indeed, Henry Manne’s seminal work on Mergers and Market for Corporate Control—the first to argue that mergers are a means of applying superior management practices to new assets—is almost never cited by contemporary researchers in this space. Our paper attempts to set the record straight.

Read the full white paper here.