ICLE White Paper

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.