Due to their global economic significance and their attitude to concentrating economic power, digital platforms have triggered an unprecedented competition policy debate, fragmented and multi-faceted in terms of both actual problems and potential solutions. Doubts concern whether “self-preferencing” behaviors constitute a new self-standing theory of antitrust liability; and, whether this being the case, what “legal standard” should be used to pursue self-preferencing theories. This paper turns the spotlight on these crucial issues. Additionally, the paper considers the digital platforms’ regulatory interventions recently announced, evaluates their relationship alongside competition rules and argues that the new regulatory framework, including the introduction of new competition tools, may significantly affect the assessment of conduct by digital platforms, even abuse of dominance cases.

By Alessandra Tonazzi & Gabriele Carovano1



Digital platforms have taken the spotlight in the competition policy debate due to their significance to the global economy and their attitude to concentrating economic power.

Due to the roles of big data, network effects, economies of scope and scale, etc., digital platforms tend to be self-reinforcing, and to lead to winner-takes-all or winner-takes-most scenarios. Because of these features, digital platforms, in addition to catching the attention of many academics, governments, and regulators worldwide, are also disorienting them.

The competition policy debate on digital platforms is fragmented and multi-faceted in terms of both actual (or would-be) problems and potential solutions. The positions in the debate range from calling for more competition2 to arguing that unrestricted competition can backfire, spiraling into a race to the bottom incapable of delivering the expected benefits.3 While some propose to expand the scope of competition law, by introducing new standards complementary to consumer welfare, others go in the opposite direction, advocating for traditional, purely economics-based antitrust enforcement.4 Some others, again, conscious of the limits of competition enforcement in dealing with digital markets, advocate for complementary regulatory intervention, limited to a subset of players with “strategic market status.”

In the context of these general policy discussions, competition authorities are taking different approaches to tackling competitive concerns arising from the conduct of digital platforms. At times authorities raise the criticism that the current “balancing of error costs” is mistakenly tilted in favour of under enforcement.5

In this complex and lively debate, so-called “self-preferencing” is one of the topics that has attracted particular attention. Specifically, it has been questioned (i) whether “self-preferencing” amounts to be a new self-standing theory of antitrust liability or not; (ii) if so, what is the “legal standard” to pursue self-preferencing theories (i.e. under what circumstances does self-preferencing by a dominant firm amount to an abuse of dominance?); and (iii) whether competition enforcement is sufficient to tackle the challenges of self-preferencing conduct, or whether a regulatory intervention is needed.



Generally speaking, “self-preferencing” refers to an undertaking treating itself, its services, or its subsidiaries favorably compared to its treatment of rival external competitors or costumers.6 While on the surface any self-preferencing conduct seems retraceable to a unique common figure as long as “dissimilar conditions” are applied to “equivalent transactions,” from a competition law perspective, it is worth differentiating at least between three discrimination types: (i) pure self-preferencing; (ii) pure secondary line differentiation; and (iii) hybrid differentiation.7

In “pure” self-preferencing scenarios, the self-preferencing firm is vertically integrated and the firm itself is the direct beneficiary of the more favorable treatment in the downstream market.8 By contrast, in “pure” secondary line differentiation scenarios, the discriminatory firm is not vertically integrated but, nonetheless, it directly benefits from its discriminatory conduct in its upstream market.9 In hybrid scenarios, regardless of whether the self-preferencing/discriminatory firm is vertically integrated, the self-preferencing/discriminatory firm benefits only indirectly from its discriminatory conduct in a market different from that in which the discrimination takes place. Consequently, the anticompetitive effects of the discriminatory conduct take place primarily in a market in which the discriminated subjects are not active themselves.

This paper, focusing on the European legal framework, will only deal with “pure” self-preferencing and “hybrid” differentiation conduct where the discriminatory firm is vertically integrated. From the above classification, it can be seen that while all “self-preferencing” conduct contain a certain discriminatory component, not all discriminations result in self-preferencing. A discriminatory conduct, to qualify as “self-preferencing,” needs to be executed by a vertically-adjacently integrated firm, regardless from whether the benefits resulting from the discriminatory conduct are directly (either upstream or downstream) or indirectly (in a third-adjacent-market) perceived by the discriminatory firm. It follows that self-preferencing does not have a pure “vertical” dimension, but it can also have a horizontal significance.



The limited number of cases (some still ongoing)10 and the absence of Court decisions make it difficult to determine (i) whether self-preferencing constitutes a new and independent theory of competition law liability under Article 102 TFEU, and (ii) which legal test to use to assess the lawfulness of self-preferencing conduct.

Whereas self-preferencing might cause anticompetitive effects, it is reasonable for an undertaking to preference (to a certain extent) its downstream services as a means to boost efficiencies, economies of scales, and recoup upstream investments. The efficiency arguments for self-preferencing are indeed similar to those that traditionally apply to vertical integration. Although this approach may differ in the digital economy, it is unclear and debated where the line should be drawn between pro-competitive and anti-competitive self-preferencing conduct.11

In debating the issue, some have argued that self-preferencing conduct by a dominant firm should be classified as abusive under Article 102 TFEU only insofar as it amounts to a “refusal to deal/supply.”12 In other words, the assessment of self-preferencing depends on the “refusal to supply test,” including the indispensability criterion, as established in Oscar Bronner13 and later jurisprudence. Additionally, the same scholarship highlights the difference between requiring a dominant undertaking not to treat its competitors “less favorably” and an obligation to grant terms that are “identical” to those offered to itself. As result, even where certain infrastructure is deemed “essential to compete,” the dominant firm’s duty to grant access to that infrastructure should only go as far as to allow the access seeker to be “commercially viable” and not to provide terms exactly “identical” to those provided to the infrastructure owner’s own services.

Others have argued that aside from a “refusal to deal,” there are several legal bases under which self-preferencing conduct may have competition law relevance, such as traditional leveraging theories including discrimination, tying, margin squeezes or even unfair pricing.14 Against such a multitude of legal tests for self-preferencing conduct, some have warned against their instrumentalization. Since the requirements to establish an abuse vary from one test to another, there is a risk that formal legal labels attributed to a given practice can be pigeonholed to frame the claim in a utilitarian manner, for instance to avoid “indispensability.”

The debate risks remaining academic since in practice there is little case law that has explicitly addressed it yet. Even in the EU Google Shopping case, the EC did not characterize Google’s conduct as “self-preferencing,” instead describing it as a “leveraging practice,” arguing that it constitutes “a well-established, independent, form of abuse falling outside the scope of competition on the merits.”15 While this is uncontroversial, it remains the case that “leveraging” is a general term, often used for various strategies (like tying, refusing to deal, or predatory pricing, etc.) that a dominant firm may use to extend its power from one market to another.16

Notwithstanding the remaining halo of uncertainty, the debate has the merit, among other things, of affirming that self-preferencing conduct may be abusive under Article 102 TFEU regardless of the applicability of the essential facility doctrine, although it does not automatically infer that self-preferencing constitute a new and independent theory of harm. All of the examples in European case law falling within the third “hybrid differentiation” category are grounded in already existing legal tests. Arguing that self-preferencing constitutes a new and independent theory of harm would expand the perimeter of application of EU competition law also to conduct that is not prosecutable under existing legal theories, i.e. conduct that despite having a discriminatory component, does not satisfy any of the existing theories of harm (refusal to deal, discrimination, tying, margin squeeze, unfair pricing, etc.). Notably, a new theory of harm will allow competition authorities (“CAs”) to prosecute “pure” self-preferencing conduct or those practices that fall within the “hybrid” category, which are currently not pursuable.17 In online “pure” self-preferencing cases, for example, the indispensability requirement could be substituted with other parameters testing the proportionality and reasonability of the self-preferencing conduct vis-à-vis the lawful aim that the dominant firm has publicly declared pursuing to its costumers/ consumers.

Another, and interesting way of looking at the role that the indispensability factor should play in assessing self-preferencing is the suggestion that one should look at the nature of the remedies. If they are proactive (i.e. requiring a positive obligation) the indispensability requirement should be considered part of the legal test.18 Such a conclusion would be consistent with Google Shopping where the Commission did not require indispensability because ending the infringement did not involve imposing a duty on Google to “transfer an asset or enter into agreements with persons with whom it has not chosen to contract”19 but simply dictated a principle to be followed, without imposing a modality to implement it.20

However, this suggestion does not seem entirely convincing. Using the case-specific needed remedy to classify the alleged anticompetitive conduct, would be as if in medicine, diseases were classified on the basis of adoptable treatments. If a drug is beneficial to several diseases, it does not make the latter identical in their characteristic features nor their differentiation useless. Unsurprisingly, the promoters of the “nature of the remedies criteria” end up arguing in favor of keeping the indispensability requirement as a proxy to balance the ex ante and ex post dimensions of competition21 to reduce interventions that would harm firms’ incentives to invest and innovate. Reasoning this way will lead to a scenario where the more complex the nature of the remedies needed to terminate an alleged anticompetitive conduct, the higher the standard of proof for authorities and claimants, and the lower would be the chances of having certain investigations commenced or certain conduct prosecuted. Such conclusions seem particularly unacceptable in digital markets, given the complexity of their dynamics and the information asymmetries existing between tech-giants, on the one hand, and CAs and claimants, on the other.

Depending on decisional practice and case law developments, “self-preferencing cases” may evolve in different directions. If a strict approach is adopted – basically requiring an “essential facility” scenario – the number of “self-preferencing cases” could be very limited. A second alternative would be to include “self-preferencing conduct” in a traditional theory of harms, without introducing a new “self-preferencing theory,” although this would require clarifying existing conditions and legal tests to assess abusive conduct. Such second alternative, in a more earth-breaking format, might even result in the fusion of all existing “legal categories” in a single, overarching “leveraging test” whose elements would need to be clearly defined without “indispensability” been one of them. Finally, the existing case law might be supplemented by the addition of self-preferencing among existing “legal categories.” To determine the perimeter of the application of the new “self-preferencing test,” however, a clear delimitation and classification of “abusive conduct” would be needed to avoid confusion and legal uncertainty.

In the authors’ view, while it seems appropriate not to require indispensability in online “pure” self-preferencing cases, there are multiple conditions that would still be needed and assessed to ensure that competition on the merits is not mistaken for abusive conduct. Among these conditions, significant are: (i) the assessment of firms’ market power (mindful that in digital markets, this assessment will probably go beyond traditional structural approaches based on market shares); (ii) the analysis of the competitive variables at stake; (iii) the impact of the alleged abusive conduct on such variables; (iv) the existence of anticompetitive effects; and (v) the absence of objective justifications. Whether this kind of assessment requires a new label, however, is for the EU Courts to say



This paper would not be complete without discussing other initiatives that could affect the legality of self-preferencing by digital platforms, that is the introduction of a new regulatory framework to address some competition law issues in digital markets that competition enforcement alone cannot solve.

Multiple regulatory interventions are currently being discussed in Europe and include different tools that would impose legal obligations on digital “gatekeepers.” If digital platforms’ responsibilities have to be expanded beyond the perimeter of Article 102 TFEU, some conduct – including self-preferencing – might fall within the scope of the new regulatory framework recently announced by the EC in one of its latest communications.22 Particularly, EU Commission Vice President Vestager, has recently highlighted her intention to review the fitness of EU competition rules for the digital age and explore the possibility of introducing ex ante rules within the context of the Digital Services Act package to ensure that markets dominated by tech platforms remain fair and contestable.23

Besides, some pieces of regulation have been already adopted with respect to platform-to-business relations. Significant in this sense is the regulatory intervention made on June 14, 2019, when the EU adopted its Regulation on fairness and transparency in platform-to-business relations (the so-called “P2B Regulation”).24 Notwithstanding this regulatory intervention, antitrust enforcement can still complement and expand the regulation. Indeed, although Regulation 1150/2019 refers to “fairness,” the majority of its provisions addresses transparency.25 Less than a month after its adoption, business-users’ protections were substantially expanded by the commitments required by the German Bundeskartellamt26 and the Austrian Bundeswettbewerbsbehörde27 on July 17, 2019, when both CAs simultaneously closed their abuse of dominance proceedings against Amazon.28

These developments show that antitrust and regulation should not be rivalry and exclusionary but rather complementary.29 One might argue that neither could competition alone solve some of the structural problems presented by digital markets (especially when markets are “tipping” and enforcement might intervene too late), nor could regulation nudge market forces as effectively as competition. Additionally, if public-utility regulation is ineffective, limited in scope, and permanent, competition enforcement also has several drawbacks (i.e. it is time consuming, resource-intensive, and not always definitive), that are dramatically accentuated in digital markets. The solution, therefore, seems to be in the middle. The problem, however, is how to best balance competition and regulation.

Keeping this question in mind, any regulation should take into account the aims of competition authorities and experts to preserve the correct functioning of markets and not to undermine innovation incentives. This contribution, for competition authorities, is not something new as they are regularly in dialogue with sector regulators and providing suggestions on how regulatory measures should not be disproportionate with respect to their objectives or unnecessarily restrict competition. Some have suggested the interesting idea that new forms of so-called “participative antitrust” or “regulatory antitrust” should be developed, whereby the private sector, industry representatives, regulators, and antitrust authorities jointly develop and evaluate short-term regulations.30

It is interesting to note that the discussion about a new and improved regulatory framework for digital platforms also includes the possible introduction of new competition tools. Although it is not clear yet what forms these new tools would take (structural interventions, special access obligations, etc.), this approach might blur the complementary nature of competition enforcement and regulation, tilting the balance towards a regulatory approach. In this new framework, the extent and limits to the enforcement of abuses of dominance might also be affected.

As result, regulatory interventions are welcomed and necessary to the extent that they strike the correct balance between competition and regulation. It follows that regulation’s scope must be proportionate and reasonable vis-à-vis its intended objectives. Otherwise, by excessively marginalizing antitrust enforcement, the risk of sub-optimal outcomes is inevitable since, as shown above, the performance and effectiveness of the each depends on the other.



While some academic contributions and public enforcement actions have shed some light on “self-preferencing conducts,” decisional practice concerning “self-preferencing cases” still needs elucidation. The picture might get clearer as more decisions are issued by competition authorities and Courts. At the same time, the announced changes in the regulatory framework for digital platforms, including the introduction of new competition tools, may significantly affect the assessment of conduct by digital platforms, even abuse of dominance cases.

1 Alessandra Tonazzi is Director of the International and European Affairs Division at the Autorità Garante della Concorrenza e del Mercato (“AGCM”). The opinions expressed in this paper are solely those of the author and cannot be interpreted as reflecting those of the Autorità Garante della Concorrenza e del Mercato. The authors can be contacted at Alessandra.Tonazzi@agcm.it. Gabriele Carovano is a PhD candidate at King’s College London. The author can be contacted at Gabriele.Carovano@kcl.ac.uk.

2 See Herbert Hovenkamp & Carl Shapiro, “Horizontal Mergers, Market Structure, and Burdens of Proof,” Faculty Scholarship, (2018).

3 See M.E. Stucke, A. Ezrachi, Competition Overdose, Harper Collins Publishers, 2020.

4 See Joint Submission of Antitrust Economists, Legal Scholars, and Practitioners to the House Judiciary Committee on the State of Antitrust Law and Implications for Protecting Competition in Digital Markets, May 15, 2020, available at https://gai.gmu.edu/wp-content/uploads/sites/27/2020/05/house_joint_antitrust_letter_20200514.pdf.

5 See UK Report Digital Competition Expert Panel (2019), “Unlocking digital competition,” available at https://www.gov.uk/government/publications/unlocking-digital-competition-report-of-the-digital-competition-expert-panel; the EU report “Competition Policy for the digital era,” available at https://ec.europa.eu/competition/publications/reports/kd0419345enn.pdf; the U.S. Stigler Center report, Stigler Committee on Digital Platforms, 2019, available at https://research.chicagobooth.edu/-/media/research/stigler/pdfs/digital-platforms—committee-report—stigler-center.pdf?la=en&hash=2D23583FF8BCC560B7FEF7A81E1F95C1DDC5225E&hash=2D23583FF8BCC560B7FEF7A81E1F95C1DDC5225E.

6 See the EU report “Competition Policy for the digital era,” available at https://ec.europa.eu/competition/publications/reports/kd0419345enn.pdf.

7 I. Graef, differentiated treatment in platform-to-business relations: EU competition law and economic dependence, Yearbook of European Law, Vol. 0, No. 0 (2019), pp. 1-52.

8 See the EU Google Shopping case (AT.39740 Google Search (Shopping), June 27, 2017), and probably the EU Amazon case (AT.40462 Amazon, 17 July 2019), and the Italian Google case (Case A529).

9 Case C-525/16 MEO, ECLI:EU:C:2018:270, para 34-36. See also Robert O’Donoghue, ‘The Quiet Death of Secondary-Line Discrimination as an Abuse of Dominance: Case C-525/ 16 MEO’, (2018) 9(7) Journal of European Competition Law & Practice, 443.

10 Among others, at European level see Cases AT.39740 Google Search (Shopping), June 27, 2017; AT.40462 Amazon, July 17, 2019; at national level see: (i) Italian cases A528—Amazon: investigation launched on possible abuse of a dominant position in online marketplaces and logistic services’ (Press release, April 16, 2019) https://en.agcm.it/en/media/press-releases/2019/4/Amazon-investigation-launched-on-possibleabuse-of-a-dominant-position-in-online-marketplaces-and-logistic-services; A529—Google: investigation launched against Google for alleged abuse of a dominant position’ (Press release, May 17, 2019) https://en.agcm.it/en/media/pressreleases/2019/5/ICA-investigation-launched-against-Google-for-alleged-abuse-of-a-dominantposition; (ii) Dutch Apple case (see ACM press release, ACM launches investigation into abuse of dominance by Apple in its App Store, available at https://www.acm.nl/en/publications/acm-launches-investigation-abuse-dominance-apple-its-app-store.

11 The EU Report, for instance, has proposed flipping the burden to platforms to show that self-preferencing has ‘no long-run exclusionary effects’ and ‘either the absence of adverse effects on competition or an overriding efficiency rationale’.

12 B. Vesterdorf, Theories of self-preferencing and duty to deal – two sides of the same coin, 1(1) Competition Law & Policy Debate, 4 (2015).

13 Case C-7/97, Oscar BronnerGmbH & Co. KG v. Mediaprint Zeitungsund Zeitschriftenverlag GmbH & Co. KG, Mediaprint Zeitungsvertriebsgesellschaft mbH & Co. KG and Mediaprint Anzeigengesellschaft mbH & Co. KG, EU:C:1998:569.

14 N. Petit, Theories of self-preferencing under Article 102 TFEU: a reply to Bo Vesterdorf, ORBi, 2015.

15 Case AT.39740 Google Search (Shopping), June 27, 2017, para. 649.

16 See Giorgio Monti, EC Competition Law (Cambridge: Cambridge University Press, 2007), 186.

17 Similarly, on this point, see I. Graef, differentiated treatment in platform-to-business relations: EU competition law and economic dependence, Yearbook of European Law, Vol. 0, No. 0 (2019), pp. 1-52; J. Drexl, ‘Designing Competitive Markets for Industrial Data Between Propertisation and Access’ (2017) JIPITEC, 257, 283–4.

18 P. Ibáñez Colomo, Indispensability and Abuse of Dominance: From Commercial Solvents to Slovak Telekom and Google Shopping (December, 2019). Forthcoming, Journal of European Competition Law & Practice.

19 Case AT.39740 Google Search (Shopping), 27, June 2017, para. 651.

20 Ibid, para. 697-705.

21 The importance of considering both the ex-ante and ex-post dimensions of competition is emphasised in E. Elhauge, ‘Defining Better Monopolization Standards’ (2003) 56 Stanford Law Review 253.

22 European Commission, Shaping Europe’s Digital Future, February 2020, available at https://ec.europa.eu/info/sites/info/files/communication-shaping-europes-digital-future-feb2020_en_4.pdf.

23 F. Cunningham, Europe’s digital roadmap indicates two-pronged approach to platform economy and competition rules, available at https://www.twobirds.com/en/news/articles/2020/global/europes-digital-roadmap-indicates-two-pronged-approach-to-platform-economy-and-competition-rules.

24 Regulation (EU) 2019/1150 of the European Parliament and of the Council of June 20, 2019 on promoting fairness and transparency for business users of online intermediation services [2019] OJ L 186/57.

25 Regulation 1150/2019 only offers some provisions to protect business users. See: (i) Article 3(2), which imposes a notice period before a platform can modify its terms and conditions; (ii) Article 4, which mandates platforms to provide a statement of reasons when restricting, suspending, or terminating its services to a business user; (iii) Article 4, which offers business users the opportunity to obtain clarifications within the framework of an internal complaint-handling procedure regarding the facts and circumstances that caused the restriction, suspension, or termination of the platform’s service (a procedure which Article 11 makes obligatory, except for small enterprises).

26 See Bundeskartellamt’s press release, 17-07-2019, Bundeskartellamt obtains far-reaching improvements in the terms of business for sellers on Amazon’s online marketplaces, available at https://www.bundeskartellamt.de/SharedDocs/Meldung/EN/Pressemitteilungen/2019/17_07_2019_Amazon.html.

27 See Austrian Federal Competition Authority report, Amazon.de Marketplace, Available at https://www.bwb.gv.at/fileadmin/user_upload/Fallbericht_20190911_en.pdf.

28 Among the most interesting achievements of such proceedings, noteworthy are: (i) Amazon’s withdrawal of Luxembourg as the exclusive jurisdiction for EU claims, thus allowing sellers to act against Amazon in their domestic courts; (ii) Amazon’s dropping of its so-called ‘parity-requirement clauses’ under which sellers had to provide products of quality as high as those used in other sale channels; (iii) Amazon’s concession to equalize the reviews of third-party-sellers with its own private-label brands, by extending to third-party-sellers its Vine reviewing program, which was initially available only for its private-label brands.

29 See OECD Workshop on regulation and competition in light of digitalisation, January 31, 2018, summary available at https://one.oecd.org/document/DAF/COMP/M(2018)4/en/pdf.

30 See J. Tirole, OECD, Global Forum on Competition, Roundtable on “Competition under fire,” Paris, 2019, available at http://www.oecd.org/competition/globalforum/competition-under-fire.htm.