The EU’s Digital Markets Act: Regulatory Reform, Relapse or Reversal?

The EU Commission’s Digital Markets Act (DMA) came into force on 7 March 2024 (Regulation (EU) 2022/1925). As part of a set of regulations proposed to make “Europe fit for the Digital Age”,¹ the DMA establishes quantitative thresholds as well as per se obligations and prohibitions for providers of online search engines, app stores or other platform services that are considered “gatekeepers” to digital ecosystems. For the moment, the DMA applies to six extra-EU companies: Alphabet, Apple, Amazon, ByteDance, Meta and Microsoft. To ensure that digital markets are becoming more fair and contestable, and to avoid legal fragmentation, unwarranted lenience and needless delays, DMA regulations complement existing EU competition rules but, unlike these rules, are entirely centrally enforced. Also, there are significant penalties to guarantee compliance. For some, the EU is seen as having finally set a new standard for how the digital economy should be regulated and function worldwide.²

For others, however, the DMA amounts to an intentional distraction of global market leaders and an infringement on their property rights in order to protect less efficient European competitors. What is more, the regulation is seen to override consumer preferences and welfare benefits and to be built on the assumption that digital markets inevitably fail (Suominen, 2022). It therefore may not be surprising that the DMA does not permit any effect-based economic assessments of specific cases, trade-offs and efficiency defences but instead establishes a set of preventive norms grounded in some vague notion of fairness and contestability. Seen this way, the DMA departs from recent advances of competition law and returns to earlier, form-based and excessively discretionary regulations; it also furthers the centralisation of these controls, undermining member states’ ability to enforce national laws with likely counterproductive consequences. Moreover, there is concern that the EU’s regulatory reform could inspire “copycat legislation”, leading to a worrisome global convergence of pertinent competition standards.

The first section of this article lists the dos and don’ts of the DMA as well as the reaction of various stakeholders to them. The next section highlights the EU’s current policy reversal in dealing with the potential abuse of dominant positions in line with Article 102 TFEU. The paper then reviews economic aspects of digital platforms, concerns about market foreclosure and essential facility as well as appropriate remedies. Finally, there is an examination of the challenges in stipulating and enforcing efficient rules to govern digital platforms.

The Digital Market Act: Context, content and
commentary

The Economist’s (2018) article on “Taming the Tech Titans” foretold the regulator’s current dilemma: if tech titans may use their power to protect and extend their dominance to the detriment of consumers, how can one restrain them without undermining consumer benefits and stifling innovation? Would utility-type regulation or a stricter enforcement of competition standards provide the answer?

Today our understanding of digital platforms may be more refined, but we are still unable to agree on a proper regulatory approach. Search engines, social media networks and marketplaces reduce transaction costs and increase price transparency; they expand consumption, production and sharing opportunities at little or no monetary cost. Digital platforms create user benefits through network effects driven by economies of scale and scope. And yet, although their data aggregation and reuse allow them to capture only part of the total value that they create, platforms quickly appear to be massively profitable, formidable gatekeepers that impose barriers to potential and actual competition. As a result, markets may tip, and the winner could take all.

But is this really so? And if yes, is the outcome the result of users’ preference for the existing, more attractive infrastructure? Or is it due to some foreclosure or the immobility of network users that limits market opportunities for followers? When is increasing the attractiveness of a platform’s offer at a larger scale and scope inadvertently limiting entry opportunities – and into which markets? Should such conduct be restricted for the sake of less efficient followers? Should key elements of platforms such as core data or algorithms be considered an essential facility and freely open for public use? Are companies obliged to help their competitors? Is the purpose of regulatory intervention the protection of competition or competitors? How should competing for and competing within a standard be separated? How does one distinguish having a competitive advantage from abusing a dominant position?

Discussing a closely related competition policy concern, vertical restraints, Nobel laureate Jean Tirole (1988, 188) suggested that

(t)heoretically, the only defensible position (…) seems to be the rule of reason. Most vertical restraints can increase or decrease welfare, depending on the environment. Legality or illegality per se thus seems unwarranted. At the same time, this conclusion puts far too heavy a burden on the antitrust authorities. It seems important for economic theorists to develop a careful classification and operative criteria to determine in which environments certain vertical restraints are likely to lower social welfare.

This is what the EU Commission tried to do with its 1999 competition policy reform. It first established a list of per se unacceptable contract clauses and a market share limit and then advised companies above that limit to prepare an “efficiency defence” that justified any of the non-listed clauses used. In addition, third parties were incentivised to monitor company behaviour, and penalties were increased. Hence, the 1999 reform pioneered the restructuring of EU competition control from ex ante authorisation to strengthened ex post control. The transformation reflected the realisation that contractual complexities limited the extent of centralised regulatory fine-tuning, which instead needed to be complemented or substituted by ways of reliably outsourcing regulation to those closest to the case at hand (Boscheck, 2000). The DMA rejects this logic and falls back on top-down rules.

DMA Article 1(1) reads:

The purpose of this Regulation is to contribute to the proper functioning of the internal market by laying down harmonized rules ensuring for all businesses, contestable and fair markets in the digital sector across the Union where gatekeepers are present, to the benefit of business users and end users (Regulation (EU) 2022/1925).

Article 3 of the DMA identifies gatekeepers as online platforms that, for the last three years, had a strong and stable position in linking large user bases and a significant number of companies across EU countries.³ Article 4 affirms that the gatekeeper status should be reviewed every three years. Article 5 lists obligations for gatekeepers to ensure interoperability with third parties: provide access to key data in order to allow third parties to generate their own gatekeepers’ platform; provide tools and information for companies to independently verify the performance of their advertisement on the gatekeepers’ platform; allow business users and consumers to transact outside the gatekeepers’ platform; ensure non-discrimination between gatekeepers’ and third parties’ offerings; allow users to remove any pre-installed software or apps; refrain from tracking not-consenting end users outside of the gatekeepers’ core platform services for the purpose of targeted advertising. Articles 9 and 10 do not allow for an efficiency defence; the Commission insists that “a distortion of fairness cannot be compensated by economic efficiency.”⁴ Noncompliance will result in fines of up to 10% of the company’s total worldwide annual turnover, or up to 20% in the event of repeated infringements. If infringements are considered systematic, behavioural and structural remedies, i.e. the divestiture of (parts of) a business may be imposed. Article 12 states that the Commission may pass delegated acts to update existing obligations; according to Article 19, the Commission may undertake market investigations into novel services to initiate new legislation.

Various analysts and lobbyists have commented on the rationale and the direction of the EU’s shift in policy, its procedural and economic impact as well as any risk of over- or under-regulation (see e.g. CompassLexecon, 2021; Ribera Martìnez, 2024; Hoffmann, 2023). Many see the EU’s policy reform motivated by a realisation that competition law had achieved too little, too late in dealing with the digital transformation of markets. The necessary economic analyses seemed to have resulted in enforcement of competition standards that was too slow and cautious without delivering suitable remedies. At the same time, platforms were seen to increasingly bias consumer preferences, which thus could no longer be trusted to deliver proper market responses. For these reasons, and given the worldwide scope of platform operations, the EU also intends to embed DMA-style ex ante regulation as the default position globally, forging Digital Partnership Agreements with Korea, Japan and Singapore. Referencing these efforts, India’s Parliamentary Standing Committee on Commerce proposed a DMA-style ex ante regulatory model and Australia’s Competition and Consumer Commission is working on binding codes of conduct for digital platform services. In the US, the Federal Trade Commission’s and the Department of Justice’s appreciation of the need for stricter, possibly DMA-type, competition controls was rejected by the US Department of Commerce and the US State Department, as expected. Similarly, competition authorities in Brazil, Taiwan and the UK remained sceptical about ex ante regulation.

While most critics voice concern about the lack of transparency, unavoidable conflicts with pre-existing national regulations and the higher likelihood of capturing centralised controls, they mainly object to the replacement of a welfarist economic approach and, with it, the elimination of the need to present a robust theory of harm. Apparently guided by the EU’s experience in regulating the telecom sector, the DMA attempts to police a (non-existent) digital market by imposing restrictions on crudely grouped but very different platforms that offer dissimilar services in different sectors with quite distinct business and monetisation models. It is an attempt to avoid the “hassle” of having to first define relevant markets, then establish platform dominance in these, and finally demonstrate abuse. Gatekeepers are simply identified in terms of size, which per se is not a sign of dominance. The presumption of tipping markets is analytically flawed; the notion of “winner takes all” is empirically wrong (see below). Similarly, Articles 5 and 6 are but a muddled listing of restrictions derived from past or ongoing investigations of specific business practices, which, quoted “out of context”, are established as general rules to be indiscriminately applied. The presumption of illegal conduct is irrebuttable, although procompetitive effects of some of the admonished behaviours are well established (Boscheck, 2000). Finally, the potentially negative economic impact of the EU regulatory reform is projected to amount to billions of dollars – mainly due to new compliance and operational costs for digital service suppliers, their business and individual customers as well as a lack of product and service innovation (Suominen, 2022). As an offshoot, Europe is apt to see a compounding of its inflationary pressures and increasing technological divide separating it from the rest of the world.

Towards “a dynamic and workable effects-based
approach to abuse of dominance”?⁵

The Commission’s background paper on its Amendments to the 2008 Guidance on Enforcement Priorities, published on 27 March 2023, provides some clues to answer this question. It starts off by describing the original Guidance as vital in

moving away from a formalistic approach to enforcing Article 102 TFEU, where cases were prioritized based on per se criteria, to an effects-based approach where priorities are set taking into account the potential effects of the given conduct, through the analysis of market dynamics, in line with mainstream economic thinking (McCallum et al., 2023, 2).⁶

But soon thereafter, the paper calls for change.

Reflecting upon a large number of decisions made by Union Courts⁷ and the Commission since 2008, and in light of the growing importance of digital markets with supposedly “winner takes all” characteristics, the Commission feels compelled to ensure an effective and fast enforcement of competition principles “before tipping occurs and entrenched market positions are created” (McCallum et al., 2023, 2). It hence endeavours to replace existing guidelines and avoid that “an overly rigid implementation of the effects-based approach (…) set the bar for intervention at a level that would render enforcement against practices that restrict competition unduly burdensome or even impossible” (McCallum et al., 2023, 4). The Commission’s finding, presented in this policy brief, however, is rather worrying: evidentiary standards are diluted, conduct is considered objectionable if probable rather than profitable; the impact on competition is extended from “as efficient” to “not yet as efficient” competitors.

The Commission’s interpretations of Union Court decisions in TeliaSonera, Google Shopping or Post Danmark II and others suggest that to find abuse, a dominant company’s conduct must merely have the potential to have an anticompetitive effect.⁸ The profitability or outcome of an alleged behaviour is also deemed irrelevant. Instead, what is to be considered is when “the dominant undertaking adversely impacts an effective competitive structure” (McCallum et al., 2023, 5). But it is not clear what may constitute such a structure. In judging pricing abuses, the Commission wants “to avoid an unduly strict and dogmatic application” of the “as efficient” competitor standard and broaden it to capture “not-(yet)-as efficient” competitors “to allow them to gain a foothold in the market, with the prospect of scaling up volumes and potentially increasing at a later stage” (McCallum et al., 2023, 5). As this conflicts with the insistence of Union Courts that “EU competition policy should not ensure that competitors less efficient than the undertaking with the dominant position should remain on the market” (McCallum et al., 2023, 5), the Commission had to find a solution. It discovered it in Unilever, Post Danmark II and Google Android, where the Courts considered efficient competitor tests (AECs) as “optional”, “one tool amongst others” or “only one of several factors”.⁹ Still, “the fact that a dominant undertaking’s pricing conduct ‘passes’ an AEC test should not be considered as a conclusive indication that such pricing conduct is not capable of negatively affecting competition” (McCallum et al., 2023, 7). As a result, companies are left without any guidance. Finally, in assessing constructive refusals, supply conditions and alleged situations of margin squeeze, the Commission considers the case law of the Union Courts to show that the indispensability of the product or service in question, part of the so-called Bronner criteria,¹⁰ does not need to apply.

Clearly, the Commission’s U-turn results in anything but “a dynamic and workable effects-based approach to abuse of dominance.”¹¹ Substituting presumptions for establishing effects, it boosts the Commission’s discretion and enforcement agenda at the cost of regulatory certainty and adequacy. The competition policy brief clarifies the reasoning behind the DMA and its censure of economic analysis.

On digital platforms, abuse of dominance and remedies

Digital platforms are markets in which the market-maker connects a variety of groups of users, offers additional products or services or tracks and commercialises user information. The attractiveness of these markets reflects the quality of both offers and users and the network benefits that affect each. Strong network effects, economies of scale and complementarities on the demand-side and supply-side may skew inter-platform competition to the benefit of a major platform (for an early discussion of this, see Katz and Shapiro, 1994). The risk of such “tipping” is mitigated by users utilising several competing platform services in parallel (so-called multi-homing), segmentation opportunities on each side of the market, negative network effects as well as technological disruptions. Conversely, strong positive network effects, proprietary user insights or valuable bundles of services may not only put followers and smaller players at a competitive disadvantage, but also enable the leading platform to abuse its dominant position. To avoid any presumption, however, relevant markets must be defined, and dominance and abuse established.

Markets typically consist of substitutes of different degrees. Facebook and Google compete in digital advertising, which as a category competes with print media and billboards. Match Group, the dominant aggregator of dating sites competes with a huge competitive fringe catering to the special dating interests of the LGBTQ community (Grindr), older individuals (Our Time), Jews (J-Date) or fans of spicy food (Hotsaucepassions) (Hovenkamp, 2021, 1999). Operating a so-called cluster market, Amazon uses its platform to offer non-competing goods or services, some of which one also finds in Walmart as well as at speciality stores. As platforms are unlikely to eliminate diverse businesses as long as user preferences value product differentiation, winner-take-all outcomes are improbable. Unfortunately, courts and regulatory authorities on both sides of the Atlantic still do not understand this. In Ohio v. American Express Co., the Supreme Court “lumped production complements into the same market, and in the process, it stymied coherent economic analysis of the problem” (Hovenkamp, 2019, 47). The EU’s DMA singles out gatekeepers irrespective of their market contexts and bars any economic assessment.

Still, scale and network effects, lack of product differentiation, or switching costs resulting from a lack of interoperability or the superiority of an incumbent’s product offerings may render that platform dominant, possibly even naturally monopolistic. Here, the best regulatory response would be a fact-based assessment of alleged anticompetitive conduct, such as the recent complaints against Facebook, Google and Apple related to exclusivity, tying, self-preferencing and default practices under the US Sherman Act (see Scott Morton and Dinielli, 2020)^. or the EU competition standards (prior to the EU’s reinterpretation discussed above). But it is important to consider the impact of diverse treatments of presumably foreclosing assets – so-called essential facilities – on the incentives to create, sustain and develop platform businesses.

The US Supreme Court for long has avoided essential facility reasoning, simply because the underlying doctrine raises more questions than it is able to answer. Areeda (1990, 841), for one, could not “find any case that provides a consistent rationale for (…) requiring the creator of an asset to share it with a rival”. For Hovenkamp (2021), “it is a detrimental doctrine that should be dismantled”. Marquardt and Leddy (2003, 848) quote US Antitrust Guidelines related to licensing, which expressly states that “(a)s with any other tangible or intangible asset that enables its owner to obtain significant supra-competitive profits, market power (or even monopoly) that is solely ‘a consequence of a superior product, business acumen or historic accident’ does not violate the antitrust law.” In addition, Marquardt and Leddy (2003) point out that any ill-defined concern for a proper business justification opens a futile debate about what constitutes legitimate strategic intent. A refusal to deal is by definition exclusionary and inspired by the desire to impose a relative disadvantage on a rival; in fact, even “(t)he desire to crush a competitor, standing alone, is insufficient to make out a violation of the antitrust laws” (Ocean State Physicians Health plan, Inc. v. Blue Cross & Blue Shield, 1989).

By comparison, the EU asserts as a general principle an “objectively unjustifiable” refusal to supply by an undertaking holding a dominant position on a market breaches EU competition rules. Unlike in the US, there is no doubt in the EU that a refusal to license may raise antitrust liability, and the doctrine of essential facility may be applied to intellectual property rights (see Peeperkorn, 2003; Elhauge, 2003). The EU suggested an “as efficient competitor” test – asking whether a dominant firm itself would survive its own exclusionary conduct.¹² As discussed above, the Commission now extends the impact test to “not yet efficient competitors.” This not only invalidates any attempt to gain a first-mover advantage or win a technology or standard-setting race, but it misconstrues competition to mean cooperation to help the runner-up.

Finally, the DMA foresees that systematic infringements will trigger structural remedies, i.e. the divestiture of (parts of) a business. But the social costs of such an imposition can be high. Platforms may lose economies of scale and scope; consumer choice and network benefits may be eliminated without even reducing market power in a particular product or service (see Hovenkamp, 2021, 2017-2019). For instance, obliging Amazon to spin off its ebook business may merely shift the company’s market power to the acquirer but cause consumers to incur higher search costs. Conversely, curtailing a platform’s business development for fear of leverage may foreclose opportunities for creating synergistic efficiencies and network effects. Following that logic, Amazon could be denied participation in streaming services, and Microsoft may be barred from expanding into computer hardware. Unfortunately, recently proposed alternatives to divestitures do not provide a convincing solution either (Hovenkamp, 2021, 2020-2032). Spreading out platform ownership among its main users, effectively creating a common, will replace suspected unilateral decision-making with more cumbersome multilateral dealings among the platform owners. Never mind the costs and time spent on coordinating cooperative outcomes under these conditions, would success not mean replacing a monopolist with an effective cartel? Similarly, mandating interoperability or pooling vital information among multiple firms constituting a platform is likely to discourage the creation of platforms ex ante and result in anticompetitive conduct in dealing with haggling and hold-up problems ex post. So, what is one to do?

Avoiding relapse: Towards an efficient rule for dealing with digital platforms

An efficient administration of competition rules must minimise the sum of two types of interrelated costs: enforcement costs incurred in assessing a case including the uncertainty and time involved and its effect on behaviour, and the costs of permitting (prohibiting) efficiency reducing (increasing) clauses. Enforcement costs vary depending on the quality of the enforcement processes; the costs of wrong decisions differ in line with the value of foregone alternatives and the extent of welfare distortion until removed. Clearly, one would expect that the closer a rule approximates the specifics of a case at hand, the higher the enforcement costs, and the lower the costs of a wrong decision. Conversely, simplifying rules by aggregating business relationships based on some shared characteristics is efficient to the extent that reductions in enforcement cost more than compensation for the likely increases in the costs of wrong decisions.

Dealing with digital platforms, the fact that both types of costs are potentially very high invalidates any per se ruling as proposed by the DMA. By the same token, a rule of reason must be excluded. As an alternative, Condorelli and Padilla (2020) propose considering conduct where enforcement would clearly result in net welfare losses or net welfare gains as legal or illegal per se respectively. Conduct with a high likelihood of a net welfare loss should be considered illegal, but the dominant firm should be given an opportunity to demonstrate efficiency effects. In the reverse case, the conduct should be considered pro-competitive, but regulators could prove there are detrimental effects (Condorelli and Padilla, 2020). While it seems that their recommendation returns to the principles spearheaded by the EU’s Competition Policy Reform of 1999, it does not establish a clear black list of per se unacceptable conduct nor does it provide a market share limit to assess and contain the potential welfare impact of that conduct and determine the responsibility for challenging that finding. It offers an approach to thinking about the problem but no blueprint for advancing a regulation that would be able to combine the benefits of economic evaluation with concerns for legal certainty and enforcement efficiency. Again, the original Commission Regulation (EC) No 2790/1999 of December 1999 on vertical restraints offers some pointers worth reconsidering.

The Regulation responded to the lack of substantive policy guidance, symptomatic for the prior legal treatment of vertical restraints in both the US and the EU, by changing procedural rules to delegate the assessment task and the choice of tools to contractors themselves. The Commission merely defined a “clear” standard to screen cases for likely negative impact and offered a framework that firms were able to take as guidance in reviewing their arrangements and building an efficiency defence. The new regulation offered authorities and companies a truncated rule of reason that helped to limit the cost of wrong decisions and improved the efficiency of enforcement. On a broader level, the regulation pioneered the restructuring of EU competition control along three vital building blocks: clear, or better recognisable and cogent, legal standards; a strengthened system of ex post control; and corporate self-assessment. The DMA is lacking all three.

Summary and conclusions

Starting 7 March 2024, the EU Commission began enforcing its controversial Digital Markets Act. This article sheds light on the recent debate. It lists the dos and don’ts of the DMA as well as stakeholder reactions to them before tracing some of the underlying motivations and reasoning that emerge from the EU’s current policy reversal with reference to dealing with abuse of dominant positions in line with Article 102 TFEU. Following this, the paper reviews economic aspects of digital platforms, concerns about market foreclosure and essential facility as well as appropriate remedies before closing with a discussion of the challenges in stipulating and enforcing efficient rules to govern digital platforms. It is argued that the DMA ought to be rewritten to stipulate clear and cogent legal standards, return to a strengthened system of ex post control and provide platforms with the opportunity to deliver efficiency defences as part of corporate self-regulation.

• 1 For an introduction and overview of the EU’s Digital Service Act, Artificial Intelligence Act, Data Act, European Cloud Service Scheme, the regulations linked to European Telecommunications Network Operators, see https://commission.europa.eu/strategy-and-policy/priorities-2019-2024/europe-fit-digital-age_en.
• 2 As projected by the rapporteur from Parliament’s Internal Market and Consumer Protection Committee, Andreas Schwab, quoted in Press Release (European Parliament, 2022). For a recent call for gatekeeper cooperation, see https://eutechalliance.eu/wp-content/uploads/2024/01/Public-statementcalling-on-Gatekeepers-to-cooperate-16-Jan-2024pdf.pdf.
• 3 A significant impact on the EU market requires “an annual Union turnover equal to or above EUR 7,5 billion in each of the last three financial years, or where its average market capitalisation or its equivalent fair market value amounted to at least EUR 75 billion in the last financial year, and it provides the same core platform service in at least three Member States” (Regulation (EU) 2022/1925, Article 3(2a). It is a core platform, which means that in the last financial year it had at least 45 million monthly active end users established or located in the Union and at least 10,000 yearly active business users established in the Union.
• 4 See Digital Markets Act, recitals 10 and 11.
• 5 McCallum et al. (2023).
• 6 The lead author of the paper, Linsey McCallum, is Deputy Director-General for Antitrust in the EU Directorate-General for Competition.
• 7 The Court of Justice and the General Court.
• 8 Case C-52/09 – TeliaSonera; Case T-612/17, Google Shopping; Case C-23/14 – Post Danmark II.
• 9 Case C-680/20 – Unilever, paragraph 62; Case C-23/14 – Post Danmark II, paragraph 61; Case T-604/18 – Google Android, paragraph 643.
• 10 Judgment of 26 November 1998, Oscar Bronner v. Mediaprint, Case C-7/97, EU:C:1998:569, paragraph 41.
• 11 McCallum et al. (2023).
• 12 In 2004, EU fined Microsoft €497 million for abusing its dominant position and refusing to supply information required to ensure the inter-operability between its Windows and competing operating systems and for bundling its Media Player with the Windows software. The company was ordered to offer an unbundled version of Windows to PC makers. According to the ECJ’s judgment on IMS Healthcare, a firm can be ordered to license its IP if the IP is indispensable for carrying on a particular business, where competitors do not merely duplicate existing products and where market demand is sufficient.