INTERNATIONAL: INTERNATIONAL FORUMS - COMPETITION POLICY - OECD - DIGITAL MARKET - REFORM

Digital antitrust: An emerging consensus?

An increasing number of jurisdictions are considering adjustments to their competition policy framework to reflect the unique challenges of digitalisation. Indeed, several studies have been formally commissioned on the subject. While their legal context, focus and findings vary significantly; some key themes have begun to emerge from these studies. They include proposals for expanded merger notification and review processes, streamlined competition authority procedures, and even new regulatory authorities. This article seeks to draw out the common elements that can chart a way forward for digital competition policy.

1. Digitalisation has become a dominant theme in competition policy discussions of late. This is reflective of the growing importance of digital firms in the economy: seven of the largest ten companies in the world by market capitalisation are tech companies. [1] It is also due to the fact that digital markets feature characteristics, which may not on their own be new, but which together pose novel challenges for competition policy.

2. Healthy debate about whether competition policy must be adapted to new digital realities has arisen in academic literature, conferences, media commentary, international forums such as the OECD, and within governments as well as competition authorities. Observers of competition regimes around the world will be familiar with the fact that, while there is widespread agreement on the fundamentals, there remain differences of opinion in terms of the interpretation of market phenomena, the thresholds for antitrust intervention, and the most appropriate remedies to potential harm, among others. The debate about digital competition is no different—the protagonists have so far agreed on the substantial benefits of digitalisation, but may have differed on whether competition harm is emerging that requires intervention.

3. Competition policymakers have sought to chart the way forward: expert panels have been commissioned by the EU, [2] Germany, [3] Japan [4] and the UK, [5] while competition authorities in Australia, [6] Canada, [7] France and Germany, [8] Italy, [9] and Portugal, [10] among others, have conducted studies of their own. In the US, the FTC held a series of hearings on a wide range of topics relevant to digitalisation. [11] Further, the G7 competition authorities issued a “Common understanding” in June this year, setting out their collective views on the opportunities and challenges of digitalisation for competition policy. [12] While the legal context, focus and findings of each of these efforts vary significantly across jurisdictions, as with many other antitrust topics, several key themes have begun to emerge. These common threads can be used to guide further research, studies by competition authorities, and the formulation of policy and legislative proposals moving forward.

4. The first common theme in many of the assessments listed above is a recognition of the substantial benefits of digitalisation. For consumers, digitalisation represents a source of new or cheaper products, reduced intermediation costs, and improved access to information about the purchases they make. For the economy more broadly, digitalisation holds substantial promise in terms of productivity and growth. Thus, the market changes of digitalisation are not really a problem to be solved, but rather a significant economic opportunity.

5. While there appears to be consensus on the promise of digitalisation, several reports touch on the more controversial subject of whether competition enforcement currently reflects an optimal balancing of error costs—that is, balancing the risk of under-enforcement in terms of competition harms, with the risk of over-enforcement that may hamper the manifestation of these benefits. Specifically, reports commissioned by the EU [13] and UK [14] (and echoed by the report of a US Committee for the Study of Digital Platforms, the “Stigler Report” [15]) suggest that the current balance errs too far in the direction of under-enforcement.

6. The second common theme is the nature of competitive dynamics in digital markets. In particular, there is broad recognition that digital products can feature low variable costs and, at least in some cases, substantial fixed costs. Network effects are essential drivers of consumer welfare, and data both collected from users and generated from their activities constitutes an important competitive asset. Data can also give rise to self-reinforcing improvements in service quality and variety as well as firm revenue. Innovation is essential for enhancing products, adding features, and entering related markets to take advantage of economies of scope. The importance of digital platforms means that there are relatively more multi-sided markets and zero-priced products than in the past. The implication of these characteristics in many markets is that firms compete for the market, sometimes referred to as “winner takes most” dynamics.

7. In the competition community more broadly, there is disagreement about how likely these dynamics are to result in competition problems. Competition “for the market” dynamics need not be a consumer welfare problem if the winning firm will face competitive constraints, and could be replaced by competitors offering better products—big may not automatically be bad, in other words. However, some are questioning whether this process is occurring in digital markets. For example, the UK Digital Competition Expert Panel (the “UK Report”) expressed a concern that firms which successfully compete for the market are enjoying more durable, less contestable market power than in the past. Further, with economies of scope, new entry in related markets may be limited. The Stigler Report highlighted the risk that increasing barriers to entry and expansion may be preventing the emergence of new large digital players.

8. More evidence may be needed to draw firm conclusions on whether the contestability of digital markets is lower than other concentrated markets in the past, whether dominant digital firms continue to face pressure to maintain their position due to the threat of competition, and whether there are consumer welfare or broader economic costs to the current patterns of churn in the tech sector. However, recent research by the OECD and others has found some warning signs. [16] Mark-ups, typically used by economists as a measure of market power, are on the rise in advanced economies, and have grown relatively quickly in digital-intensive sectors. [17] At the same time, the entry rate of new firms into the economy is falling. [18] These phenomena could have both troubling explanations (for example, markets are becoming less contestable than before) and benign ones (the composition of the economy is simply changing).

9. The third common theme to emerge is that there are some broad competition policy principles for which there appears to be some degree of consensus. One such principle is the importance of case-by-case analysis for merger, abuse of dominance and vertical restraint cases. Another is the need to avoid attempting to fit all digital concerns into a competition law enforcement framework, and recognise the role for consumer protection and sector regulation in some cases. The importance of competition authority co-operation across national borders, and with other regulators, is also commonly noted. Finally, the usefulness of market study or market investigation tools to assess conditions in some digital markets is also often recognised.

10. Beyond these broad themes, however, are specific proposals for changes to competition policy that have elicited more controversy. The sections below will highlight some of the common proposals made in several recent reports (primarily the Australian, [19] EU, UK and Stigler reports), as well as their differences, in order to identify possible opportunities for consensus on digital competition policy reform; namely:

  •  Establishing a new digital regulator
  • Developing new enforcement approaches
  •  Ensuring anticompetitive digital sector mergers are notified and assessed
  • Streamlining competition authority procedures
  • Enhancing our understanding of consumer behaviour

A Table is available in the pdf version of this article.

I. Establishing a new digital regulator

11. The first set of proposals made in several reports reflect a recognition that there may be limits to what competition enforcement can achieve alone in digital markets, and that enhancements to competition policy more broadly may be necessary. In particular, these proposals suggest that an entity focused on digital markets with explicit regulatory powers is needed to complement the current competition policy framework.

12. The UK and Stigler reports propose that some regulatory provisions under this new model would apply only to a specific set of platforms with strategic market status, or bottleneck power. While the specific definition would need to be defined (the Stigler Report suggests this determination power should be granted to the digital regulator [20]), this situation would generally arise if a platform has market power and acts as a gatekeeper, controlling the access of different sides of a market to one another. Contributing factors to this power could include high barriers to entry, substantial switching costs, and a tendency by consumers to use a single platform at a time (as opposed to multi-homing). By applying only to a subset of firms, the measures seek to address concerns about durable market power while avoiding the imposition of a regulatory burden on new entrants which may hamper competition.

13. Data would be a key area of focus for a new regulator (or potentially new powers for an existing regulator). Given the importance of data as an asset in digital markets, and thus as a potential barrier to entry, several reports propose that consumers should be able to move their data among firms (data portability), and that standards for that data could ensure it is compatible across systems and firms (data interoperability). In other words, these measures seek to promote market contestability by ensuring that firms compete based on current offerings, rather than based on which firm has had the most consumers in the past and thus the most data. Requirements may vary based on whether a firm has strategic market status or not. For instance, the EU Report proposes imposing specific data portability requirements on dominant firms where there are substantial “lock-in effects” for consumers, using the current EU data protection framework. [21] In addition, the Report suggests that further regulatory measures to enhance data portability may need to be applied via specific sector regulation in order to enhance competition for complement and secondary markets. [22]

14. There are some examples of these efforts already underway. In the UK, the CMA has sought to promote competition in the retail banking sector by requiring firms in that sector to develop and implement “Open Banking” standards (for data, security, and application programming interfaces [23]). Further, some of the largest digital companies (Apple, Facebook, Twitter, Microsoft and Twitter) have also launched a project on their own initiative to promote data portability for consumers (the Data Transfer Project [24]), and Mark Zuckerberg has published an opinion article calling for regulation in this area. [25]

15. The UK Report suggests that a digital regulator may need to go beyond data portability and interoperability when “opening up some of the data held by digital businesses and providing access on reasonable terms is the essential and justified step needed to unlock competition.” [26] For example, some data may be generated from user activity, aggregated and anonymised, and may thus not be covered under data portability requirements that focus on data explicitly provided by a user (e.g., user profile information).

16. Significant conceptual challenges may arise in the application of regulations requiring certain firms to provide data to competitors, in particular determining when such data is essential (since multiple datasets can be used for a similar purpose [27]). The results of the data portability efforts described above may be informative in determining how significant the entry barrier posed by data truly is, and whether improving firms’ access to this data would increase the number of competing firms in a market. Further, data privacy risks from these efforts may need to be managed. The magnitude of these challenges is reflected by the UK Report, which emphasises that data access measures should be subjected to careful analysis and applied in limited circumstances only. The UK Competition and Markets Authority (CMA) has indicated it is considering a market study into digital advertising which could be used to inform the establishment of a digital regulator. [28]

17. The Stigler Report indicates that there may be cases in which concerns about data and competition can be addressed through less intrusive efforts. In particular, the Report suggests that a new digital regulator could use behavioural “nudges” to, for example, encourage consumers to think more carefully about the data collection they agree to, since the acceptance of defaults could enhance incumbency advantages. [29] Other potential digital regulatory measures identified in the Report include unbundling and non-discrimination requirements to further address exclusionary conduct by firms without a consumer welfare justification.

18. In a more limited proposal, the Australia Report recommends the establishment of a “specialist digital platforms branch” within the Australian Competition and Consumer Commission (ACCC) responsible for, among other things: “proactively monitoring and investigating instances of potentially anti-competitive conduct and conduct causing consumer harm by digital platforms, which impact consumers, advertisers or other business users (including news media businesses).” [30]

II. Developing new enforcement approaches

19. Another set of proposals pertains to competition law enforcement, and particularly abuse of dominance cases. The EU, UK and Stigler reports all identify measures for more rigorous enforcement of competition law to make up for what could be gaps or insufficient efforts so far.

20. The importance of vigorous enforcement of existing competition law standards is discussed in several reports. For instance, the EU Report highlights the potential harm of vertical restraints, such as most-favoured nation clauses (MFNs), and the importance of a careful case-by-case approach. It opines that narrow MFNs may not cause competitive harm in some cases, but could in the presence of competition for the market dynamics in which an online platform’s only competitors are other sales channels (e.g., hotel direct sales). [31]

21. The development of new economic research and analytical tools is identified as a key priority in the Stigler Report. The Report suggests that the current level of digital antitrust enforcement in the US is reflective of excess caution in avoiding over-enforcement, and this can only be overcome through a better understanding of the unique conditions of digital markets. Better tools are specifically needed, according to the Stigler Report, to assess: zero-price transactions, consumer welfare in the presence of exploitative content; consumer biases; innovation; potential competition (including from early-stage firms); and the optimal approach balancing welfare across different sides of a market. [32]

22. Novel approaches to established competition problems were also proposed in the Stigler and EU reports. In particular, the Stigler Report indicates that refusal to deal, predatory pricing, loyalty rebates and exclusive dealing agreements should all be assessed more closely, and perhaps with more scepticism than before. [33]

23. The EU Report focuses more extensively on concerns about the role of platforms as regulators in a market, and their ability to leverage their position into other markets, for example by bundling or self-preferencing in terms of access to the platform. The Report notes that this conduct can be procompetitive and consumer welfare enhancing in some cases, and so a case-by-case analysis will be required to determine whether there are objective justifications. However, given the potential anticompetitive effects, the Report proposes placing the burden on dominant firms to demonstrate the efficiency effects. [34] This is partially a reflection of the challenges inherent in computing the harm to consumer welfare from some conduct—a challenge that the Report proposes be surmounted by acting on conduct with the potential to harm competition, and placing the analytical burden on dominant firms. Given the available economic literature on foreclosure concerns, demonstrating consumer benefits (e.g., through the elimination of double marginalisation) may not be an insurmountable burden, although it is not clear whether the change would have a significant effect on conduct. [35]

24. In addition, the EU Report discusses new approaches to concerns about access to data. Contrary to the UK and Stigler reports, it suggests that these concerns could be addressed under competition law standards rather than through the creation of a new regulator. In particular, the Report suggests that while the “classical” standards for the application of the essential facilities doctrine may not be met, data that is indispensable for effective competition could still be addressed through Article 102 cases. [36] Refusals to grant standardised access to data by dominant firms acting as gatekeepers to certain markets would be the focus of such efforts, per the Report. [37]

25. The EU Report also addresses adaptations to analytical tools, noting that there may need to be less emphasis on market definition in digital platform cases and more emphasis on considering the impact of conduct on multiple markets (e.g., product ecosystems and aftermarkets). [38] Similarly, because some anticompetitive conduct by platforms may effect dynamic as opposed to static competition, the Report proposes that authorities extend their analytical timeframes, and more carefully consider the effect of conduct on potential competition (based on concerns about the risks of under-enforcement).

26. To learn from past decision-making and analysis, the UK Report recommends a retrospective analysis of cases not brought by the CMA. Such assessments could help determine whether the approach taken to abuse of dominance cases should be adapted based on the evolution of markets following a decision not to take a case.

27. Finally, several reports emphasise an uncontroversial point: the need for competition authorities to have sufficient technical capacity to monitor and assess potential anticompetitive conduct in digital sectors. This is reflected in some recent announcements by competition authorities: for example, the Competition Bureau of Canada has appointed a Chief Digital Enforcement Officer, [39] and the ACCC has recommended the development of a Specialist Digital Platforms Branch to build expertise within that organisation. [40]

III. Ensuring anticompetitive digital mergers are notified and assessed

28. A commonly cited concern regarding digital markets is the risk that large incumbents may purchase nascent firms in order to prevent them from becoming competitors in the future, and that these acquisitions are not captured by revenue-based merger notification rules (or, in a more limited set of circumstances, by the UK share of supply test). [41] There is a particular concern that these acquisitions may deprive consumers of certain disruptive innovations, or the procompetitive benefits of a new entrant in the market. These concerns seem especially relevant given that merger activity has increased significantly, particularly in the digital sector. [42] The competition authorities in Germany and Austria have introduced additional thresholds based on transaction value, although it is too early to say whether this has been an effective measure. [43]

29. The EU Report recommends waiting until the effectiveness of alternative notification thresholds becomes clearer before contemplating changes at the EU level, particularly given the system for referrals from national competition authorities to the European Commission. The Report also casts doubts on the frequency with which “killer acquisitions” resulting in the elimination of potential innovations actually occur.

30. The Australia, UK and Stigler reports take a different approach to concerns about the notification of acquisitions by incumbent digital firms. These reports suggest that large digital firms (“large digital platforms” in the Australia Report, those with “strategic market status” in the UK Report, and those with “bottleneck power” in the Stigler Report) be subjected to an enhanced level of oversight for their merger and acquisition activity. Specifically:

  •  The Australia Report proposes that the ACCC agree with large digital platforms a specific transaction notification protocol indicating the types of transactions that must be notified (potentially including minimum transaction value) and the notice period for such acquisitions.
  •  The UK Report recommends that firms with strategic market status be obligated to notify all of their acquisitions to the CMA.
  •  The Stigler Report advises that its proposed digital regulator be notified of all transactions made by firms with bottleneck power.

31. In addition to changes to notification procedures, several reports also propose that specific theories of harm be developed, or emphasised, in digital sector merger reviews. The Australia Report recommends that legislative changes to the merger review law be made to better enable the consideration of dynamic competition issues as well as technology assets such as data. Specifically, the Report proposes adding as factors of assessment: (i) “the likelihood that the acquisition would result in the removal from the market of a potential competitor” and (ii) “the nature and significance of assets, including data and technology, being acquired directly or through the body corporate.” [44]

32. The EU Report calls for an increased emphasis on conglomerate theories of harm, given that dominant firms may use acquisitions of nascent firms to add to their product ecosystem (as opposed to simply discontinuing those firms’ innovations to avoid the cannibalisation of existing sales). [45] In other words, further consideration should be given to the potential for firms outside the market to exert competitive constraints on the acquiring firm in the future. This would apply in cases where the acquirer is a dominant platform or “ecosystem” with an entrenched position due to network effects or other entry barriers. [46] Acquisitions of competitors with a growing base of users could be used by dominant platforms to protect and defend their current network effects-based position.

33. This theory of harm is particularly consistent with the idea that many online platforms compete for users’ attention, or a user base, despite offering differentiated products. The Report addresses a common concern of proposals to capture start-up acquisitions in merger control—that it is a deliberate and perhaps the only exit strategy for many small firms—by emphasising the importance of assessing the acquisition relative to the counterfactual (whether the acquisition target could have grown in order to compete with the incumbent in the future). The Report refers to this approach as “inject[ing] some ‘horizontal’ elements into the ‘conglomerate’ theories of harm [47] by asking the following questions:

(i) Does the acquirer benefit from barriers to entry linked to network effects or use of data?

(ii) Is the target a potential or actual competitive constraint within the technological/users space or ecosystem?

(iii) Does its elimination increase market power within this space notably through increased barriers to entry?

(iv) If so, is the merger justified by efficiencies? [48]

34. The UK Report focuses its recommendations on: (i) prioritising the review of mergers that will have innovation or potential competition implications; (ii) modifying the UK’s merger assessment guidelines to add further discussion on multisided markets, switching costs, zero pricing, loss of future innovation and potential competition, and to signal greater openness to vertical merger theories of harm; and (iii) implementing a “balance of harms” approach. The latter proposal reflects an effort to address uncertainty when considering a merger’s harm to dynamic competition, and to alter the balance between over- and under-enforcement risks. Specifically, the Report proposes taking into account not only the likelihood of harm, but also the magnitude of that harm when reviewing merger decisions. [49] In other words, the proposal suggests that uncertain harms that will have a high potential impact should have a more determinative role. The CMA, while broadly supportive of the UK Report’s conclusions, has indicated that a “balance of harms” approach is unnecessary, and could result in challenges for practical application as well as transparency. [50]

IV. Streamlining competition authority procedures

35. The UK and Stigler reports also highlight concerns with some procedural aspects of competition law enforcement in their respective jurisdictions. They make proposals to streamline judicial review and improve the speed of enforcement interventions, both of which may be particularly important in digital markets given the complexity of the issues involved and the need to conduct a case-by-case analysis. While these proposals are specific to the framework in each jurisdiction, the underlying concerns are more broadly applicable.

36. With respect to judicial review, the Stigler Report indicates that it can be challenging for generalist judges to hear complex competition cases if they only rarely appear on their docket. Thus, the Report proposes developing a specialised competition court that can become sufficiently familiar with the complexities and different schools of antitrust thought. [51]

37. While decisions in the UK are reviewed by the specialist Competition Appeal Tribunal, the UK Report expresses concern that appeals may be unduly burdensome and limit freedom of action. In particular, the Report notes that antitrust cases in the UK are subject to a “full-merits review,” and are easier to appeal than merger decisions. [52] The Report thus recommends that the grounds and standards for review be limited. [53] It should be noted, however, that the CMA has indicated this change may not be needed. [54]

38. Again seeking to address concerns about the speed of enforcement intervention, the UK Report proposed more extensive use of interim measures. Digital markets may be particularly susceptible to the risk that, by the time a final decision is issued concerning anticompetitive conduct, the harm to competition will be irreversible. The Report indicates that the legal threshold for the CMA to use interim measures was lowered in 2014, but the CMA has not yet used these measures. [55] In its response to the Report, the CMA has identified further changes that could facilitate the use of interim measures. [56]

V. Enhancing our understanding of consumer behaviour

39. A final crosscutting theme in several of the reports cited above is the challenge posed by complexities in consumer behaviour—sometimes called behavioural biases. These biases can include a sensitivity to the framing of choices, heavy discounting of future payoffs, loss aversion, choice overload and a tendency to favour the status quo. [57] Digital markets may be particularly prone to these biases, since consumers may consume zero-price products without carefully considering potential trade-offs and product quality (sometimes referred to as the “free effect”). [58] However, while these behaviours have been observed in some circumstances, including laboratory experiments, it is not always clear how they apply in a given digital market. There is an open question, for example, whether there is truly a “privacy paradox” (consumers reporting that they care about privacy but not reflecting this in their consumption decisions), or whether consumers have simply made a rational decision to use the status quo due to limited alternatives and information overload. [59]

40. The need for competition authorities to be aware of these behavioural characteristics is highlighted in several reports. The Stigler Report indicates that firms may in fact take advantage of these biases in order to entrench their market power, including through selective framing and machine learning. [60] More generally, consumer biases could substantially contribute to market power on their own. For example, if consumers do not critically think about quality when they obtain a zero-price good, new entrants may face an insurmountable barrier to entry. This would especially be the case if an entrant has an alternative business model with positive prices, since consumers may be unwilling to pay for something they currently obtain at a price of zero—even if they could obtain significantly more value for a minimal payment.

41. Thus, measures to promote competition could be designed by taking behavioural biases into account, for instance by addressing the use of default options or promoting multi-homing. [61] For example, through the Australia Report, the ACCC requested that Google provide consumers with a choice of internet browsers in its Android operating system and a choice of search engines in its Chrome browser, respectively, without the use of defaults (indicating that it would propose mandatory government action if this request is not accepted). [62]

VI. Conclusion

42. A consensus on the way forward for digital competition policy has not yet been reached. The preceding discussion of proposals for reform purposefully omits a range of criticisms that have arisen from stakeholders and, at least in some cases, the disagreement of competition authorities or policymakers. Further, even the relatively small sample of reports reviewed above varies, for example in terms of the theories of harm that should be prioritised for abuse of dominance cases, or the changes needed to merger regimes.

43. However, these reports do chart a path forward on some key ideas that stand a chance of reaching broad acceptance. Competition enforcement should be complemented with procompetitive regulation, and specific measures may need to be imposed on certain firms. More effort is needed to refine tools for assessing anticompetitive conduct, and it may be necessary to reverse the burden of proof in some cases. The acquisitions of dominant digital firms may need to be subjected to greater scrutiny. Competition enforcement procedures could be accelerated in at least some jurisdictions. Competition policy should pay more than lip service to the complexities in consumer behaviour.

44. In addition, there are some uncontroversial measures that can be undertaken in the short-run, and are in process in several jurisdictions; namely, better and faster co-operation among competition authorities, and further use of market studies to understand competition problems in digital markets outside the framework of enforcement action. There are a range of active enforcement investigations by competition authorities around the world seeking to grapple with the competition concerns identified in the introduction of this paper using existing legislation, with perhaps the most recent such investigation being the one announced in the US. [63]

45. Further, some controversial issues not dealt with in these reports, or which are the subject of disagreement, help to identify the future frontiers of the antitrust policy debate, for which practical measures are not yet clear. Should competition policy care about monopsony power in labour markets, and how should it balance consumer and worker welfare when they move in opposing directions? Do vertical and conglomerate mergers really generate substantial competition harm? Where should the line be drawn between competition, consumer protection, and data protection law?

46. There are concrete policy proposals on the table, a clear identification of where further empirical research is needed, and a set of more complex competition policy questions for the long term. The challenge for the competition community now is to keep the momentum and press forward. 

— The views expressed in this paper do not represent official views of the OECD, its Competition Committee or any OECD members.

PDF Version

Author

  • OECD - Competition Division (Paris)

Quotation

James Mancini, Digital antitrust: An emerging consensus?, November 2019, Concurrences N° 4-2019, Art. N° 92053, www.concurrences.com

Visites 969

All reviews