In her new five-year mandate as Competition Commissioner, Commission Executive Vice-President Vestager will be considering changes to EU antitrust enforcement in relation to big data in several respects: the role of data in defining markets and assessing market power; imposing data-access remedies in abuse-of-dominance cases; and clarifying rules on sharing and pooling data. In merger review, pressure for reform is coming from many directions, but (perhaps surprisingly) not particularly in relation to the assessment of big data combinations. Based on the May 2019 special advisor report, the most significant legal changes may come in the form of a new and less stringent approach to data access remedies, but such cases are few and far between and often take years to resolve. In parallel, Vestager will be working on new regulatory frameworks for artificial intelligence and digital services. Vestager’s views on the potential for big data to confer market power and the pros and cons of mandating access to data may ultimately have a greater role in shaping EU regulation than EU antitrust enforcement.
By Jay Modrall1
I. INTRODUCTION
The antitrust implications of big data have been a key concern for international antitrust authorities since at least 2016, when the French and German authorities released a study on competition and big data2 and the OECD organized a roundtable.3 Since then, antitrust authorities’ focus on big data has only increased, with major studies on big data – as well as other digital economy issues – around the globe, from Australia4 to the United States5 to the United Kingdom,6 among many others.
At the European Union (“EU”) level, Commission Executive Vice-President Vestager, who recently began a new five-year mandate as Competition Commissioner, has spoken extensively on big data since a major 2016 speech7 on the topic. Vestager has, however, taken a more nuanced approach than many commentators. In September 2016, she noted that “companies need to make sure they don’t use data in a way that stops others competing. But that doesn’t mean there’s a problem, just because you hold a large amount of data. We don’t just assume that holding a large amount of data lets you stop others competing.”
More recently, Vestager has sometimes sounded more concerned, commenting that companies “sometimes depend on data, to be able to compete”8 and that “to keep markets open and competitive, we may need to require companies that control [big] data – such as platforms – to share the data they hold.”9 On the other hand, both in 2016 and her most recent speeches, Vestager has noted that the European Commission (“EC”) has examined mergers of companies with important data assets without finding problems due to the combination of the parties’ data. “Either the data involved in the merger wasn’t that important,” she noted, “or similar data was still available to rivals, despite the merger.”
So how will the EC enforce EU competition law in relation to big data, now that its new mandate has begun? On the one hand, the EC is rightly afraid of leaping to conclusions, and has examined numerous mergers without identifying specifically big-data-related concerns. On the other hand, like many other authorities, the EC is concerned about the competitive impact of big data and may well feel political pressure to turn up the heat on companies that accumulate and use big data.
This article discusses some specific changes that we may expect to see in the EC’s approach to big data issues, drawing in particular from the May 2019 report delivered by three high-level advisors (the “Digital Era Report,” or the “Report”),10 which recommend significant changes to EU antitrust policy, including notably in relation to big data.
Perhaps following Vestager’s lead, the Digital Era Report takes a rather nuanced approach to big data. The Report notes that “[t]he significance of data and data access for competition will . . . always depend on an analysis of the specificities of a given market, the type of data, and data usage in a given case.” In relation to proposals to require dominant companies to grant competitors access to data under Article 102 of the Treaty on the Functioning of the European Union (“TFEU”), the Report stressed the need for care, noting that “it is necessary to distinguish between different forms of data, levels of data access, and data uses.”
II. MARKET DEFINITION AND MARKET POWER
Although not specifically in connection with big data, the Digital Era Report argues that the digital economy requires a fundamental re-thinking of the concepts of market definition and market power. “In the digital world,” it argues, “it is less clear that we can identify well-defined markets. Furthermore, in the case of platforms, [market] interdependence . . . becomes a crucial part of the analysis whereas the role of market definition traditionally has been to isolate problems. Therefore, in digital markets, less emphasis should be put on the market definition part of the analysis, and more importance attributed to the theories of harm and identification of anti-competitive strategies.” The Digital Era Report proposes introducing the concept of a digital “ecosystem” as a framework for analysing anti-competitive effects, particularly in relation to after-markets.
The Digital Era Report also argues for a broadening of the concept of market power, noting that there can be market power even in an apparently fragmented marketplace. The Report notes that market shares are often not a useful concept, particularly in the context of online platforms. In addition to the traditional statistics-based approach to analysing market power, the Report advocates using the concept of intermediation power in relation to online platforms and the power conferred by the collection of data not available to competitors.
III. MANDATING DATA ACCESS AND ARTICLE 102 TFEU
Much of the Report’s big-data discussion focuses on conditions in which antitrust law can be used as a tool to require dominant companies to give competitors access to their data. The Report examines arguments for and against mandating data access in different contexts, depending on how the data are obtained (whether they are volunteered, observed, or inferred) and how data are used (use of non-anonymous individual-level data, anonymous individual level data, aggregated data, or contextual data).
The Report notes that the “debate is mostly framed as a debate on whether the criteria of the so-called “essential facilities” doctrine (“EFD”) are met.” In the data access context, however, the Report argues against application of the EFD and for the application of a broader interest-balancing test. In particular, the Report argues against application of the requirement that a claimant under the EFD show that it requires access to the essential facility to develop a “new product.” The goal of the “new product rule” is to preserve a dominant firm’s investment incentives by ensuring a sufficient degree of appropriability of profits for the firm. But the Report argues that in some “access to data” cases, the danger that the firm will lose the incentive to invest in data collection and innovation is lower. Thus, rather than a formal “new product rule,” authorities applying Article 102 TFEU to “access to data” cases should verify that an access mandate will not eliminate the appropriability of benefits resulting from successful investments. According to the Report, this consideration may be especially important outside the platform context.
As under the “standard” EFD, the Report argues that a precondition for data access remedies under Article 102 TFEU should be indispensability. To determine whether data access is indispensable for Article 102 TFEU purposes, it is important to distinguish access to a data set from access to the data in close to real-time where no substitutes exist. The non-substitutability of data may result from the richness (“number of columns”) and size (“number of rows”) of the dataset, in particular where machine-learning algorithms play a role. The frequency of data generation also needs to be taken into account when discussing data access, both for providing a service to the person who generated the data and for aggregated applications.
The balancing of interests needs to take account both of the need to protect the dominant firm’s investment incentives and the need to ensure that strongly entrenched positions of market power, protected by high barriers of entry, remain contestable. In some settings, such interest-balancing can result in a duty to grant access to data in a form that allows competitors to compete effectively in neighbouring markets, which may include a duty to ensure data interoperability.
The Report outlines several data access “scenarios” to identify situations in which data access mandates would be more or less likely to be appropriate. Access mandates would more likely be appropriate where the data controller holds a gatekeeper position (i.e. access to its data is essential for competing on one or more neighbouring markets) or data access requests are somewhat standardised. Access mandates would be less appropriate where a firm requests data for the purpose of training algorithms for uses unrelated to the fields of activity of the data controller, although even in such case an access mandate could be imposed “where the dominant firm is a large player in data markets and has an infrastructure for data access requests in place.”
In the balancing of interest context, the Report discusses whether the threshold for granting access to data should be lowered as compared to access to infrastructure or access to IPR cases, at least where data is produced as a by-product of another activity and incentives to generate such data will persist irrespective of a possible access mandate, for instance in relation to machine-generated sensor data in the Internet of Things context. While the value of data to a dominant data controller may be taken into account in the original price (and therefore need to be protected under competition law), competition law should not protect an after-market monopoly.
Indeed, the Report supports a more interventionist approach to data access – and possibly data interoperability – for use in complementary markets or after-markets, which are part of the broader “ecosystem” served by the data controller. The EC and EU courts have frequently had occasion to address the relevance of secondary markets and after-markets in the Article 102 TFEU context. But the Report argues that “some of the specificities of data could imply that the competition policy treatment of access to data should be different from that of standard aftermarkets… but there is very little economic analysis of these issues, so [their] conclusions should be considered as very preliminary.”
In the specific case of individual-level data, the Report discusses the implications of the data portability right under the EU’s General Data Protection Regulation (“GDPR”), but notes that data portability in the GDPR does not provide a right to continuous data access or interoperability, but simply a right to receive a copy of some accumulated past data. Thus, it may facilitate a data subject’s switching between services but not facilitate multi-homing or the offering of complementary services, which frequently requires continuous and potentially real-time data access. Accordingly, “more demanding” data access regimes than the GDPR could be mandated under Article 102 TFEU for dominant firms, or through sector-specific regulation in particular to open up secondary markets. Tension with the GDPR may also arise where an access mandate involves personal data. The Report notes that the UK and French competition authorities have resolved this tension in past cases by ordering data access on an opt-out basis or permitted access to data only for specified purposes and specified acts of processing.
Thus, in relation to big data, the Report does not support a generally applicable right of access to data under EU antitrust law, but proposes a new analytical framework to make it easier for the EC to impose data access mandates under Article 102 TFEU, especially in relation to after-markets. The Report recommends moving away from the established EFD analysis, applying a more general interest-balancing test.
IV. SHARING AND POOLING DATA AND ARTICLE 101 TFEU
Although much of its discussion focuses on the potential imposition of data-sharing requirements, the Report also discusses the antitrust treatment of voluntary data sharing and pooling. Data sharing and data pooling arrangements will frequently be procompetitive, enabling firms to develop new or better products or services or to train algorithms on a broader, more meaningful basis. Such arrangements can be anti-competitive, however, for example where some competitors are denied access or granted access only on less favourable terms; data sharing amounts to an anti-competitive information exchange including competitively sensitive information; data sharing or pooling discourages competitors from differentiating and improving their own data collection and analytics pipelines; or where the granting of access to data on nonfair, reasonable and non-discriminatory (“FRAND”) terms results in an exploitative abuse. Where a dominant, vertically integrated platform provides privileged data access to its own subsidiaries, this can constitute a form of self-preferencing. Where a data pool has market power and gives its members a significant advantage, the pool may be under a duty to give access to others, perhaps on FRAND or similar terms.
The Report references the treatment of data access issues under Article 102 TFEU and merger control, as well as established rules on R&D agreements and patent pools, but notes that it may be more difficult to establish market power of a data pool based on market shares due to the multi-purpose use of data. Thus, an access regime may need to differ depending on the type of use, and the duty to give access should be proportional to the pool’s market power. For example, a group of smaller players pooling their data to gain a competitive advantage should not be forced to give their pooled data to a much larger player. Where there is a FRAND or similar duty to provide data access, and the pool’s data format standard is proprietary, the standard owner should not be able to raise its fees over time as the pool becomes more important in the market (by analogy to “patent ambushes”), and such duties should apply to both access to the data pool and the use of the data format standard.
The Report notes that the competition law treatment of data sharing and pooling arrangements is relatively new and under-researched, and recommends that the EC use guidance letters and “no infringement” decisions, as well as the next version of the EC’s Guidelines on horizontal cooperation. Indeed, a separate block exemption regulation on data sharing and pooling may be called for.
In an open-ended but interesting aside, the Report notes that legislation is required to give firms guidance on what is allowed and not allowed to encourage fuller development of data markets, data sharing and data pooling arrangements. Such a general legal framework would, for example, define contractual rules for access to non-personal data and empower institutions to facilitate the management of consent into the processing of personal data. Such a legal framework would also reduce the pressure on competition law to revolve concerns about access to data.
V. MERGER CONTROL
Perhaps surprisingly, the Report does not discuss the analysis of big data in the merger review context. Rather, the Report focuses on cases in which a dominant platform and/or ecosystem benefitting from strong positive network effects acquires a target with low turnover but a large and/ or fast-growing user base and a high future market potential. The competitive concerns arise in markets characterised by high concentration and barriers to entry, resulting from strong positive network effects and possibly reinforced by data-driven feedback loops. The Report notes that this scenario concerns a relatively small group of cases and rejects the common label, “killer acquisitions,” noting that the targets in this scenario are typically integrated into the acquirer’s ecosystem, not shut down.
Back in 2016, the Franco-German Report expressed concern that a combination of “data troves could raise competition concerns if the combination of data makes it impossible for competitors to replicate the information possessed by the merged entity.” As mentioned, however, the EC has investigated this theory a number of times in recent years but have not yet found a cause for concern. The issue will no doubt continue to arise in future cases, but there does not appear to be strong pressure for a change in the EC’s current approach.
VI. QUO VADIS, EC?
With the Digital Era Report in the EC’s arsenal, a new mandate, and political support for scrutiny of big data, what changes can we expect in EU competition law enforcement in relation to big data? As Yogi Berra supposedly said, it’s tough to make predictions, especially about the future. But we can make some educated guesses in each of the four areas discussed above.
First, the Digital Era Report recommends that the EC review its approach to defining antitrust markets and assessing market power. Indeed, Vestager recently announced11 a review of the EC’s 1997 notice on market definition.12 Vestager noted that “digitisation . . . raises some more challenging questions for the way we define product markets. Very often, we find that big digital businesses . . . [provide] consumers with an ecosystem of services, that are all designed to work together well. And as the special advisers pointed out in their report, it can be difficult for consumers to switch from one ecosystem to another.” Though the results of the EC’s review won’t be known until late 2020 or 2021, we can expect the EC to consider a wider perspective on markets that are connected with one another in “ecosystems” and to focus on the role of data in locking customers into such ecosystems.
Second, in relation to the possibility that Article 102 TFEU may require companies to share their data, the EC can be expected to explore new theories of harm, or revisions to existing theories on a case-by-case basis. Such mandates would only be imposed on companies found to have market power, as is the case today, but changes to the EC’s traditional assessment of market definition and market power may make findings of market power more frequent. Moreover, if the EC follows the Digital Era Report’s recommendation to relax the traditional EFD criteria, the EC may be more likely to impose data access remedies in the coming years. If so, companies active in complementary or downstream markets are more likely to benefit than head-to-head competitors of the dataholder. On the other hand, any such changes to Article 102 TFEU enforcement in relation to big data will emerge, if at all, only through individual cases, which are few and far between and normally take years to resolve.
Third, in relation to the treatment of data sharing and pooling under Article 101 TFEU, the Digital Era Report called on the EC to provide additional guidance. In fact, in late 2019, the EC launched a review of its horizontal block exemption regulations and guidance on horizontal cooperation, including notably sharing of competitively sensitive information.13 This review was triggered by the upcoming expiration of the relevant block exemptions, not by the Digital Era Report recommendations, but it seems likely that the EC will take this opportunity to clarify the treatment of data sharing and pooling. A separate block exemption on data sharing and pooling, as proposed in the Report, could be very helpful.
Fourth, merger control seems to be the area in which major changes in relation to big data are least likely. The EC has developed a methodology for analyzing big data combinations in recent years, and there seems to be no strong pressure for change. Each case is different, but the EC’s practice so far suggests that the combination of merging parties’ “data troves,” without more, are unlikely to raise concerns.
In summary, in the new EC mandate we are likely to see an increased focus on big data in EU competition law enforcement. A number of independent initiatives already under way could facilitate change. However, any such changes are likely to be incremental. A senior EC official recently told me that there are more conferences on big data than cases. That may continue to be true.
On the other hand, competition law enforcement is not the only game in town. Indeed, Executive Vice-President Vestager observed in September 2019 that “data is not just an issue for competition. . . So we may also need broader rules to make sure that the way companies collect and use data doesn’t harm the fundamental values of our society.”14 Indeed, Commission President von der Leyen has charged Vestager – in her parallel capacity as Commissioner for making Europe fit for the digital age – with coordinating work on new legislation on artificial intelligence and digital services. The scope of such legislation is not yet clear, but ultimately Vice-President Vestager may have a greater impact on the collection and use of big data in Europe in her regulatory capacity than in her antitrust enforcer capacity.
1 Norton Rose Fulbright LLP, Brussels.
2 Autorité de la Concurrence and Bundeskartellamt, Competition Law and Data, May 10, 2016, available at https://www.bundeskartellamt.de/SharedDocs/Publikation/DE/Berichte/Big%20Data%20Papier.pdf?blob=publicationFile&v=2 (the Franco-German Report).
3 See https://www.oecd.org/competition/big-data-bringing-competition-policy-to-the-digital-era.htm.
4 Australian Competition & Consumer Commission, Digital Platforms Inquiry Final Report, June 2019, available at https://www.accc.gov.au/system/files/Digital%20platforms%20inquiry%20-%20final%20report.pdf.
5 Federal Trade Commission, Hearings on Competition and Consumer Protection in the 21st Century, available at https://www.ftc.gov/policy/hearings-competition-consumer-protection.
6 Digital Competition Expert Panel, Unlocking Digital Competition, March 2019, available at https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/785547/unlocking_digital_competition_furman_review_web.pdf.
7 Margrethe Vestager, Competition and Big Data, September 29, 2016, available at https://wayback.archive-it.org/12090/20191129222113/https://ec.europa.eu/commission/commissioners/2014-2019/vestager/announcements/big-data-and-competition_en.
8 Margrethe Vestager, Defining markets in a new age, December 9, 2019, available at https://ec.europa.eu/commission/commissioners/2019-2024/vestager/announcements/defining-markets-new-age_en.
9 Margrethe Vestager, Digital power at the service of humanity, November 29, 2019, available at https://wayback.archive-it.org/12090/20191130155750/https://ec.europa.eu/commission/commissioners/2014-2019/vestager/announcements/digital-power-service-humanity_en.
10 Jacques Crémer, Yves-Alexandre de Montjoye and Heike Schweitzer, Competition Policy for the Digital Era, available at https://ec.europa.eu/competition/publications/reports/kd0419345enn.pdf.
11 Margrethe Vestager, Defining markets in a new age, December 9, 2019, available at https://ec.europa.eu/commission/commissioners/2019-2024/vestager/announcements/defining-markets-new-age_en.
12 Commission Notice on the definition of relevant market for the purposes of Community competition law, OJ C 372, 9.12.97, pp 5-13.
13 See https://ec.europa.eu/competition/consultations/2019_hbers/index_en.html.
14 Margrethe Vestager, Security and trust in a digital world, September 13, 2019, available at https://wayback.archive-it.org/12090/20191130063011/https://ec.europa.eu/commission/commissioners/2014-2019/vestager/announcements/security-and-trust-digital-world_en.
