Designing an EU Intervention Standard for Digital Platforms

A consensus is emerging around the world about the need for policymakers to address certain characteristics and competitive tendencies that are generated by digital platforms or digital ecosystems, with a view to reforming the public policy instruments currently in place so that they are fit for the digital age. The paper starts by reviewing the relevant precedents under EU competition law and economic regulation upon which this reform could be based. The paper then puts forward recommendations to adapt competition rules, in particular as regards the determination of market power (e.g., by better taking into account the effects of ecosystems, the impact of potential competition and the role of innovation) and the application of theories of harm (i.e. by focusing on leveraging and envelopment behaviour, access to key innovation capabilities, discrimination and self-preferencing and the violation of normative regulatory principles). The paper then proceeds to propose a cumulative 'three criteria test' to determine the types of digital platforms upon which competition rules, and possibly complementary regulation, should focus. These three criteria require an assessment of: (i) the existence of market structures which are highly concentrated and non-contestable; (ii) the presence of digital gatekeepers which act as unavoidable trading partners; (iii) and, for the purposes of ex ante regulation, the lack of effectiveness of competition rules to address the identified problems in the market. The paper also considers the types of remedies that could be imposed on those identified digital platforms, including: interoperability and access to key innovation capabilities such as data; the prohibition of anti-competitive discrimination; and the facilitation of consumer switching. Given the rapid evolution of technology and market uncertainty, consideration should be given as to whether these remedies should be imposed in a participatory manner with the industry stakeholders directly affected by the measures. Finally, the paper deals with a number of procedural and institutional issues raised by the adoption of such a legal standard, proposing to adapt existing antitrust guidelines, to extend the power of DG Competition to conduct fully fledged market investigations (as is the case in the UK and Australia) and possibly to work closely with National Regulatory Agencies, coordination with whom at EU level arguably needs to be strengthened.


Introduction
A consensus is emerging around the world from a series of international and national reports and studies 1 about the need for policymakers to address certain characteristics and competitive tendencies that are generated by digital platforms or digital ecosystems. 2 In its new Digital Strategy, the European Commission indicates that that it will evaluate and review of the fitness of EU competition rules for the digital age and launch of a sector inquiry and that it will explore ex ante rules to ensure that markets characterised by large platforms with significant network effects acting as gatekeepers, remain fair and contestable for innovators, businesses, and new market entrants. 3 It is therefore appropriate to consider the extent to which a legal regime can be developed within the European Union ("EU") so as to justify intervention against such digital platforms or ecosystems, either under ex post principles and/or under ex ante regulation, and to identify the particular elements of such a regime. In exploring these options, we consider that the end product should reflect an evolution of existing practice and principles, rather than a revolutionary step-change in public policy.
To this end, this paper is structured as follows: After this introduction, Section 2 analyses whether existing legal doctrines are sufficiently flexible to provide a sound analytical basis for intervention under competition rules, or are more apt to serve as the basis of a new standard of regulatory intervention, and the robustness of the theories of harm explored by the European Commission in its existing and pending case practice and the role that is played by ex ante regulation and the nature of the regulation-style remedies that can be harnessed by policymakers to address concerns about digital markets. Section 3 reviews the competition doctrines relating to theories of harm that should be explored further if effective enforcement is likely to be realised. Based on the evaluation of the issues listed above, Section 4 formulates a standard of intervention that embraces both ex post and ex ante disciplines, consistent with Robert Schuman Centre for Advanced Studies Working Papers the policy guidance set forth in the earlier parts of this paper. Then Section 5 deals with the remedies applicable to firms meeting the standard for intervention standard and Section 6 deals with implementation and institutional issues. Finally, Section 7 concludes with a summary of our main recommendations.

Relevant Legal and Economic Standards in the Ex post Context
A number of the international studies express the view that traditional competition policy may have a number of shortcomings in addressing many of the competitive harms allegedly generated by digital platforms and Internet ecosystems. However, there are a number of existing competition law doctrines whose application in a digital platform environment needs to be explored further, as they potentially provide important analytical bases upon which to ground the more complex theories of harm that arise in the digital platform context.

The doctrine of special responsibility
In our view, intervention in relation to digital platforms lends itself to the application of the doctrine of "special responsibility", in such a way that ultimately protects consumers. The doctrine of special responsibility is a loosely defined doctrine dating back to the 1983 Judgment in Michelin I, and has been applied in a wide range of Commission Decisions since that time which have in turn been endorsed by the European Courts. According to the Court of Justice in Michelin I: "A finding that an undertaking has a dominant position is not in itself a recrimination but simply means that, irrespective of the reasons for which it has such a dominant position, the undertaking concerned has a special responsibility not to allow its conduct to impair genuine undistorted competition in the Common Market." 4 The essence of the doctrine lies in the understanding that firms with market power are obliged to be mindful of the anti-competitive effects of their unilateral commercial practices, given that such effects are unlikely to arise from comparable actions taken by firms without market power. 5 The application of the doctrine is consistent with the idea that the concept of an "abuse" is an objective one which does not turn on the subjective intention of the dominant firm. 6 The degree to which the doctrine of special responsibility applies to a firm with market power will often turn on the extent of its market power, as a firm that is on the verge of holding monopoly power (or "super-dominance") will have a heightened sense of responsibility to respect this standard. Its association with very significant levels of market power might explain why the application of the doctrine comes very close, in the eyes of at least some commentators, to condemning even what would otherwise be "competition on the merits" if the impact of such conduct is sufficiently anti-competitive. 7 We do not believe that the doctrine should be used to censure activities which amount to competition on the merits. To this end, it is arguable that, as a means of restricting the scope of the doctrine, the Wolf Sauter has put forward the proposition that the natural extension of the doctrine of special responsibility into the online world is to impose a "duty of care" on dominant firms. 9 Whereas the traditional doctrine has been focused on the foreclosure of rivals (exclusionary behaviour), 10 this commentator argues that the doctrine should also extendgiven the multi-sided nature of a digital platformto the exploitation of relationships with direct customers and end consumers. In our view, stretching the traditional doctrine to fit the contours of digital platforms may, provided sufficient safeguards are in place, be an appropriate and proportionate response on the part of policymakers in light of the competition concerns that arise from market power generated in relation to such platforms.
Given that so much of the commercial behaviour associated with digital platforms involves different forms of contractual consideration beyond monetary exchanges, concerns information asymmetries which may be very high, and is directed towards influencing consumers in their positive competitive choices, reliance on the doctrine to substantiate an exclusionary abuse appears to be reasonable and proportionate where it is the exploitative behaviour that fuels the exclusionary conduct. 11 The application of the doctrine would thus facilitate reliance on theories of harm which involve practices such as discrimination, self-preferencing, adversely affecting consumer choice, and the abuse of procedures (including the lack of transparency), all of which have a particular relevance in the digital platform environment and all of which reflect a link between the various sides of the market intermediated by a digital platform.
Unlike the negative duty to abstain from potentially anti-competitive behaviour, however, a duty of care in the context of a digital platform might need to involve positive steps to be taken in order to avoid a market failure (i.e., steps of a pre-emptive in nature), which has more in common with regulatory intervention (i.e., addressing problems based on industry structure) rather than intervention under competition rules (i.e., addressing problems driven by strategic market behaviour). It is in these types of situations where it is arguably more appropriate for ex ante instruments to intervene (i.e., whether in the form of consumer protection, data protection and so forth) than for ex post censure to occur.
Many of these types of behaviour are already addressed in legislation which establish new sets of normative behaviour in terms of transparency vis a vis customers with insufficient bargaining power and in terms of discriminatory behaviour. 12 It is not unreasonable to have these regulatory principles being able to serve as legitimate normative yardsticks which can guide antitrust enforcement in the context of digital platforms. However, in relying on the doctrine of special responsibility in this way, it is reasonable and proportionate that any intervention should be limited to the pursuit of the principle of efficiency, rather than the pursuit of wider distributional welfare goals. 13 Of course, the precise scope of the doctrine of special responsibility will be clearly different in response to the very different business models which prevail in the digital world, as they have very different ways of bringing together the multiple sides of any given digital "market". 14 Our preliminary conclusion is that the particular balance that needs to be sustained between the different sides of a platform in order to achieve optimum welfare results lends itself to the application of the doctrine of "special responsibility", which can take into account impacts on the various sides of the platform. In doing so, however, we believe that competition regulators should be able to articulate how the exploitative abuse on one side of a digital platform (i.e., an action directed at the expense of consumers) can support or sustain a foreclosure abuse on another side of that platform (i.e., an action directed against competitors).
Most importantly, the overarching application of that doctrine can also be understood by reference to a number of other key concepts such as that of a "bottleneck" or a "digital gatekeeper", the implications of dealing with an "unavoidable trading partner", and the related idea that less powerful traders or customers might be in a situation of "dependency" vis a vis many digital online platform providers. Some of these doctrines are longstanding in national and EU competition policy, while others have evolved more recently in the digital era.

Bottlenecks
The economic concept of a "bottleneck" is often considered to be comparable to the "digital gatekeeper" concept, but arguably differs insofar as a bottleneck facility is universally treated as a point of congestion which has the potential to lead to objective inefficiencies. By contrast, a digital gatekeeper (see discussion below), is very capable of generating efficiencies, 15 but the public policy dilemma will be whether those efficiencies are overborne by the potential anti-competitive effects generated by a particular online platform.
Because of the traditional understanding that many historical monopolies in the electronic communications sector could manipulate access relationships because of the potential scarcity in capacity and restrictive network design, a bottleneck facility has often been treated in EU regulatory policy as the equivalent of an "essential facility" 16 and has widely been considered to provide the analytical basis for the regulation of access to electronic communications networks. 17 More recently, a US study has proposed that the regulation of digital platforms should be directed to those operators which have "bottleneck power". 18 The bottleneck concept has, on balance, more in common with the regulatory tradition than with the competition policy tradition, where it is likely to be treated as being synonymous with the control over an essential facility. As such, its value is arguably more relevant if used in a regulatory context, rather than running the risk of blurring existing antitrust standards.
14 For example, a sales portal has very different characteristics to a social network, a search engine, an apps-hosting ecosystem, a ride-sharing business, and so forth. All of these platforms strike a different balance between the various sides of the market in relation to which they operate, in order to achieve optimal results for the various sides of the platform involved. 15 As occurs with respect to many digital platforms, which benefit from economies of scale and scope by being able to provide a wide range of services. 16 For a wide-ranging discussion of the concept of "essential facilities" and its relationship with the bottleneck concept, see

Designing an EU Intervention Standard for Digital Platforms
European University Institute 5

Digital Gatekeepers
The concept of a "digital gatekeeper" is synonymous with the advent of the Internet, and its origins in EU competition policy date back to the tenure of Commissioner Mario Monti. 19 A loose definition of a "gatekeeper" would be that it is an economic agent that can control access by a group of users to some goods or another group of users. To be more precise, we can distinguish two different types of situation.
In the first situation, the gatekeeper controls access by third-party firms to its users. 20 For example, an online social network has, to some extent, control over access to its users by online advertisers, particularly for those consumers who spend most of their time on that social network. In the second situation, the gatekeeper controls access to content, products and/or services. 21 For example, a search engine controls the access of users to Web content via its ranking algorithm, while a music streaming service controls access to its large catalogue of music titles through its personalised recommendations, etc.
Thus, in the context of digital platforms, the "gatekeeper" concept is associated with that party having a privileged relationship with a customer (which may be an end consumer) which is critical in directing the customer's access to services or apps, while at the same time allowing that provider to take advantage of the consumer's frequency of use of its digital platform so as to tailor ever more sophisticated and varied services to that customer. In this sense, the larger the network effects generated by the digital gatekeeper, the more difficult it is for consumers to avoid dealing with them.
Unlike the concept of a "bottleneck", Commission's understanding of the "digital gatekeeper" concept should not be conflated with its traditional thinking about what constitutes an "essential facility", as the latter concept serves as the usual antitrust standard used to determine whether access should be mandated to a dominant firm's infrastructure. 22 By contrast, the digital gatekeeper concept would be a less onerous legal standard to satisfy when compared to the essential facility doctrine, which is tantamount to declaring an undertaking to be a quasi-monopolist. 23 This in part explains why the role of the digital gatekeeper is not only seen by the Commission through the prism of potential market power under Article 102 TFEU, but also by reference to anti-competitive agreements caught by Article 101 TFEU. Irrespective, however, of the terminology used to characterise a physical facility, it is nevertheless clear that if access to such a facility is mandated by ex ante regulation, the pursuit of a competition law action will not be barred simply because that facility falls short of being classified as an 'essential facility'. 24 Accordingly, the mandating of access by regulation to a facility will be treated, to all intents and purposes, as a mandatory access obligation under EU competition rules. 25 It is clear that the doctrine of special responsibility sits comfortably with the idea that a digital platform may be playing the role of a "gatekeeper". Both concepts fall short of proving that the online platform constitutes an essential facility, which is identified with a view to determining the legality of a refusal to denial (i.e., the denial of access 'bottleneck' situation existed), but where the market power associated with the platform has the potential to distort competition on one or more sides of the platform.
Thus, although pregnant with meaning in terms of its significance in establishing the existence of dominance, the 'digital gatekeeper' concept nevertheless has as yet no jurisprudential foundation under Article 102 TFEU to constitute a self-standing legal test of dominance. As such, it is arguably a concept that is best utilised in a regulatory context as the basis for potential targeted obligations, whether symmetric or asymmetric, precisely because it does convey the idea of relative market power capable of restricting competitive outcomes.

Unavoidable trading partners
The "digital gatekeeper" concept is also closely aligned to the idea that dominance (i.e., market power) is often identified with situations where an undertaking is considered by customers to be an "unavoidable trading partner". By being an unavoidable trading partner, an undertaking is able to satisfy the criteria of being able to behave independently of customers, as is required under the Article 102 jurisprudence if an undertaking is defined to be dominant. 26 In the digital online platform context, the Cremer Report has already considered that classification as an unavoidable trading partner is usually associated with the existence of "intermediation power". 27 The application of the unavoidable trading partner 28 concept means that market power might be capable of being exercised even in fragmented markets if there are multiple sides to the market in which the platform operates, at least to the extent that groups of customers feel they have little option other than to deal with particular providers (which also explains in part why customer inertia through recourse to default settings is so widespread in relation to digital platforms). However, where widespread multihoming is available and is used by customers, the leveraging of market power becomes increasingly difficult for platform providers to achieve. 29 The German Cartel Office appeared to rely in part on the doctrine of "unavoidable trading partner" to base its Decision to intervene against Facebook's alleged practice of sharing customer data between its various social media and communications operations, 30 although its approach is arguably just as much an example of customers being deemed to be "dependent" on Facebook according to traditional German antitrust doctrine (see below). Particular manifestations of the doctrine of an unavoidable trading partner lie in the twin ideas that:  Unilateral market power might exist in relation to "must have" content where certain TV programmes are critically important commercial inputs in order to be able to attract advertising and subscribers. 31  28 The ACCC Report refers to Google and Facebook as "unavoidable trading partners" for a significant number of media businesses, in the sense that they are important channels through which consumers access news, with many news businesses being dependent on them as key sources of referral traffic. 29 Ibid., page 49. products enjoy a "must have" quality. 32 This approach to conglomerate market power issues is particularly relevant in the context of digital platforms, given that the most widely applied theory of harm is based on the threat of leveraging into adjacent or neighbouring markets, which is facilitated where a platform provider has a wide portfolio of services (see discussion below).
The implications of dealing with an "unavoidable trading partner" are clear in a digital platform environment where the platform operator has market power. Unless multi-homing is a realistic commercial option which is acted upon by consumers in a meaningful way, several digital markets are dominated by individual firms for whom many consumers cannot envisage a viable alternative on an "all or nothing" basis because a market has tended to "tip" in favour of the incumbent digital platform provider (i.e., creating a "winner takes all" situation). Moreover, even where other competitive alternatives exist, many digital platforms are still characterised by "winner takes most" situations, at least where even multi-homing has its limitations in terms of providing a credible, sustainable competitive alternative to the incumbent platform provider. However, where a firm is an unavoidable trading partner but is nevertheless not able to satisfy the 'dominance' standard under Article 102 TFEU, it may be more appropriate to utilise the concept of an unavoidable trading partner in tandem with the concept of a 'digital gatekeeper' as the basis of a future regulatory standard of intervention where the circumstances are appropriate.

Situations of economic dependency
The digital gatekeeper concept has not only found fertile ground in the concept of the 'unavoidable trading partner', but also in the idea that dominant firms have a special responsibility when dealing with smaller traders which are dependent upon them for their economic viability. Thus, the longstanding doctrine of "economic dependency" under German antitrust law, 33 which has also been embraced by several Member States 34 such as France 35 and, more recently, Belgium, 36 is relied upon to prevent undertakings from exercising unfettered commercial freedom in those situations where their customers do not have realistic solutions in selling or purchasing other products or services in the market. 37  The state of economic dependence requires that it be a demonstrated that it is impossible for the plaintiff to resort to another undertaking for the supply, or the sale, of a given product or service, due to technical or economic reasons. In essence, four types of economic dependence have been addressed by the French Competition Authority, namely: (i) scarcity-based dependence; (ii) dependence associated with long-lasting business relationships; (iii) assortment-based dependence; and (iv) demand-based buyer power dependence. 36 Law of 21 March 2019 amending the Code of Economic Law as regards abuses of economic dependence, abusive clauses and unfair market practices between companies. 37 The concept of dependency is also increasingly used in several regulatory instruments related to the digital sector (such as generally understood that the concept of economic dependency might apply in one of two circumstances, namely: (i) a high level of concentration in the market (i.e., market dominance); or (ii) the special features of a bilateral relationship between the undertaking in question and its individual customers. 38 The concept of dependency identifies a range of relationships which can trigger public intervention, including relationships based on: product ranges or strong brands (e.g., "must have" products or product ranges); 39 large volumes of business; product shortages; the strength of the buyer; and technical standards or specifications set by undertaking in question (e.g., for spare parts). 40 Arguably the most comprehensive legal definition of dependency is found in the new Belgian legislation, which specifies that economic dependency is characterised by: "the absence of reasonably equivalent alternatives available within a reasonable period of time, on reasonable terms and at reasonable costs, allowing it for each of them to impose services or conditions that could not be obtained under normal market conditions." 41 Thus, the overriding principle behind the application of the doctrine is the view that the dependent customer or competitor has "insufficient and unacceptable means of switching to other providers." 42 In light of the relative importance attached by all empirical studies on the reluctance of customers using digital platforms to choose alternative options to their default settings and the concerns about the legitimacy of customer consent to the use of their data, the German Cartel Office applied the doctrine as a key element of its action against Facebook (see Table 2 below in Section 2.2.2). However, given that one alternative limb of the doctrine may apply without recourse to a finding of dominance, it may be more appropriate as a policy option to address such issues through regulatory measures rather than by a diluted measure of dominance that is sui generis to online platforms. In this respect, the doctrine can be associated with a doctrine such as that of "special responsibility", at least where a position of market dominance can be established. In the alternative, it can be driven by the existence of a position of 'digital gatekeeper' which, to all intents and purposes, amounts to an 'unavoidable trading partner' where an alternative regulatory standard might need to be established. It is also important to note that some of the thinking associated with the concept of dependency turns on the application of conglomerate effects theory that finds fertile ground in those markets driven by digital platforms (i.e., the existence of broad product ranges and "must have" products).
In practice, there will be many instances in practice where the application of the concept of "economic dependence" under antitrust rules has little to distinguish it from the concept of an "unavoidable trading partner", with both expressions often being used interchangeably in the context of a wider antitrust analysis. In a digital platform environment where competitive choices may often seem to be illusory because of consumer inertia, it is a concept which already has achieved significant traction relationships in the agricultural and food supply chain, OJ [2019] L111/59). Moreover, the concept is increasingly being relied upon outside the EU. Thus, Japanese legislation has long ago embraced a similar concept as part of its antitrust tradition. More recently, the Indian Competition Commission has implicitly endorsed the concept in its Key Findings and Observations, Market Study on E-Commerce in India, 8 January 2020, where it emphasises "bargaining power imbalance and information asymmetry" at the core of many of the issues that arose in its study (at para. 112). 38  There are a number of existing doctrines in the EU competition law tradition which can be harnessed to better understand the competitive implications of digital platforms and Internet ecosystems. Doctrines such as that of "special responsibility", "unavoidable trading partners" and "dependency" all appear to have an important role to play in understanding why certain commercial relationships might lead to problematic commercial outcomes. However, in order to better understand how these doctrines can apply as integral aspects of a competition law enforcement policy, it seems inevitable that some form of guidance is required from the European Commission as to how those doctrines should apply to particular digital platforms.
There are also a number of other concepts used in the context of antitrust proceedings, especially those relating to "bottlenecks" and "digital gatekeepers", which the authors feel can play an important role in the potential formulation of a new regulatory standard for targeted policies that can complement or supplement competition enforcement, especially when combined with the other doctrines with stronger roots in competition policy such as those described above.

Theories of Harm: Recent Precedents in Competition Law
In determining whether EU competition rules are sufficiently robust in order to address perceived market abuses, one also needs to turn to the Commission's existing administrative practice in the application of Article 102 TFEU to digital platforms and Internet ecosystems.

The Google precedents
In the Google Shopping Case in 2017, the European Commission concluded that Google had abused its dominant position in search engines, 46 by conferring an anti-competitive advantage to another one of Google's services, namely, comparison shopping. It did so by having its search results confer the most prominent placements to its own comparison shopping service while simultaneously demoting rival services, thereby stifling competition on the merits in comparison shopping markets. 43 On appeal, although the Decision was overturned (see discussion later), the Judgment of the Düsseldorf Court has nevertheless not challenged the extent to which the Cartel Office relied upon the doctrine of economic dependency. 44 As noted elsewhere, however, new German legislation has been proposed to strengthen and widen the scope of application of the doctrine of dependency, thereby arguably rendering moot any judicial overturning of the Cartel Office's Decision against Facebook. 45 At paragraph 84 of that Guidance, the Commission notes that, although it will in principle deal with de novo refusals to supply in the same manner as it would to terminations of previously negotiated relationships, it is nevertheless the case that "termination of an existing supply arrangement is more likely to be found to be abusive than a de novo refusal" to supply. In our view, this approach is driven in large measure by the idea that the dominant firm holds such overwhelming bargaining power vis a vis its customers that the latter is in a position of economic dependency on the former. It is also consistent with the idea that a dominant firm that takes the commercial decision to open its network to competitors will render itself vulnerable to an Article 102 action in the event that it chooses to terminate supply at some later date (the doctrine is also relevant to the analysis of 'self-preferencing' practices by those firms which have opened up their platforms for use by downstream competitors). 46  In a subsequent investigation, the Commission sanctioned Google in 2018 for its various anticompetitive practices vis a vis Android device manufacturers which were designed to strengthen Google's dominance in search engines. 47 In its Android Case, the Commission identified three distinct types of problematic commercial practices, namely: (i) obligations on manufacturers to pre-install the Google Search app and browser app (Chrome); (ii) payments to certain large manufacturers and mobile network operators on condition that they pre-install exclusively the Google Search app on their devices; and (iii) preventing manufacturers wishing to pre-install Google apps from selling even a smart mobile device running on alternative versions of Android that had not been approved by Google.
Most recently, in early 2019, the Commission concluded in the AdSense Case 48 that Google had engaged in abusive practices in online advertising, primarily by imposing a number of restrictive clauses in contracts with third-party websites which prevented Google's rivals from placing their search advertisements on these websites. The Commission's case was based on the understanding that Google was implementing this practice in a bid to foster a de facto exclusivity relationship.
The Commission's approach in these cases is based on adaptations to more traditional EU competition law doctrines whether based on: a particular application of the non-discrimination principle to embrace self-preferencing practices (Google Search); the application of traditional leverage theory to sanction restrictive obligations imposed on device manufacturers (Android); or quasi-exclusive relationships with publishers designed to dis-intermediate Google's search advertising rivals (AdSense). While the respective market definition and market analysis exercises undertaken in each case were anything but straightforward, the Commission's approaches to theories of harm in relation to digital platforms has been previewed in a number of IT sector cases in the recent past. 49 Pending the resolution of European Court appeals in the various Google Cases in the Commission's favour, 50 it would appear that traditional competition law doctrines have been sufficiently robust to allow the Commission to sustain its investigations into Google's practices. The Commission's focal point has been to identify the source of Google's market power and the ways in which that market power is being leveraged into neighbouring or adjacent areas, while at the same time reinforcing its market power in search. Moreover, when assessing the anti-competitive impacts of self-preferencing strategies, the Commission has insisted that it was justified in not subjecting Google to the test of whether it held the position of an essential facility. Finally, in each case, the Commission remained unconvinced by the defendant that its actions generated sufficient efficiencies to overcome their anti-competitive effects, nor was it convinced that Google's restrictions were indispensable in achieving such efficiencies.

Designing an EU Intervention Standard for Digital Platforms
European University Institute 11 While the theories of harm explored by the Commission in the respective Android and AdSense Cases are based on established, well understood theories of harm, the abusive practice of self -preferencing is arguably a more controversial theory of harm relied upon by the Commission. This is essentially because its critics feel that the prohibition of self-preferencing undermines the benefits of vertical integration and the efficiency gains that can be realised by integrated undertakings. By contrast, however clear support for self-preferencing practices being classified as anti-competitive practices in their own right Robert Schuman Centre for Advanced Studies Working Papers can be found in: 52 (i) the case-law on margin squeezes in the telecommunications sector; 53 (ii) a large number of precedents adopted under the Commission's powers under in Article 106 (1)TFEU relating to potential leveraging and conflicts of interest, when applied in conjunction with Article 102 TFEU; 54 (iii) the adoption of a very recent Decision of the French Competition Authority. 55

Facebook in Germany
The German Cartel Office challenged the behaviour of Facebook in early 2019, 56 concluding that Facebook had engaged in an exploitative abuse of consumers through its process of gathering and combining data from all of its various business units 57 and other third party sources without having obtained the explicit consent of its users to do so. In the circumstances, the breach of data protection rules amounted to abusive behaviour under Section 19 GWB, being classified by the Cartel Office as "unfair trading conditions". In the eyes of the Cartel Office, Facebook did not offer consumers a genuine basis upon which to provide their informed consent.
The sweeping terms in which German antitrust rules are formulated arguably offered the Cartel Office a level of flexibility in dealing with this case that might otherwise not be available to the European Commission in its application of Article 102 TFEU.
The finding of the Cartel Office is novel in a number of ways:  First, it is striking insofar as it takes a very broad approach in determining what constitutes "freely given consent" for data protection purposes, and concluded that an infringement of data protection rules constitutes an antitrust infringement.
 Second, the Cartel Office has taken a view as to what a reasonable user would expect in terms of the handling of its personal data, which arguably stretches what even data protection authorities would be willing to conclude as regards the reality of "consent".
 Third, the logic of the Decision is driven by the understandingmade express in German competition legislationthat consumers were in a position of "dependence" vis a vis Facebook in the role of the latter as a dominant provider of social media services. Also the ongoing investigation of the Turkey's Competition Authority into Google's search engine, https://www.reuters.com/article/us-turkey-google/turkey-to-investigate-whether-google-violated-competition-law-idUSKCN1P11AQ. 56 57 In this regard, refer also to the recent CMA Report on online platforms and digital advertising, in which it expresses its concern that Facebook's position may have become entrenched with negative consequences for the people and businesses who use these services on a daily basis; available at : https://www.gov.uk/government/news/cma-lifts-the-lid-on-digitalgiants.

Designing an EU Intervention Standard for Digital Platforms
European University Institute 13 As noted earlier, the doctrine of dependency finds fertile ground in the antitrust cultures of certain Member States, 58 and is arguably capable of being harnessed by the Commission and incorporated into its Article 102 TFEU analysis. However, it does not, of itself, amount to a finding of dominance, while it is equally compatible with an approach that is regulatory in nature, or at the very least consumer protection-oriented.
On appeal, the Higher Regional Court of Düsseldorf overturned the Decision of the Cartel Office in August 2019. 59 In doing so, it observed that the Decision had inappropriately strayed from EU standards of review where market power allegations were concerned. According to the Court, the Cartel Office's Decision was prone to serious legal doubts inter alia because it had: failed to establish that Facebook was dominant in online advertising or that its market power had allowed it to exploit users' data; (ii) not explained how Facebook's alleged violations of EU Data Protection rules constituted an infringement of competition rules; and (iii) not distinguished between the appropriate acts of collection, processing and use of consumer data, nor what constituted excessive acts of data collection.
The Court was unimpressed with the Cartel Office's lack of reasoning to support its findings, 60 nor was it impressed with the Cartel Office's conclusion that users had no effective choice in releasing "all" of their data in the circumstances. What remains to be seen is whether the Appeal Court will see fit to allow the Cartel Office to configure an antitrust theory of harm based on exclusionary abuse that is based on a breach of data protection rules, 61 or to support the conclusion that Facebook was truly an "unavoidable business partner".
Most importantly, the Court censured the approach of the Cartel Office in taking two analytical shortcuts, by: (i) attributing too much weight to the "indifference or inconvenience" of users when considering Facebook's actions; 62 and (ii) by not drawing a causal connection between data protection violations and competition harm, as the doctrine of "special responsibility" only extends to competition matters, 63 and not to a simple data protection infringement. 64 By the same token, the Düsseldorf Court was open to being persuaded that an infringement of data protection rules might constitute an exploitative abuse at the expense of consumers. Moreover, our understanding is that the logic of the Court suggests that the Cartel Office's allegations were flawed, insofar as the necessary probative evidence was lacking to support its case, rather than being misconceived. 65 As the Judgment is itself to be appealed, it will no doubt become clearer as to the extent that data protection-based infractions will be deemed capable of constituting an essential element of an antitrust theory of harm. Moreover, another interesting line of enquiry before the Appeal Court should be whether or not an exploitative abuse on the consumer's side of the digital platform can be used to generate an 58 For example: Germany, France and Belgium. While not rising to the level of a "general principle of law" over time (TFEU, Article 15(3)), it is nevertheless a doctrine that is compatible with an antitrust analysis of market power and its likely exercise. 59 Facebook v. Bundeskartellamt 64 Or at least to specify the nature and scope of those violations which would be so grievous as to amount to an exclusionary abuse. 65 Other issues which will no doubt be addressed by the Appeal Court will be the extent to which the concepts of consumer "dependency" and the status of the defendant as an "unavoidable trading partner" establish a legal standard that is compatible with that used under EU enforcement. exclusionary abuse on another side of the digital platform, especially if the two types of abusive behaviour are linked causally and by reference to the overarching duty of 'special responsibility' on the part of the digital platform operator. Given that 'markets' need not be driven by the transfer of financial consideration but by data transfers, with trade-offs being made between end-users' privacy and the commercial value attributed to the value of their attention by the platform operator, the Appeal Court's Judgment will no doubt prove to be an important addition to the jurisprudence. Having said that, the proposed amendments to the Act against Restraints of Competition before German Parliament which were tabled in January 2020, suggest that the views of the Appeal Court might ultimately be moot. 66

Market for social networks in Germany
The Cartel Office concluded that there was little direct competition from rival social networks. Linguistic preferences were critical to the identification of the relevant geographic market 1) Exploitative abuse of consumers through the gathering and combination of data from various sources (including from third parties) without obtaining users' explicit consent (users were faced with a 'take it or leave it' policy).
The abuse under Section 19 of the GWB was based on the breach of data protection rules, and treated as "unfair trading conditions" for antitrust purposes 2) Facebook's data collection from its other corporate services (WhatsApp, Instagram) and from third party websites or apps should only be possible under conditions of voluntary consent * Contrary to EC practice, which has focused to date on exclusionary behaviour rather than exploitative behaviour * First application of logic of Court of Justice in Allianz Hungary Case, Case C-32/11 (2013) * Driven by understanding that Facebook was an 'unavoidable trading partner' and that its customers were 'dependent' upon it 1) The condition in Section 19 GWB, to the extent that a weaker party was subject to disadvantageous provisions because of its incompatibility with general legal principles, has no counterpart in Article 102TFEU. This focus on unbalanced negotiating position also applied to data protection law 2) Wide interpretation to the meaning of "freely given consent" under Article 6(1) of the GDPR, given Facebook's bargaining position 3) Finding that Facebook's data processing was not necessary to achieve the purpose for which it was supposedly undertaken 4) Reasonable user would not expect data from Facebook, WhatsApp and Instagram to be integrated (i.e., separate registration, user profiles and company profiles) 5) Notion that behaviour was abusive because it raised entry barriers, thereby strengthening dominance 6) Investigation can just as readily focus on the exploitative abuse on the customer-facing side of the market * Cease and desist order, with Facebook given 12 months in which to adapt its policies, after remedy proposal lodged within 4 months (effective functional split of data within Facebook companies) 66 Refer to the text of the draft amendment, published at: https://www.wmbl.de/Redaktion/DE/Downloads/G/gwbdigitalisierungsgesetz-referentenwurf.html. The amendments seek inter alia to : (i) subject platforms that are dominant in one product market to a specific list of five prohibitions (including a ban on self-preferencing); (ii) reverse the burden of proof on those operators found to be dominant where their practices hinder interoperability or the ability of competitors to use data acquired in the operator's core market; (iii) extend the notion of "dependence" so that it includes 'relative market power', thereby allowing the Cartel Office to challenge those platforms which act as gatekeepers that abuse that power by denying the access of customers to certain markets; (iv) create a concept of "intermediation power", which can be actionable alongside more traditional concepts such as buying or selling power; (v) modernize the 'essential facilities' doctrine so that access to the data held by digital platforms can be rendered more simply; and (vi) lower the current legal standard used to establish the availability of interim relief. failing to establish dominance in the online advertising market or that Facebook's market power had been used to exploit users' data; and  not distinguishing between what was or was not appropriate in the collection, processing and use of consumer data and the excessive collection of such data.
A further appeal is pending.

Post-Google investigations
In the aftermath of the Commission's Decisions in relation to Google -and without prejudice to a number of further investigations widely understood to have been already lodged in relation to other Google practices -the Commission is also reviewing two high profile complaints in relation to fundamentally different digital platforms.

The Amazon investigation
It was confirmed in July 2019 that the Commission had opened a formal investigation into Amazon's alleged preferential treatment of its own products at the expense of merchants using its sales portal through the use of merchant sales data, facilitated by: (1) Amazon's standard form contracts with merchants; and (2) the role played by data in the choice of the "buy box" which allows consumers to add items from specific retailers into their shopping carts. 67 In parallel, Amazon has announced changes to its third party seller service agreements in light of another antitrust investigation by Germany's Cartel Office, which are designed to improve the contractual positions of merchants. 68 Public announcements confirm that, at the heart of the Commission's investigation of Amazon's commercial practices is the relevance of its dual role as a marketplace platform and as a competing retailer on that platform. Insofar as the Commission pursues this theory of harm, it would appear to be a particular application of the Commission's self-preferencing (i.e., discrimination) theory of harm explored in the Google Search Case. As noted above, such an approach raises issues similar to the concerns which prompted EU policymakers to regulate access by electronic communications providers in ways which avoided both internal and external forms of discrimination. 69 In addition, the parallel investigation brought by Germany's Cartel Office against Facebook suggests that German antitrust 67 See Commission Press Realase of 17 July 2019, available at: https://europa.eu/rapid/press-release_IP-19-4291_en.htm. 68 Bundeskartellamt, Amazon (online sales), B2-88/18, 17 July 2019. In this regard, refer also to the recent antitrust investigation launched by India's Competition Commission on 14 January 2020 into the practices of Amazon and Flipchart in terms of their discounting, exclusive brand launches and preferential treatment which they have allegedly offered to certain mobile phone sellers. See economictimes.indiatimes.com. Refer also to the Competition Commission's findings in this regard in its Market Study on E-Commerce in India, op. cit., esp. at para 86. 69

The Apple investigation
The announcement was made on 3 June 2019 that the Commission was considering its response to a complaint lodged by the music streaming service Spotify lodged earlier in the year, to the effect that Apple allegedly: (1) unfairly limits competitors in their access to the Apple Music streaming service; and (2) imposed a 30% fee which it levied on content-based service providers for using Apple's in-app purchase system (IAP) for any subscriptions sold in its Apple store. 71 In the public statements of Spotify, the levy is described as a "tax" which allegedly unfairly targets music subscriptions while excluding other apps such as Uber or Deliveroo. 72 In addition, Spotify also alleges that Apple prevents it from implementing "experience-enhancing upgrades" related to various functions/apps supported by Apple. In its response, Apple has emphasised that its App Store "ecosystem" provides the very platform and support framework which allows Spotify to pursue its business model and that it should be expected to make an appropriate contribution to sustain that ecosystem. 73 While one can identify in the Spotify complaint some lines of antitrust challenge which are similar to those raised in the Google Cases and the Amazon Complaint (namely, the alleged favouring of its own Apple Music service), the context appears to be different. First, the differentiation practised by Apple is based on differences between apps for payment and free apps, as opposed to alleged discriminatory treatment between paid apps. 74 Second, Apple operates a closed ecosystem that generates economic efficiencies, which means that there are differences between the ways in which Apple's ecosystem operates when compared to the digital platforms of parties operating open, merchant platforms.
While these precedents suggest that competition rules may, indeed, be sufficiently durable in many cases to be able to address many of the policy concerns that have also been expressed by regulators, there are a number of issues endemic to the operation of digital platforms in relation to which traditional competition law tools may require adaptation, to which we turn in Section 3.

Preliminary conclusions
The theories of harm put forward by the Commission are based on well-worn antitrust theory or regulatory practice. The final resolution of the investigations into various digital platform/Internet ecosystem practices by the Commission should clarify the extent to which traditional competition rules require any re-tooling to be able to address the particular types of competitive concerns raised in digital markets. One should not be surprised if any reluctance on the part of the European Courts to endorse the Commission's antitrust enforcement policy is met by European legislators with recourse to instruments of ex ante regulation. What is interesting, however, is the support received for: the Commission's self-preferencing theory of harm from the recent Decision of the French Competition Authority; and the German Cartel Office's policy of 'ringfencing' data so that it cannot be shared 70  between the various business units of a social media company, a policy position confirmed by a very recent CMA policy statement and reflected in existing telecommunications sector regulation.

The Possible Need for Complementary Regulation
Irrespective of whether or not competition rules are effective in curbing anti-competitive behaviour in the market, they are unquestionably less effective in addressing the sorts of problems that might arise where competition is essentially for the market. This is because markets may have already "tipped" before effective intervention is possible, or they may be characterised by market failure rather than abusive strategic market behaviour, or even because customer inertia reinforces entrenched market positions and renders those positions less contestable. In each of these situations, it may be ex ante regulation rather than competition law that is best placed and most effective to address competition concerns. Moreover, it is precisely because these situations are so difficult to address through the application of the essential facilities doctrine and the inherent limits of competition policy to mandate interoperability, that regulation may appear to be the most viable means of intervention.
Thus, in parallel with competition rules, several complementary regulatory tools already apply in the digital sector and beyond. Some rules apply asymmetrically to those undertakings designated as having a level of market power which is so significant and entrenched that it cannot be addressed effectively by competition law. Other rules apply symmetrically to all undertakings in a given sector which meet specific defined conditions and aim at remedying market failures.
There is also the possibility that 'smart' versions of symmetric regulation can be adapted. Thus, while recognising that a broad range of market players needs to satisfy certain basic obligations, these types of regulations could also identify smaller market actors as being capable of exemption from the full weight of regulation by reason of their relatively small impact on the market. 75 The electronic communications sector provides a series of relevant examples of both symmetric and asymmetric regulatory models for intervention, driven by the particular nature of the identified competition problems that need to be addressed.

Asymmetric regulation for electronic communications networks and services
EU policymakers in the electronic communications sector, which displays certain common characteristics with digital platforms (which sit above electronic communications networks in the digital value chain), have opted to subject dominant network providers and service providers to regulatory obligations which are not only inspired by a competition policy analysis, but which are also designed to work in tandem with the application of competition policy.
In addition to the symmetric obligations outlined below in Section 2.3.2 which apply to all undertakings that satisfy certain legal conditions, a range of asymmetric obligations apply to those undertakings in relation to which NRAs have determined that they enjoy Significant Market Power ("SMP"). The SMP designation is inspired by the dominance standard used under EU competition rules. The procedure undertaken is as follows: 76 75 This can occur, for example, with the exclusion of certain firms from regulatory obligations on de minimis grounds (for instance when the platform is a small entreprise accrding to EU law as occurs in Art.11(5) of Regulation 2019/1150 on promoting fairness and transparency for business users of online intermediation services) or through the use of differentiated remedies which can take due account of the market circumstances of smaller operators (as has occurred since 2002 under the decisional practice of NRAs when applying the EU regulatory framework for electronic communications: art.68(4) EECC). 76  (ii) a market structure which, taking account of those barriers to entry, does not tend towards effective competition within the time horizon of the market analysis (usually three, but now up to five years); and (iii) the insufficiency of competition law alone to address adequately the market failure(s) identified under the first two criteria. 77  Second, the NRA defines the product and geographic boundaries of the selected markets.
 Third, the NRA then determines whether one or several undertakings enjoy an individual or collective dominant position on those identified markets and, when that is the case, designates them as holding SMP status.
The rationale behind the SMP regulatory framework is the establishment of a competition law-style of analysis when imposing regulatory obligations. The asymmetric obligations are imposed on undertakings with market power, with regulation being as much driven by the need to ensure connectivity and interoperability because of the structural nature of electronic communications networks as the desire to regulate 'essential facilities' (including a symmetric obligation ensuring interconnection between all market actors, along with number portability and access to the inside wiring of homes).
In turn, asymmetric obligations must be imposed under the EECC by reference to an established set of remedies, including mandated access obligations, the fixing of tariffs for termination services, nondiscrimination obligations, transparency obligations, accounts separation obligations and cost accounting obligations, and even the functional separation of business units where the more traditional access-focussed remedies are found to be ineffective in promoting effective competition.

Symmetric regulation for electronic communications and digital services
Symmetric rules are aimed at resolving the competitive concerns that arise from various types of dependency relationships, or the need to ensure interoperability. Accordingly:  The EECC imposes on providers of number-independent interpersonal communications services a range of obligations in order to render their services interoperable, including by relying on standards if: (i) those providers reach a significant level of coverage and user up-take; (ii) the Commission has identified an appreciable threat to end-to-end connectivity between end-users and has adopted implementing measures specifying the nature and scope of any obligations that may be imposed by the National Authorities; and (iii) the obligations imposed are necessary and proportionate to ensure the interoperability of interpersonal communications services. 78  In addition, the EECC imposes on providers of Conditional Access Systems (CAS) providing access to digital television and radio services, and the access services upon which broadcasters depend to reach any group of potential viewers, the obligation to offer to all broadcasters on FRAND terms technical services enabling the broadcasters' digitally-transmitted services to be received by viewers or listeners and to keep separate financial accounts regarding their activities as CAS providers. 79  In turn, the Open Internet Regulation imposes a range of non-discrimination requirements on all Internet Services Providers (i.e., Net Neutrality obligations), in terms of the way they handle Internet traffic, without reference to whether or not they hold market power. 80  The P2B Regulation imposes transparency and non-discrimination obligations on the providers of online intermediation services and, to a lesser extent, on the providers of online search engines. 81  The PSD2 Directive 82 subjects providers of payment services to a number of obligations regarding the release of personalised security credentials of their customers to other payment service providers.
All of these legislative forays into market dynamics have been driven by the acknowledgement that traditional competition tools are compromised when dealing with various examples of market failure brought about by a combination of shifting market boundaries, indirect network effects, customer acquiescence and information asymmetries with their digital providers, and varying levels of dependency on key players which undermine the exercise of effective countervailing bargaining power. The policy imperative for introducing such legislative changes is more compelling where interoperability is desirable and where the fruits of enduring market power are no longer driven by competition on the merits. Insofar as these regulatory regimes promote the principles of transparency and non-discrimination, they are geared towards ensuring that market failures do not materialise or are not exacerbated, whereas data portability measures are primarily designed to ensure that markets remain contestable. 78 Article 61(2c) EECC. As noted by the Commission, this need could arise from a significant decline in usage of particular numbers-based communications systems, so that the public interest in end-to-end connectivity can no longer be assured through that system -either because a single number-independent Inter-personal Communications Services becomes the predominant mode of interpersonal communication or because of market fragmentation characterised by a large number of different, non-interoperable communications applications: Executive Summary of the

Participatory design of remedies
When an information asymmetry exists between competition or regulatory authorities, on the one hand, and regulated firms, on the other, or when the effects of proposed remedies are uncertain, the authorities might also involve the regulated firms in the design of the remedies. This involvement should obviously be addressed cautiously and with sound safeguards being put in place, given that the natural incentives of the regulated firms in the case of abuses of market power and market failures is to be able to satisfy the lightest possible remedies. In recent competition cases, several interesting approaches have been developed which may prove to be successful in digital markets if appropriately adopted, as is discussed below.

Standard Essential Patent (SEP) Cases: Access on FRAND terms and good faith negotiations
Under an SEP regime, undertakings that participate in the setting of a standard might own "essential patents", with respect to which a licence is required by anyone seeking to participate in the market in compliance with that standard. In such circumstances, owners of SEPs have a very strong bargaining position vis-a-vis their competitors who are working with the standard.
In order to prevent the phenomenon of licence "hold-up" from SEP owners, the SEPs are licensed on FRAND terms in order to guarantee access to the patent in question, despite the fact that its ownership resides exclusively in the patentee. This form of licensing is considered to strike a reasonable balance between the needs of competitors and the importance of rewarding the patent holder so that they continue to invest in R&D and standardisation activities. 83 According to the Commission's SEP Communication, 84 a standardisation agreement of this kind is generally considered to be compatible with Article 101 TFEU because of the fundamentally competitive impact of the FRAND licensing regime in the relevant market. However, where the patentee departs from FRAND terms, its behaviour might constitute a violation of Article 102 TFEU.
The Commission's approach was first expressed in 2014, in its Motorola Decision, when it found that it was anti-competitive for an SEP holder to seek to exclude competitors from the market by seeking injunctions against them if it had already committed itself to licensing its SEP on FRAND terms. 85 This approach was taken a step further that same year in the Samsung Decision, 86 where the Commission obtained concessions from the SEP owner committing: (i) not to seek an injunction for a period of at least five years; (ii) prescribing a maximum negotiation period of 12 months; and (iii) establishing a dispute settlement mechanism in those situations where the parties could not agree to FRAND terms. This approach was endorsed by the Court of Justice in 2015 in the Huawei v ZTE Case, 87 where the Court's rationale for holding that the failure to license an SEP on FRAND terms was based on the principle of "legitimate expectations". Accordingly, without prejudice to a patentee being able to enforce its legitimate IP rights, the commitment to license an SEP on FRAND terms carries with it certain consequences for the patent holder -in certain circumstances, the pursuit of an injunction to enforce an SEP may of itself amount to the abuse of a dominant position.
When the English High Court was asked in 2017 to determine whether an SEP owner's licence terms were FRAND, 88 it determined this question by reference to three potential royalty benchmarks, namely: a rate that is governed by the value of the patentee's portfolio; terms in comparable licences; or crosschecking to determine the patentee's share of relevant SEPs, and applying that share to the total aggregate royalty for a standard.
The approach taken in the EU thus far in relation to SEPs might prove to be very instructive in the digital platforms context. Analogous to the SEP situation, a digital platform provider's commitment to abide by a set of normative rules set forth in a Code of Conduct might be capable of being treated as the equivalent in competition terms of a commitment to license on FRAND terms in an IP context. Where the refusal of digital platform provider to abide by the terms of a Code of Conduct has the potential to result in foreclosure, the quasi-regulatory approach adopted in relation to injunctions being sought to enforce SEPs would appear to apply by analogy. In effect, it would mean that the onus of proof as to whether or not the conduct question was anti-competitive would no longer lie with the Competition Authority but with the digital platform. This is because a material departure from the express terms of a Code of Conduct would be presumed to result in anti-competitive effects, given the circumstances in which such a Code was adopted.
The analogy with an SEP scenario is strengthened by the fact that comparable access regulation imposed on electronic communications sector operators is inevitably premised on the understanding that access terms will be non-discriminatory. 89 The potentially normative nature of obligations set forth in a Code of Conduct is, in turn, supported by the fact that a raft of obligations already set forth in the P2B Regulation insist upon the fact that digital platform providers, if discriminating against their competitors, must satisfy transparency standards which at least explain why such discrimination is taking place. 90 Beyond these general principles, however, some form of dispute resolution mechanism will need to be considered in order to resolve effectively the inevitable disputes that would arise in relation to the scope of the FRAND remedy.
In the Huawei Case, the Court of Justice imposed, under a competition law analysis of an SEP situation, a dispute resolution framework that involved several steps for good faith and timely negotiation between the various market stakeholders: 91 first, the facility owner should present a written offer specifying, in particular, the access price and the way in which it is to be calculated; (ii) second, the access seeker should diligently respond to that offer, in accordance with recognised commercial practices in the field and in good faith and, if they cannot accept the offer, should submit promptly and in writing a specific counter-offer; and (iii) finally, if the parties cannot agree, they may, by common agreement, request that the price be determined by an independent third party, by decision without delay.
This type of negotiation framework could also serve as the basis for a remedy where the criteria of the threshold test have been satisfied. 89 The concept of non-discrimination is usually associated with the FRAND standard while the FRAND standard is itself usually associated with an overarching idea of 'fairness' or 'equity' while a traditional non-discrimination remedy under competition rules is relatively agnostic when it comes to the notion of 'fairness'. 90 Article 7 of the Regulation 2019/1150 on promoting fairness and transparency for business users of online intermediation services. It follows, however, that if such discrimination is unjustified objectively, it might lead to a competition infringement claim. 91 Huawei Technologies (C-170/13), at paragraphs 55-71. Article 60 of the EECC also imposes a good faith negotiation obligation in case of access and interconnection disputes in telecommunications. In TeliaSonera Finland (C-192/08) EU:C:2009:696, paras 36 and 51 to 55, the Court of Justice decided that an NRA may consider that the good faith negotiation obligation has been breached where the facility owner proposes to the interconnection seeker unilateral conditions likely to hinder the emergence of a competitive market at the retail level. Robert Schuman Centre for Advanced Studies Working Papers

Voluntary commitments
Competition law cases may be resolved with voluntary commitments proposed by the investigated firms and accepted by the Commission after they have been subject to a market test. 92 A resolution of the competition concerns by the offer of commitments was attempted in the Google Search Case, but these commitments were ultimately abandoned in favour of a prohibition Decision being adopted. Commitments may resolve competition issues more quickly than the adoption of prohibition decisions, which are often appealed. Commitments often also benefit from the insights of the investigated firms. However, while they require a very rigorous assessment process by Competition Authorities, given that they are usually not appealed, they do not contribute to the establishment of a clear case-law on what is permitted and what is prohibited by competition law. Most importantly, it is notoriously difficult for Competition Authorities to enforce behavioural obligations strictly unless they take the form of commitments, whose breach can at least be prosecuted separately. 93 The possibility for such commitments has also been introduced in the EECC with a procedure inspired by competition law. 94 Under that procedure: (i) the SMP operator may propose commitments provided they are sufficiently detailed, in particular with regard to the timing and scope of their implementation and their duration; (ii) thereafter, the NRA assesses those commitments, particularly in terms of whether they are fair and reasonable, as well as whether they are open to all market participants, whether they ensure the timely availability of wholesale access under FRAND terms and, more generally, whether they enable sustainable competition in downstream markets. For the purpose of this assessment, the NRA performs a market test by conducting a public consultation with interested parties, in particular those which are directly affected; (iii) the NRA then communicates to the SMP-designated operator its preliminary conclusion and the operator may revise the commitments accordingly; (iv) when the NRA is satisfied that the commitments comply with the objectives and the criteria being assessed, it may adopt a decision to render the commitments binding, wholly or in part, for a specific period (which may be the entire period for which they have been offered); and (v) the NRA then monitors compliance with the commitments that it has made binding and, in the case of a failure to comply, imposes penalties. The relative importance of voluntary commitments in addressing competition concerns in relation to digital platforms lies at the heart of the proposals for reform supported by the Competition Authorities of the Benelux region.

Competition Law: the need to adapt enforcement principles
Notwithstanding the existing examples of antitrust intervention by the European Commission in relation to digital platforms (see Section 2.2 above), antitrust analysis raises particularly complex issues regarding the essential analytical building blocks of any infringement actionnamely, the processes of market definition and the assessment of market dominance. 95 This is not only because of the multi-sided nature of digital platforms, 96 but also because so many of the transactions conducted over such platforms do not involve the transfer of monetary consideration. 97 Moreover, intervention is rendered even more complex because of the dynamic nature of competition in relation to such markets and the need to weigh ambiguous competing harms against potential distributional efficiencies, rather than reliance on the usual more static models of competitive harm.
These complexities inevitably mean that the assessment of competitive harm is a much more arduous exercise than might otherwise be the case in relation to markets with more common patterns of supply and demand. 98 As a result, the assessment of competition concerns is, in the minds of many policymakers, not sufficiently rapid to be able to arrive at meaningful decisions before markets "tip" and are no longer susceptible to the stimulus usually afforded by new entry. It is thus not surprising that many of the recently published reports on digital platforms conclude that important analytical short-cuts need to be explored in order to be able to circumvent the dangers created by late and/or ineffective intervention. A number of these proposed shifts in enforcement strategy are discussed below.

The assessment of market power
To the extent that competition rules are relied upon for intervention, Competition Authorities should, as suggested in the Cremer Report, prioritise the potential for market power generating competitive harm arising in digital platforms, and determine whether there are any realistic constraints on such market power being exercised, rather than over-emphasizing the traditional first analytical step of determining the existence of a relevant product market to which the antitrust allegation must relate. However, jurisprudence has developed in such a way over the past sixty years that it would be difficult to conceive the market definition exercise either being removed altogether or significantly diminished in importance in the assessment of competitive harm. Having said that, the authors feel that it is also the case that existing European Commission practice provides a number of alternative avenues of market power analysis whose relative importance in a digital platform/Internet ecosystem context have yet to be fully explored.
3.1.1 Digital platforms and Internet ecosystems: Market power measured by reference to conglomerate effects Many digital markets are connected either on the input side where the capabilities may be shared to develop different products (shareable inputs in modular design product development) and/or on the output side where products may be connected within the same ecosystem. 99 To capture some of the characteristics of the range of available connected products, one of the practices has been to define "after-markets" or a single functional market for "systems". For instance, in the CEAHR Case, the General Court decided that: "(…) to be able to treat the primary market and the after markets jointly, possibly as a single unified market or 'system market', it must be shown (…) that a sufficient number of consumers would switch to other primary products if there were a moderate price increase for the products or services on the after markets and thus render such an increase unprofitable (…)". 100 97 As has been acknowledged under EU law, the absence of monetary consideration will not prevent the definition of a Similarly, in its Market Definition Notice, the Commission states that: "A narrow definition of market for secondary products, for instance, spare parts, may result when compatibility with the primary product is important. Problems of finding compatible secondary products together with the existence of high prices and a long lifetime of the primary products may render relative price increases of secondary products profitable. A different market definition may result if significant substitution between secondary products is possible or if the characteristics of the primary products make quick and direct consumer responses to relative price increases of the secondary products feasible." 101 By contrast, given that the definition of markets is problematic in a digital platform context because the dominant undertaking is present on a number of neighbouring or adjacent markets whose edges are often anything but clear, it may be appropriate in some circumstances to identify dominance in one discrete product market which is critical to the fulfilment of the digital platform's business model 102 while also identifying its intrinsic links to other potential markets in question. Such links might consist of at the supply-side level, the use of common or shareable inputs to develop different products or services; and at the demand-side level, the sale to the same customers of different products or services which are rendered inter-operable.
This would require a more nuanced analysis of relevant "markets' that is intrinsically linked to the idea that market power is likely to be exercised because of the connection between markets or submarkets, coupled with the fact that many such markets or sub-markets are linked through the manipulation of common, or overlapping sets of data. Such an analysis can be identified not only in the administrative practice of the European Commission when assessing mergers with conglomerate effects, 103 but also under its significant body of administrative practice under Article 106 TFEU. 104 Examples of commercial links which can trigger the exclusion of "as efficient" competitors include: product and/or service proximity which facilitates bundling and tying practices; exclusive dealings in other areas preventing customers from effectively switching; the definition of a single functional market covering both upstream and downstream levels. The Guidelines first explain that "the purposive nature of market definition can require the product or geographic dimension of a market to be extended beyond what can be substituted for products of the merger parties to include other functional levels in the vertical supply chain or other products that are typically purchased or supplied together with those of the merger parties". The Guidelines then explain that "where merger parties are vertically integrated or compete against vertically integrated firms, the ACCC must determine whether competition analysis is best conducted in the context of one relevant market encompassing the whole vertical supply chain or a series of separate markets each comprising one or more stages of the chain. This delineation depends on the economics of integration. Importantly, there need not be trade between the relevant stages of the vertical supply chain for there to be separate markets-the potential for exchange can be sufficient. However, where there are overwhelming efficiencies of vertical integration between two or more stages in the vertical supply chain, the ACCC will define one market encompassing all those stages",. 102 It being well understood that the business model driving a ride-sharing platform such as that of Uber differs greatly from the business model adopted by the search function performed by Google. 103 A. Lindsay and A. Berridge , The EU Merger Regulation: Substantive Issues, 5 th ed, Sweet & Maxwell, 2012, discussion in Chapter 12, where the authors emphasise in particular the role of "must have" products and the portfolio range of products which act as a magnet for the offers of bundled services by the dominant merged firm. 104 While one should not equate the existence of a statutory monopoly with a dominant position earned by a private firm in a competitive environment, there is nothing to suggest that the Article 106 practice that has developed over the years is not as persuasive as anything developed under Article 102 as regards the incentives of a dominant firm to leverage its dominance into related markets, consistent with an approach based on conglomerate effects. The application of the "leveraged dominance" concept under Article 106 TFEU is, however, broader than its Article 102 TFEU counterpart, insofar as it relies only on structural market considerations and does not require that anti-competitive effects be proved. breadth of a product portfolio which incentivises discrimination (or self-preferencing) in favour of its own services; discrete product niches where predatory pricing can be profitably engaged in, subsidised by profits in the principal market; margin squeeze practices, where the platform operator minimises the gap between a wholesale and a retail offering; refusal to supply or license a critical input, which is important to being able to compete effectively in a downstream market; use of superior access to data by merchant platforms to affect downstream competition adversely in online retailing; anti-competitive strategies to lock-in customers, driven by the lack of interoperability or interconnection between platforms; and the use of Internet Protocols in ways which aggregate more information in the hands of a small group of operators, effectively undermining the multi-layered architecture of the Internet.

Figure 1: Market Analysis where Markets are Connected
By focusing on the links between markets or market segments and, in turn, on the incentives available to operators to take advantage of those links through leveraging strategies, such an approach would remove (at least in part) the need to engage in complex fact-finding about the scope of all of the other markets falling within the scope of the digital platform in question. The identification of such links would also add weight to a theory of harm based on the concept of leveraging, which is an overriding concern in such situations, and which has played a critical role in the administrative precedents adopted by the Commission to take into account its enforcement strategy under Article 102 TFEU.
The proposed approach would draw upon the logic behind tying and bundling offences (which relies on the links between different products and services) under the Commission's assessment of conglomerate mergers and margin squeeze cases (which rely on the link between the wholesale and Robert Schuman Centre for Advanced Studies Working Papers retail functional levels of competition) effected by dominant undertakings under Article 102 TFEU, 105 and both of which form part of the administrative practice of the Commission.
At the heart of any fundamental re-calibration of EU antitrust tools to deal with digital platform/Internet ecosystem issues is the notion that leverage theory should be understood to operate in a manner that is consistent with the foreclosure incentives usually associated with conglomerate effects under a merger review. Although the logic of conglomerate effects in an Article 102 TFEU context has no administrative precedents as yet, there is a wealth of precedents from which competition law authorities can draw from Article 106 TFEU and from the Commission's merger practice under the EU Merger Regulation.

More dynamic analysis and focus on potential competition
When assessing the existence of market power, Regulatory Authorities and Competition Authorities are also entitled to shift part of their focus from existing to potential competition. In that regard, the Commission notes in the Horizontal Merger Guidelines that: 106 "For a merger with a potential competitor to have significant anti-competitive effects, two basic conditions must be fulfilled. (i) First, the potential competitor must already exert a significant constraining influence or there must be a significant likelihood that it would grow into an effective competitive force. Evidence that a potential competitor has plans to enter a market in a significant way could help the Commission to reach such a conclusion. (ii) Second, there must not be a sufficient number of other potential competitors, which could maintain sufficient competitive pressure after the merger." Indeed, the essence of ex ante regulation and merger reviews 107 is to be forward-looking in their analysis to be able to take into account potential competition, with any Article 102 TFEU analysis also taking due account of the likely impact of new entry. Although the loss of potential competition under EU competition law practice does not match the level of intervention capable of being exercised by US antitrust authorities where they feel that there is an "attempt to monopolize" under Section 2 of the Sherman Act, it cannot be denied that the legal basis for intervention already exists under EU competition law where potential competition is allegedly being foreclosed.
However, the lack of precedents in this area has meant that we have little empirical guidance on how to measure the potential loss of future competition. 108 Perhaps this empirical gap in our analysis can be filled, at least in part, by the model developed by Michael Porter, who has proposed to apply his famous five competitive forces analysis 109 in the determination of market power in antitrust. In doing so, the author shows that, in conjunction with the current rivalries captured by market shares and concentration

Designing an EU Intervention Standard for Digital Platforms
European University Institute 27 ratios, potential competition by new entrants in offering the same product or substitute productscaptured by the level of entry barriers and the degree of market contestabilityis also a key element in understanding the intensity of competition. 110

Defining and analysing innovation markets
In applying Porter's "five forces" style of analysis to digital platforms, one needs also to take account the role played by innovation, as any assessment of welfare should be dynamic in nature, rather than focusing on static observations. A focus on the dynamic nature of competition need not, however, be considered too speculative to yield tangible results. 111 In practice, more dynamic criteria have already been designed for innovative sectors. As regards market definition, the Commission has developed the concepts of competition in innovation markets which refers to R&D poles which may compete between each other depending on the "the nature, scope and size of any other R&D efforts, their access to financial and human resources, know-how/patents, or other specialised assets as well as their timing and their capability to exploit possible results". 112 The Commission observes that R&D poles may be identified when the process of innovation is well structured, as occurs in the pharmaceutical industry, but that the concept will normally not be used when the process of innovation is not clearly structured. In Dow/DuPont, the Commission also defined the concept of an innovation space, which is "not a market on its own, but an input activity for both the upstream technology markets and the downstream products markets". 113 In relying on those concepts, the first step is to determine whether market power is being exercised upstream from existing products/outputs to the capabilities/inputs necessary to develop and diffuse improved or totally new products.
Therefore, a dynamic standard analysis based on products/output markets can be complemented by an analysis based on capabilities/input markets, in order to better reflect upon the importance, the rate and the uncertainty of innovation and the key role played by innovation capabilities. As has been suggested by David Teece, "when innovation is high, capabilities are more stable than products." 114 This type of approach is nevertheless challenging because the characteristics of innovation capabilities and their role in product innovation are complex, particularly in those industries where the innovative process is not clearly structured. However, as has been convincingly argued by Wolfgang Kerber and Benjamin Kern, 116 there are existing, sufficiently robust theories and empirical studies in strategic management, evolutionary economics and innovation economics disciplines to allow us to construct methodologies to define innovation markets of sufficient relevance and with the appropriate degree of legal certainty. As David Teece observes: "the tools for assessing capabilities may not be well developed yet, but they are developed enough to allow tentative application. Clearly, product market analysis can be unhelpful and misleading in dynamic contexts. Using the right concepts imperfectly is better than a precise application of the wrong ones." 117 This new approach may be more difficult to apply in the digital sector than, say, in the pharmaceutical sector, because the innovation process is less structured and shorter in the former than in the latter. However, it is not an impossible task, and innovation markets may be defined by reference to the main capabilities of the digital sector such as data, certain types of engineering skills, high computing power and highly risky capital ventures.
When assessing market power, antitrust authorities should focus their attention primarily on the barriers to accessing those key capabilities. To do that, authorities can rely on the four conditions identified by Jay Barney for a firm's resources to be considered as a source of competitive advantage (i.e., inimitable, rare, valuable and non-substitutable). 118 As regards barriers to data, Anja Lambrecht and Catherine Tucker have already applied the four Barney conditions. 119 However, their conclusions are arguably too general, as the conditions need to be assessed for specific datasets used for specific types of algorithmic applications. As regards barriers to risk capital and skilled labour, Nicolas Petit has proposed a close antitrust scrutiny of the exclusivity clauses in corporate venture capital funds and of the non-compete clauses used regularly in labour contracts for highly skilled staff. 120 The antitrust economics and strategic management literature suggests that markets characterised by innovation should not be dumped into the "too hard" basket alongside markets which are characterised by competitive interactions across many sides of a digital platform. Lessons can be learned from the fact that there is a body of literature developing from different innovation economics sources which is capable of being harnessed by competition authorities to serve as the basis for an analysis of digital markets from the perspective of the "innovation markets" model.

The Assessment of the Anti-competitive Practices -Theories of Harm
Based on the available literature and EU precedents, we need to consider the extent to which competition rules are sufficiently adaptable to be 'fit for purpose' and the extent to which these procedural rules which support them are too restrictive. In contrast to what has been suggested, 121 we would be reluctant to recommend a sweeping changes to the onus of proof for abusive practices under Article 102 TFEU, as this may be very difficult to reconcile with sixty years of jurisprudence of the European Courts and would also run counter to the presumption of innocence which prevails under human rights legislation. The United Kingdom (UK) arguably has greater flexibility in changing this critical procedural rule than would the EU 27 in relation to Regulation 1/2003, especially since the standard of review by the UK courts is currently on the merits and not merely on the basis of the lighter standard of judicial review (i.e., "manifest error") that applies before the EU Courts. We would, however, suggest that there is a need for the Commission to articulate clearly its various theories of harm in the context of digital markets. To this end, the updating by the European Commission of its existing Market Definition Notice and its Enforcement Priorities Guidance seems to be necessary.
However, one can nevertheless draw a distinction between the "ultimate burden" of proof in establishing the anti-competitive effects of a practice and the "evidentiary" burden on a defendant to provide sufficient evidence in relation to any issues which a party may believe to be material in the determination of the ultimate factual issue under investigation. In other words, it is still the case that the defendant (or a notifying party, in the case of a merger), cannot keep to itself information which it believes to be relevant to the competitive appraisal of its actions. 122

Bundling, envelopment strategies and conglomerate effects in digital ecosystems
While not proposing to modify the existing obligation on the European Commission to prove its case, we would nevertheless propose that the dynamic of digital platforms is more consistent with the exploration of theories of harm whose roots lie in the concept of conglomerate effects. Accordingly, it makes sense for the Commission to build its case under Article 102 TFEU under this theory.
Conglomerate competition concerns might arise in a merger context, for example, where the merging parties are actual or potential competitors or where they produce goods or services which might be upstream or downstream from one another. Conglomerate mergers may also generate positive effects which increase consumer welfare. In some circumstances, however, conglomerate mergers may be harmful to consumers through the marginalisation or elimination of competitors. 123 This requires an understanding of the merging group's incentives and capabilities to pursue foreclosure strategies in relation to actual and potential competitors, ultimately to the detriment of consumers. Such an analytical framework seems most appropriate in a digital platform / Internet ecosystem context. The Commission's Non-Horizontal Merger Guidelines 124 provides guidance on the approach taken in the assessment of conglomerate mergers. The classic example of a conglomerate merger is one which involves the combination of producers of complementary products or services, with the principal antitrust concern being that the merged entity will be able to foreclose competitors through the leveraging of its market power from one market ("the leveraging market") into another market ("the leveraged market"). This inevitably involves an analysis of the connections between two markets, a determination of whether a sufficient degree of market power exists in one of these markets, 125 and the likely negative effects on consumers brought about by the resulting foreclosure of competitors, while also taking due account of the efficiencies that might be produced by the investigated conduct. The customer bases for the two products or services must be linked in such ways that facilitate a foreclosure strategy that is capable of affecting demand for the leveraged product. In order to be actionable, the foreclosure strategy must be shown to be capable of impairing the ability of actual or potential  126 In its analysis, the Commission takes into account the ability of other market participants to deploy an effective and timely counter-strategy. 127 Bundling, tying and mixed bundling strategies are classic examples of leveraging strategies. There is a growing body of economic literature which suggests that tying or bundling practices may be profitable for exclusionary reasons. 128 Other leveraging theories might involve exerting pressure on customers to buy a full portfolio of services, 129 exclusive dealing 130 or the cross-subsidisation of its product range generated by its promotion of "must have" products. 131 In digital markets, a specific form of anti-competitive bundle is what is what Eisenmann, Parker and Van Alstyne refer to as "platform envelopment". 132 According to the authors, "envelopment" occurs when a dominant platform enters a new market pioneered by that entrant's platform and forecloses the new entrant. Overlapping user bases between the dominant platform's primary market and the new market and shared components or modules between the two products, render entry into the new market feasible for the dominant firm. First, the dominant platform can leverage its customer base from its primary market to the new market, and thus benefit from significant network effects when it enters the new market. These network effects can be of much larger magnitude than those enjoyed by the entrant. Second, because of shared components or modules, operating the two platforms together might generate significant economies of scope.
Within this framework, platform envelopment corresponds to "pure" bundling practices: namely, the dominant platform bundles its existing platform service with a new platform service similar to the platform that it wishes to envelop. 133 Through pure bundling, the dominant platform can then foreclose the rival platform in the target market. When the dominant firm's platform and the target's platform are weak substitutes or unrelated to one another, Eisenmann et al argue that significant economies of scope are a pre-condition for envelopment by bundling to succeed. In digital markets, when platforms have acquired a large customer base in a primary market, they may adopt an envelopment strategy to expand into other markets.
In antitrust practice, the scope of leveraging theory applying in a conglomerate markets context has been given further impetus through a recent precedent adopted in the context of Article 106 TFEU. In the OPAP Case, 134 the Commission, while declining to pursue a case against Greece's de facto monopolist in gaming and betting because the affected market (horse race betting) was not large enough to trigger a "Community interest" to justify the Commission's intervention, the Commission was nevertheless willing to entertain an action based on the ability and incentive of OPAP to leverage its position in horse race betting into other adjacent betting markets. In the context of digital platforms or, even more broadly, digital ecosystems, leveraging in the context of a conglomerate market structure has an important role to play in understanding the nature of potential theories of harm. As noted in the Commission Guidelines on the assessment of non-horizontal mergers, foreclosure effects are "likely to be more pronounced in industries where there are economies of scale and the demand pattern at any given point in time has dynamics implications for the conditions of supply in the market in the future". 136 Bundling may have both positive and negative competitive effects and, in the digital sector, both effects are often amplified. At a general level, the competitive effects of bundling depend on whether bundling creates efficiencies, which may be the case in particular via supply-side and demand-side synergies, and whether those efficiencies are passed on to consumers. This, in turn, depends on the particular market characteristics and conditions at issue, in particular the competition between digital firms and the possibility of consumers moving or multi-homing between those firms.
In our view, conglomerate effects theory, especially given its potential to rationalise tying and bundling practices engaged in by firms with market power, provides a fertile base upon which to build compelling theories of harm in a digital platform context. Such theories have the added advantage of being able to address concerns about the stifling of potential competition, a theory of harm which already finds support in merger practice (albeit very little used to date in relation to behavioural practices). 137 There is also a wealth of administrative practice upon which the Commission can draw from its Article 106 TFEU experience, while also acknowledging that the approach adopted under Article 106 conveys greater prominence to the presumed anti-competitive effects flowing from industry structure (rather than on the actual effects on competition in an action brought under Article 102 TFEU). 138

Access to key capabilities
The control of key capabilities may motivate the formation and expansion of digital conglomerates and may be one of the reasons for their competitive edge when such components form the basis of modular innovation. The control of those key components may lead to different competitive concerns, in particular a refusal to share those components, thereby impeding the entry of innovators.
To the extent that it is possible to identify those key capabilities, one potential remedy is to allow competition to emerge and to ensure market contestability by requiring access to such capabilities. If technically feasible, compulsory access obligations will allow entrants, on the one hand, to enjoy the same economies of scope in product development as the incumbent firm and, on the other, to generate demand-side synergies of similar magnitude when integrating the key components in their product ecosystems.
However, as explained by the Commission 139 and the Court of Justice, 140 compulsory access always involves a trade-off between short-term competition, which it aims to stimulate, and the innovation incentives of the various market players, in particular the dominant firm subject to the provision of access. One should be particularly sensitive to the possibility that the mandating of access to content 137 As noted earlier, this approach also reflects the US approach where "attempts to monopolise" are alleged under the Sherman Act. 138 Theories of potential harm developed under the Article 106 practice of the Commission will need to be complemented by an "effects" analysis that satisfies the standard set forth in the Intel Case (Case C-413/14 P). Whereas the Court of Justice in Greek Lignite (Case C-554/12 P) considered that proof of effects is unnecessary within an Article 106 framework, it should also follow that the opposite is true when an action is being pursued under Article 102 TFEU. 139 Commission Guidance on the enforcement priorities in applying Article 102 TFEU, at para. 75. Robert Schuman Centre for Advanced Studies Working Papers runs the risk of discouraging the content creator / aggregator from innovating and investing in further innovation. This trade-off should be assessed against the specific characteristics of the digital key inputs whose access is being required. 141 It reminds us of the long-held debate in the electronic communications sector regarding the benefits of network-based competition, as opposed to services-based competition (with preference invariably for the former). In certain situations, it might therefore be much more appropriate to mandate some form of interoperability, as opposed to mandating access to content.
If the key input analysed is a dataset, 142 the trade-off between short term and long-term competition should be assessed against the particular characteristics of the data involved, in particular the nonrivalrous nature of the data and the general purpose nature of the technology. On the one hand, the costs of compulsory access are smaller for non-rival products than for rival products because the owner of the former can share them without being deprived of their use. 143 On the other hand, the benefits of compulsory access are higher for general purpose technologies than for other products because of the pervasiveness, the inherent potential for technical improvements and the innovational complementarities of the former. 144 Therefore, when applying the same trade-offs between short and long-term competition and innovation, the conditions for imposing data sharing under competition law might in many instances be lower for data than for other products. Accordingly, Heike Schweitzer et al. suggest that: "the threshold for finding that a refusal to supply data constitutes an abuse may be somewhat lower than the threshold for finding an abuse in cases of a refusal to grant access to infrastructures or to intellectual property rights."' 145

Discrimination and self-preferencing
Insofar as a dominant undertaking operates a digital platform that is open to all traders, acts of discrimination, self-preferencing and other related acts of leveraging may be unlikely to generate efficiencies which outweigh the restrictions to competition arising from such acts. This is the presumption which grounds the legislative prohibition of self-preferencing in the telecommunications regulatory context. By contrast, as noted earlier, the operation of closed platforms of digital ecosystems may be more likely to generate positive efficiencies, unless consumers do not have sufficient competitive options to choose alternative platforms and switching between such platforms is not feasible. 146 Having said that, policymakers are entitled to question whether patterns of commercial behaviour suggest that customers switching between platforms is not viable or attractive, especially where access to key competitive data is not available to all operators. In other words, where the circumstances so justify, presumptions about the vibrancy of inter-platform competition might need to give way to theories of harm which are confined to individual platforms, at least where it is not demonstrated that the efficiencies do not outweigh the restriction on competition that has been identified, and where switching is not effective. This same approach has been taken to its extreme in the field of call termination on the networks of electronic communications operators, where the process of call 141 As it was proposed in the Cremer et al. Report, pages 98-107. 142 Potential other key components include algorithms, software or search technologies. 143 The costs of compulsory sharing in reducing data collection incentives are also lower when the data was collected as by-

Designing an EU Intervention Standard for Digital Platforms
European University Institute 33 termination is considered to be specific to each network that has been allotted its own numbering sequence. 147

Violations of regulatory principles
Given the particular characteristics of many digital platforms, one can envisage that a dominant digital platform provider's persistent departure from normative principles established under regulation or from industry operating standards (especially those established in the context of remedies imposed by competition law or regulatory authorities) might be capable of forming the basis of a theory of harm for an action brought under competition rules. As the AstraZeneca precedent 148 has demonstrated, the existence of regulatory obligations alongside competition rules does not mean that competition rules are no longer relevant. On the contrary, as the Telia Sonera precedent 149 confirms, a regulatory norm can assume the status of a competition norm where the circumstances are appropriate and where the policy drivers for such principles of "regulatory antitrust" are strong. Moreover, as the Telekom Polska 150 precedent demonstrates clearly, the failure of regulation to address effectively what might otherwise be an antitrust violation provides strong policy motivation for antitrust rules to intervene. The history of margin squeeze precedents 151 also pays testimony to the fact that regulatory compliance does not equate to competition law compliance.
It should equally follow that regulatory non-compliance, where it is systematic and unable to be addressed effectively by regulation, can be equated with an antitrust violation where the nature of that violation is capable of foreclosing competitors, whether by impeding their effectiveness or by excluding them from the market altogether. Recent European case-law also confirms that the broader legal context in which a commercial practice occurs (i.e., the existence of regulatory obligation) may be relevant in determining whether an anti-competitive practice has occurred. 152 The administrative practice of the Commission thus already establishes a number of robust principles surrounding the application of traditional theories of harm. Given the unique commercial dynamic of digital platforms, it is open to policymakers to increase the scope of such theories of harm so that they embrace the 'currency' of digital platforms (i.e., data transfers).

Preliminary Conclusions
A review of existing competition law practice suggest that, at the very least, the enforcement competition law principles could be clarified and adapted in the following ways: focusing on potential competition and on the control of the key innovation capabilities such as essential data, computing power and related skills or risky and patient capital.
 For the assessment of anti-competitive practices, the Competition Authorities should focus on the most problematic practices in the digital economy, namely, bundling and envelopment strategies within a specific digital platform or eco-system, refusal to grant access (including interoperability) to key inputs and innovation capabilities, discrimination and self-preferencing, and persistent violations of normative regulatory principles. 153 In doing so, the guidance should clarify the scope of the doctrine of 'special responsibility' in the context of digital platforms.

Complementary regulation: Elements of an Intervention Test
To the extent that the goal of policymakers is to ensure contestability between digital platform or ecosystem alternatives, the only realistic possible tool for intervention might need to take the form of ex ante regulation. This is especially the case where it is an instance of market failure that is under the policy spotlight, rather than an example of specific strategic anti-competitive behaviour. Several thresholds for intervention under ex post (extended) competition law or ex ante regulation have been proposed in the different policy reports and legislative initiative adopted recently. All of those proposals are based on a traditional economic policy framework. However, as Orla Lynskey observes, viewing the functioning of Internet gatekeepers only through economic lens may be too restrictive and might not sufficiently take into account concerns relevant to the preservation of individual rights. 154 Thus: -The Furman Report in the UK proposes the test of significant market status, defined as enduring market power enjoyed by a firm over strategic bottleneck market or a position to exercise market power over a gateway or bottleneck in a digital platform, where they control others' market access. 155 -A recent policy paper of the French telecommunications regulator, ARCEP, proposes a test of systemic digital platforms, defined on the basis of three main criteria: the existence of bottleneck power, a certain number of users in the EU (or as a proxy, sufficiently high EU turnover), the existing of integration of that firm into an ecosystem enabling leverage effects; and four secondary criteria, such as, gatekeeper position, access to many high quality data, market shares for online advertising (where relevant), and the market value of the platform. 156 153 Where the infringement of these regulatory principles is persistent and is capable of fuelling anti-competitive strategies, it would not be unreasonable for policymakers to equate such ex ante violations with ex post infringements of a dominant position under an Article 102 TFEU action, assuming that one can prove a nexus between the various sides of a platform that are being affected by the same practice. for a significance/systemic power analysis based on the following criteria: (i) the size -total value of its assets exceeds € 30 billion; (ii) the economic importance for the specific Member State or the EU economy as a whole; (iii) the size of the cross-border activities -the total value of its assets exceeds €5 billion and the ratio of its cross-border assets/liabilities in more than one other participating Member State to its total assets/liabilities is above 20%; or (iv) the direct public financial assistance when the bank has requested or received funding from the European Stability Mechanism or the European Financial Stability Facility. The banks meeting the "significance" threshold are regulated at the EU level by the Single Supervisory Mechanism while the other banks (namely, the "less significant" institutions) continue to be supervised by their national supervisory bodies: -The proposed 10 th Amendment of the German competition law introduces the concept of undertaking of paramount significance which would be determined on the basis of five criteria: a dominant position on one or more markets; financial strength or access to other resources; vertical integration and activities on otherwise related markets; access to data relevant for competition; importance of activities for third parties' access to supply and sales markets; and related influence on third parties' business activities. 157 It is also interesting to note that recent examples of EU legislation applicable to digital platforms differentiate their obligations according to the size of the platforms in terms of users, their annual turnover or their age (the number of years the platform has been active in the EU). 158 Our view is that the intervention threshold should be a natural extension of the policy orientation reflected in existing regulatory instruments. In addition, a regulatory threshold test should be onerous, as is the case in the electronic communications sector with the use of the Significant Market Power test where markets satisfy the "three criteria test" (as explained above in Section 2.3.1.) As such, regulatory intervention would be less susceptible to the criticism that it constitutes an unnecessary additional burden which cannot satisfy a cost-benefit analysis and which would be susceptible to the generation of Type 1 errors. In other words, its principles should be compatible with, or be a natural extension of, inter alia, the relevant principles contained in the EECC, the Open Internet Regulation, the P2B Regulation and the PSD2 Directive (as explained in Section 2.3.2). In this regard: (i) Any form of intervention should be based on existing economic or legal standards derived from ex ante or ex post practice or in the academic literature which relates to these disciplines.
(ii) To the extent that commercial practices systematically infringe the principles set forth in those regulatory instruments, these practices can legitimately constitute the analytical basis for an ex post action brought under Article 102 TFEU where such behaviour tends to promote or reinforce anti-competitive harm. To the extent that Competition Authorities can draw a causal nexus between such breaches of normative rules with anti-competitive harm, it should not be problematic for legislators and policymakers to equate such conduct with an antitrust theory of harm.
Any new regulation, however, should be proportionate and should not exceed what is necessary to achieve its objectives. In this regard, it is anticipated that ex ante regulation would only address the sorts of issues which competition policy cannot address effectively.

A three-criteria test for ex post competition law and complementary regulation
In relying on the above principles as the basis for any model of regulatory intervention, a digital platform-specific version of the "three criteria test" currently used in the electronic communications context might be relied upon to justify intervention which aims to ensure the existence of a level playing field which promotes market contestability. The criteria crafted for a digital platform environment would consist of the following:

Criterion 1: Non-contestable Concentrated Market Structure
Under this first threshold criterion, policymakers would need to identify the following elements in the marketplace in which the digital platform operates, namely: (i) The existence of a well understood and defined digital "platform" or Internet ecosystem of products which lies above regulated telecommunications networks, which generates direct and indirect Robert Schuman Centre for Advanced Studies Working Papers network effects that might be capable of deterring switching by the creation of "positive feedback loops", or which is vulnerable to the phenomenon of "tipping" in favour of a particular operator (or limited group of operators).
There is no specific definition of an online platform under EU law. It is in principle an "Information Society service", as defined by Directive 2015/1535, 159 which has, according to the Commission, the following characteristics: (i) the ability to create and shape new markets, to challenge traditional ones, and to organise new forms of participation or conducting business based on collecting, processing, and editing large amounts of data; (ii) they operate in multi-sided markets but with varying degrees of control over direct interactions between groups of users; (iii) they benefit from 'network effects' where, broadly speaking, the value of the service increases with the number of users; (iv) they often rely on information and communications technologies to reach their users, instantly and effortlessly; (v) they play a key role in digital value creation, notably by capturing significant value (including through data accumulation), thereby facilitating new business ventures and creating new strategic dependencies. 160 (ii) On the input side, such platforms benefit from high entry barriers which, in particular, may be due to the control of key innovation capabilities. As explained by the Commission, barriers to expansion or entry can take various forms. In particular, they may take the form of advantages specifically enjoyed by the dominant undertaking, such as economies of scale and scope, privileged access to essential inputs, important technologies or an established distribution and sales network. 161 (iii) On the customers' side, such a platform's market position benefits from the prevalence of singlehoming, which may be due to strategic behaviour (e.g., exclusivity clauses, tying, MFNs) or from ineffective multi-homing options due to consumer inertia.
In addition, when there are elements of "free" provisioning, which may increase information asymmetries and have the tendency to mask commercial incentives, the rationale for public intervention may increase. This can be the case when 'price' is a digital representation of value, in particular in the form of personal data that is being provided in exchange for the supply of a service. 162

Criterion 2: Digital Gatekeeper which is an Unavoidable Trading Partner
Under this second criterion, the digital platform has the following commercial and structural characteristics: (i) The digital platform acts as a "gatekeeper" or as a provider of some form of a "bottleneck facility", which renders it an "unavoidable trading partner" (comparable to a "must have" product in a conglomerate market environment) for other operators on that digital platform or on that digital ecosystem. This criterion requires that the platform in question has achieved a significant level of user uptake and that the access seekers depend on the platform to reach users. 163 (ii) The digital platform's business model is associated with the provision of some form of intermediation between two types of users. This implies that the platforms in question should allow a user to offer goods or services to another user with a view to facilitating transactions between them. 164 (iii) The environment in which the digital platform operates is characterised by the lack of effective countervailing bargaining power being exercised by the underlying regulated networks over which the ecosystem operates, nor by its immediate or end-user customers. As explained by the Commission in other contexts, such countervailing buying power may result from its customers' size or their commercial significance to the dominant undertaking, and their ability to switch quickly to competing suppliers, to promote new entry or to integrate vertically, at least where they can credibly threaten to do so. 165 (iv) Existing regulatory obligations are not effective in preventing the digital platform in engaging in exclusionary behaviour. To the extent that a digital platform is exploiting a regulatory loophole, or where regulation is not capable of exerting a meaningful market discipline on the platform, competition rules can rely on the normative rules prescribed under regulation upon which to base a theory of harm under competition law.

Criterion 3: Effectiveness of competition rules
On the assumption that either a finding of dominance cannot be established 166 in relation to a particular digital platform (despite positive findings usually associated with market power) or the identified infringements cannot be effectively addressed pursuant to appropriate remedies under ex post competition rules and within a reasonable period of time given the speed of technology and market evolution, there is a risk that the changes effected to the marketplace are irreversible and the possibility of innovation from other competitors is foregone.
In the electronic communications regulatory framework, competition law interventions are deemed to be insufficient where, for instance, the compliance requirements of an intervention to redress persistent market failure(s) are extensive or where frequent and/or timely intervention is indispensable. 167 At the stage of granting remedies, the need for interoperability obligations and contestability provisions may mean that certain remedies need to be imposed systematically across all relevant market actors, alongside asymmetric remedies which are designed to address specific market power issues.

Consequence of the test
The fulfilment of first two criteria Insofar as the first two criteria are satisfied, ex post intervention is likely to be appropriate, although certain procedural changes to Regulation 1/2003 168 might arguably also be necessary to address abuses of market power, in order to render such intervention effective and timely. competitive harm when the two first criteria are met. While no presumption of illegality should, in our view, flow from such a situation, the defendant would, in those circumstances, be expected to tender the type of evidence necessary either to dispel the view that anti-competitive consequences would follow or that consumer welfare benefits would outweigh any competitive detriments. This would be consistent with a number of regulatory norms that are already in place, which require transparency measures in order to determine why such problematic practices are being put into effect, and establishes the basis upon which a potential defendant would raise pro-competitive/pro-efficiency arguments consistent with the criteria set forth in Article 101(3) TFEU. Thus, while the ultimate burden of proof would remain with the European Commission, this is not to say that a defendant has no positive role to play in the justification of its actions. This is especially the case since the acts of disintermediation or selfpreferencing are, according to the regulatory instruments already discussed, potentially distortive of competition even in the absence of market power.
In this regard, systematic behaviour which transgresses clear regulatory norms (e.g., the nondiscrimination provisions of the Open Internet Regulation and the P2B Regulation, the 'free consent' requirements of the GDPR) or Codes of Conduct that may be prevalent in an industry can legitimately establish the basis of a theory of harm which supports a competition law infringement action. Business justifications put forward by defendants to justify their impugned practices should be subject to the test set forth in Article 101(3) TFEU, namely, they should constitute the minimum required to achieve their legitimate commercial goals (which is also consistent with the overarching policy requirement that a dominant undertaking has a duty of "special responsibility" to the marketplace under Article 102 TFEU). 169 This range of infringements should not be subsumed within the refusal to deal doctrine (which would embrace exclusivity relationships), 170 and it should continue to be governed by ex post competition rules that are subject to a high standard of judicial review, which is appropriate given the importance of avoiding a Type 2 error. 171 The fulfilment of the three criteria Insofar as all three criteria are satisfied, some form of targeted ex ante intervention may be appropriate to address market failures. To the extent that Regulation 1/2003 can be amended accordingly, appropriate action can be taken by the Commission under a revised Sectoral Inquiry power which allows greater freedom for the European Commission to conduct enquiries closer in form to that of the UK's "market investigation" regime, and to impose regulatory-style remedies possibly to be implemented by a regulatory authority as a by-product of such an investigation.
One alternative would be to draw inspiration from NRAs responsible for implementing access obligations and ensuring interoperability in the electronic communications sector. In this way, regulatory obligations can be tailored to the characteristics of the particular SMP-designated operator in a manner which is reminiscent of the procedure followed in EU competition law investigations. To this end, the delegation of the monitoring role to ensure that such behavioural remedies are respected could even be made to National Regulatory Authorities, who would cooperate with DG Competition or National Competition Authorities to ensure their effective enforcement.

Designing an EU Intervention Standard for Digital Platforms
European University Institute 39

Preliminary conclusions
Further to the adaptation of competition law enforcement in accordance with the commercial dynamics of digital markets, as discussed in the previous Section, the process of competition law enforcement needs to be accelerated and to be strengthened when digital markets satisfy the proposed three criteria test.
Thus, if the market structure is concentrated and non-contestable and the digital platform under investigation is a digital gatekeeper, the four commercial practices listed in the previous Section (namely, bundling and envelopment strategies, refusals to provide access or interoperability to key inputs and innovation capabilities, discrimination/self-preferencing, and violations of key regulatory principles) have a high probability of having an anti-competitive effect. Accordingly, Competition Authorities should prioritise investigations into such conduct, as the standard of proof for an antitrust infringement will often be met in such cases. As noted earlier, however, this does not absolve the Commission of the need to update its guidance on the enforcement of Article 102 TFEU.
If, in addition to those two criteria, ex post competition law intervention is not sufficiently effective in remedying the identified market failure, in particular because it is too slow or because the remedies to be imposed require constant monitoring, ex post competition rules probably need to be complemented by the introduction of targeted ex ante intervention. Such ex ante intervention may be crafted by the Competition Authorities with powers to conduct market investigations and to impose remedies. In the alternative, there is the possibility that they can be implemented by Regulatory Agencies (or even imposed by the Regulatory Agencies themselves) which will also monitor these remedies.

Choice of Remedies
In terms of the remedies that might be imposed appropriately in furtherance of the three criteria outlined above being satisfied, there should be no illusions to the effect that the crafting of remedies will be a straightforward exercise. The effectiveness of most of the remedies assume a minimum degree of transparency of market practices, as is now required by the P2B Regulation and other legislatives instruments. Given the intricate nature of commercial relationships in a digital platform environment, thought should nevertheless be given to, inter alia, a number of remedy options.

Range of Remedies
When considering appropriate remedies to address perceived market failures or market abuses in the field of digital platforms, one is confronted with two options that are usually associated with regulatory intervention, namely:  the breaking up of very large platform operators in an effort to disrupt the anti-competitive incentives associated with vertical integration;  the mandated sharing of essential innovation capabilities (such as data) among competitors in order to promote interoperability and to promote innovation; (which may also lead to the conferral of greater powers on individuals to port their own data in order to ensure the contestability of digital markets). Robert Schuman Centre for Advanced Studies Working Papers structural or functional separation should not be adopted because many of the benefits and efficiencies generated by digital platforms might be lost if their businesses were to be separated. Structural separation should only be imposed in very exceptional circumstances when the digital platform in question is very mature (in terms of the business model used and the acceptance of consumers of that model), demonstrates persistent indications of market failure, and behavioural remedies under ex post and ex ante disciplines have been demonstrated to be ineffective over a relevant period of time. Therefore, behavioural remedies imposed under competition law enforcement which can be effected in a timely manner or (when competition law is not sufficiently effective) under regulation should be preferred.
Having taken the view that a structural remedy is likely to be too onerous, and more prone to lead to a Type 1 error, we consider below the efficacy a range of other possible remedies.

Access to key capabilities and interoperability
The existence of market power in digital markets inevitably gives rise to requests for access by competitors, whether that access request relates to the fundamental building blocks of digital marketsessential or indispensable data -or to requests designed to ensure that a competitor's services can coexist, interact with or complement the services of, the digital platform/ecosystem provider (loosely referred to as 'interoperability', while not subject to the usual bilateral relationship associated with an interconnection relationship).

(a) Access to ensure interoperability
With regard to the issue of interoperability, any intervention standard would need to strike the appropriate balance between, on the one hand, interoperability to avoid consumer lock-in and, on the other, sufficient flexibility such that platforms could continue to compete on the basis of differentiation and innovation. Innovation might be more effectively delivered, for example, if competition occurs between networks or digital platforms, rather than between service providers operating over individual networks or platforms. However, when the three criteria explained above are satisfied, the benefits of compulsory interoperability should in principle exceed its costs, unless proved to be otherwise.
In principle, the development of interoperability standards should be left to industry participants, with some form of independent oversight, such that the agreed-upon standards can indeed deliver a meaningful level of interoperability. Moreover, the standards would need to be open, thereby allowing them to be freely adopted by all industry participants.
Licensing should, as a general principle, be prescribed on FRAND terms, unless the competitive harm is identified to be so significant, and the bottleneck control being exercised to be so extensive, that it justifies some form of LRAIC terms. The fact that digital platforms and ecosystems have developed in free market conditions suggests to the authors that the imposition of LRAIC terms may, however, be more likely to result in a Type 1 error. 173 Moreover, to the extent possible, regulators should demonstrate sensitivity in accommodating the existence of multiple standards if feasible, as this would arguably support greater innovation.

(b) Access to data capabilities
With regard to access to content or access to data, several horizontal rules (such as the General Data Protection Regulation and the Free Flow of Non-personal Data Regulation) 174 or sectoral rules (e.g., 173 By contrast, the LRAIC standard is appropriate in the telecommunications environment, where fixed incumbent operators had benefited from decades of monopoly and public investment: EECC, Annex III. 174  those adopted in the banking, transport and energy sectors) have been introduced recently and are only starting to take effect. Time may be required to determine the true value of such policies, and even the extent to which they might act as a deterrent in many cases to interoperability and data sharing. 175 Should the process of data sharing become more widespread, a separate set of competition issues are likely to arise in terms of high levels of concentration in the data collection process and as a result of the possibilities of downstream coordination between competitors or downstream foreclosure occurring where data is particularly granular and 'raw'. 176 In the era of 'big data', any comprehensive data pooling or data sharing regime will need to be assessed in terms of its overall compatibility with the principles set forth in the Commission Guidelines on horizontal co-operation agreements. 177 As regards data which amounts to a critical strategic asset (i.e., key innovation capabilities which are indispensable to a competitor's ability to compete or innovate) in the hands of an operator, to the extent that such data is critical in the exercise of the operator's 'gatekeeper' or 'bottleneck' functions, they could be assessed by reference to the same criteria outlined above in Section 4.
One can also foresee that the withholding of access to data relating to a business user's products and services for anti-competitive aims, or the leveraging of data from one market into another in order to dominate other product markets, will require appropriate remedies in appropriate circumstances. To this end, there may be circumstances where the sharing of data by a digital platform between distinct user groups or applications might be deemed to be anti-competitive. In such circumstances, the systematic failure to respect data protection rules might be considered to constitute an antitrust violation in its own right. 178 However, as was demonstrated by the Düsseldorf Court on appeal from the Facebook Decision of the German Cartel Office, one should anticipate that those connections need to be proven rather than merely asserted.
Ultimately, given the privacy concerns that will inevitably arise from the mandated sharing of consumer data (let alone the human rights implications of such a practice), coupled with the coordinated effects concerns that might arise from the sharing of data among competitors under an Article 101 analysis, and the difficulty of proving that data is indispensable under an Article 102 analysis, a pragmatic option by which to mandate data is to opt for a regulatory solution. It is inevitable that the focal point of any regulatory intervention will, in turn, be driven by the relative importance of data as a proprietary asset of consumers as well as its competitive significance as a potential barrier to entry or expansion. As a counterweight to this, however, is the fact that data protection rules themselves foresee the difficulties of sharing a consumer's data in a digital platform context. In addition, the ability to separate and isolate an individual's data from the other data which the digital platform has manipulated, consolidated and restructured will constitute a very complex (if not impossible) exercise in many circumstances. Robert Schuman Centre for Advanced Studies Working Papers

Prohibition of discrimination and self-preferencing
The approach explored by the Commission in the Google Shopping Case serves as an example of what could be achieved when implementing a non-discrimination remedy, insofar as the risk of discriminatory treatment needs to be answered by positive justifications from the operator with market power about the welfare-enhancing aspects of such behaviour. 179 In these very limited circumstances, it is arguably legitimate for the onus to shift on to that operator so that it can affirmatively prove that its actions are not discriminatory in their impact but, rather, based on objective and transparent criteria of differentiation. In this respect, such an approach goes very little further than the need under existing regulation for digital operators or those with intermediation power to justify instances of selfpreferencing, even in the absence of "market power" in the hands of the platform provider. 180 In the telecommunications sector, the concept of discrimination extends, by definition, to issues of selfpreferencing, at least where an operator with market power is vertically integrated. 181 The concept of discrimination needs to be assessed both by reference to quantitative (price) and qualitative (terms and conditions, display formats, etc.) criteria. When determining whether the qualitative aspects of a commercial policy has resulted in discrimination, competitors can have recourse to the standards anticipated by KPIs (Key Performance Indicators), which should be offered by the operator with market power. Insofar as the industry works through the use of Codes of Conduct, it will inevitably be more straightforward to have recourse to KPIs (along with an associated fast-track arbitration procedure designed to resolve disputes between the operator with market power and its competitors).
Finally, as occurs in the regulatory framework affecting telecommunications networks and services, the introduction of secondary remedies which promote greater transparency will inevitably render the implementation of a non-discrimination remedy more effective and timely.

Facilitating consumer switching and the prohibition of exclusivity
Where appropriate, action can be taken to prevent digital platforms or Internet ecosystems with market power from exerting pressure on business customers to deal with them exclusively or from generating inappropriate economic incentives which dissuade customers from switching between digital platforms or Internet ecosystems. Traditional EU competition law administrative practice appears to be sufficiently robust to be able to deal with such issues where market power is involved.

Methods by which to impose remedies
Given the novelty of many issues that may need to be remedied in the digital economy, combined with the important information asymmetries which prevail between the relevant Authorities and the digital firms, it may be critical in order to ensure the effectiveness of the public intervention to involve the 179 In its Decision 19-D-26 against Google, the French Competition Authority considered that the Google Ads rules imposed on advertisers should be applied under objective, transparent and non-discriminatory conditions. The opacity and the lack of objectivity of these rules made their application by advertisers very difficult. At the same time, Google had full discretion to modify its interpretation of the rules in a way that was difficult to predict, thereby allowing Google to apply its rules in a discriminatory and inconsistent manner. 180 Refer to discussion in Section 2.3.2 regarding the various regulatory instruments already in place. regulated firms in the design of regulatory remedies. To this end, the Nobel prize winner Jean Tirole is also calling for more participatory antitrust in digital industries. 182 This explains why the establishment of Codes of Conduct are recommended in several EU legal instruments relating to digital platforms. 183 These Codes of Conduct seek to arrive at industry consensus on the appropriateness of commercial behaviour insofar as they may tend to support non-discriminatory operational standards. Such Codes of Conduct might be adopted in response to an enquiry triggered under the procedure set forth above, or adopted voluntarily by market participants in advance of any public intervention.
In this regard, the Commission has developed, by a process of open consultation, certain principles for better self-and co-regulation that have been tested by a pilot Community Practice. 184 These principles relate to the formulation of the rules, insofar as they should: be prepared openly and by as many as possible relevant actors; and set clear targets and indicators, and be designed in compliance with EU and national laws. In turn, certain principles also relate to the implementation of the rules, insofar as they should be: monitored in a way that is sufficiently open and autonomous; improved in an iterative manner (learning by doing); and non-compliance should be subject to a graduated scale of sanctions.
Departures from such a Code of Conduct could, in turn, be treated in a similar fashion to a dominant firm's departure from the terms of its agreement in SEP context. In both cases, competition law would be capable of intervening because of the presumption that the balance between proprietary rights and competition policy needs to be interpreted in favour of the latter, given that welfare loss would be greatest if competition rules were not enforced fully (i.e., resulting in the risk of a Type 2 error). The participation of an operator with market power in a Code of Conduct context should, in these circumstances, be actionable if the operator departs materially from the terms of that Code of Conduct. In these circumstances, the Code of Conduct plays a comparable role to a Reference Offer in the telecommunications sector, insofar as it establishes minimum standards which can be improved upon but which cannot be degraded. 185 By introducing some form of Code of Conduct to regulate certain key strategic aspects of the relationship between the platform provider and a competitor on one side of the market in its capacity as a "customer" of the platform's services, a situation is created which is comparable to that addressed by the Commission under EU competition rules when dealing with Standard Essential Patents ("SEPs") that are licensed on FRAND terms. 186 Also, consistent with the remedies imposed in a SEP case, the relevant Authorities may impose the obligation on parties to enter into negotiations that should be effected in good faith and without undue delay. The relevant Competition or Regulatory Authorities with jurisdiction on such matters could impose such obligations in order to incentivise the parties to agree to arrive at a balanced solution. Finally, the possibility of commitments being adopted by regulated firms and being validated by Regulatory Authorities under a regime which is similar to that recently introduced under the EECC, might also extended to the regulation of digital platforms.

Implementation
The implementation of the proposed information regime for digital platforms will inevitably require some changes to the procedural bases for intervention and the institutional arrangements used to support such intervention. To the extent that a correct balance is struck between the types of obligations borne by platform providers and the strengthening of procedures to enforce those obligations, the extent of institutional change might be minimised.

Ex post Competition Law
The Commission could bind itself through guidance and adopt Decisions based on such guidance, which would render its decisions no less immune to effective judicial review by the European Courts even where it has recourse to a truncated approach to market definition. In particular, the scope of the doctrine of "special responsibility" across multiple sides of a market, the importance attached to a platform of ecosystem being an 'unavoidable trading partner' and the application of conglomerate effects theory are all important matters that require further clarification through a Communication, Notice or Guidelines.
To this end, the modernisation of several instruments (such as the Notice on Market Definition, 187 the Guidelines on Horizontal Agreements 188 and on Vertical Restraints, 189 the Communication on the Commission's Enforcement Priorities in Applying Article 102 TFEU, 190 and the Guidelines on Non-Horizontal Mergers 191 ) may be appropriate in order to take due account of enforcement realities in the digital platforms context. As previously discussed, these realities include, but are not limited to: the application of the doctrine of special responsibility across the different sides of a platform; greater recourse to conglomerate effects theory, both as the basis by which to limit the scope of the market definition exercise, and as the basis upon which to better understand the effects of anti-competitive leveraging practices; the adoption of a more dynamic approach to competition, including the development of tools to understand the impact on potential competition; an explanation of developing theories of harm and the rationale by which they are likely to cause anti-competitive effects; and the range of remedies best adapted to address such theories of harm. The adoption of such an approach should remove much of the delay inherent in the conduct of antitrust proceedings and should render Commission decision-making less susceptible to Type 1 and Type 2 errors.
Moreover, a procedural change to Regulation 1/2003 that would arguably be more defensible would consist of the clarification of the burden on the Commission to satisfy the current legal test for the grant of interim relief. In particular, it might be clarified that the risk of "irreparable damage" to competition currently required under Article 8 of Regulation 1/2003 is more easily satisfied in those cases that are susceptible to market tipping (as has already arguably been recognised by the European Courts in 187 Commission notice on the definition of the relevant market for the purposes of Community competition law, O.J. [1997] C 372/5. 188 Commission Guidelines on the applicability of Article 101 TFEU to horizontal co-operation agreements.. Microsoft). 192 Having said this, the recent European Commission proceedings brought against Broadcom provide us with the possibility of a practical examination of the robustness of the present EUlevel regime for injunctive relief. 193 To the extent that existing doctrines are appropriately interpreted and applied in a digital world, we do not believe that any fundamental changes are required to the standard of proof or to the burden of proof borne by the Commission under Article 102 TFEU, nor to the standard of judicial review to which the Commission is subject before the European Courts. However, to the extent that the various criteria set forth in the discussion in Section 4 have been met by a firm which is dominant, it appears that practices such as the refusal to grant access to essential inputs, self-preferencing practices 194 and departures from the terms of Codes of Conduct, 195 are capable of meeting the standard of proof for anticompetitive conduct. It is important that the Commission elaborate upon its rationale for drawing such conclusions in the form of updated guidance, at the very least by amending its Enforcement Priorities Guidelines.

Ex ante Competition Law or Complementary Regulation
Competition law need not only look backwards in resolving existing competition problems, but can look forward ex ante in the manner effected under merger control rules. This ability to apply ex ante thinking to antitrust analysis is taken further in jurisdictions such as the UK and Australia, where hybrid forms of antitrust/regulatory methods of investigation and remedy creation are endorsed.
One possibility would therefore be to modify Regulation 1/2003 to give powers to the European Commission to conduct full market investigations (rather than the more narrowly focused "Sectoral Inquiries") and apply a form of ex ante competition law in response thereto. 196 In doing so, the stigma of investigation and the of firms can be avoided, to be replaced by a more consensual form of remedy application.
In the alternative, if it is felt that specific legislative instruments need to be adopted, one can envisage the following alternatives, namely: 197 (i) an extension of the P2B Regulation with new asymmetric obligations (i.e., imposed on firms with market power) based on the three criteria test described above in Section 4, while at the same time broadening the scope of the Regulation to all intermediation platforms; (ii) a new piece of EU legislation which introduces the essential procedural, substantive and institutional changes foreseen above; or (iii) a new piece of comprehensive legislation, in the form of a European Digital Framework or a Code for Online Platforms, which would merge and amend all the existing economic and non-economic rules currently applicable to online platforms. In particular, this could include the e-commerce Directive, the Open Internet Regulation, the P2B Regulation, the AVMS Directive, elements of the DSM Copyright Directive and aspects of the NIS Directive. 192 Case T-201/04, Microsoft v. Commission, EU:T:2007:289, para 562: "In this case, the Commission had all the more reason to apply [Article 102 TFEU] before the elimination of competition on the work group server operating systems market had become a reality because that market is characterised by significant network effects and because the elimination of competition would therefore be difficult to reverse." 193 See IP/19/3410 of 26 June 2019. At the time of writing, it appears that Broadcom will file an appeal to the General Court seeking the annulment of the Commission's Decision. 194 As occurs in the regulated field of electronic communications. 195 As occurs in the context of SEP disputes. 196 Similar to the manner proposed in the Joint Memorandum of the Benelux Competition Authorities. 197 The Communication from the Commission of 19 February 2020, Shaping Europe's digital future, COM(2020) 67 notes (at page 10) that: 'The Commission will further explore, in the context of the Digital Services Act package, ex ante rules to ensure that markets characterised by large platforms with significant network effects acting as gate-keepers, remain fair and contestable for innovators, businesses, and new market entrants'. Robert Schuman Centre for Advanced Studies Working Papers In considering these various options, policymakers will be mindful of the fact that the formulation and passage of EU legislation is both a complex and a time-consuming exercise when one departs from examples of 'soft law' such as Recommendations, Notices or Guidance. Although the precise legal form of ex ante intervention to be adopted is a policy choice which should be left open to the EU legislator, arguably the easiest way forward may be to re-consider the subject-matter scope of the P2B Regulation and incorporate its logic into normative antitrust rules.

Institutional Design
Any new form of complementary regulation should preferably be implemented by an existing regulatory agency, so that the incremental change to its existing powers is reasonable and proportionate. Given the broad and horizontal scope of such regulation, such an Authority should have horizontal competence. Moreover, implementation should ideally occur at the EU level or should involve the EU's institutions, given the likely broad territorial scope of activities of the most significant online platforms. In the alternative, it should at the very least be sufficiently harmonised at EU level.

Extending the powers of DG Competition
Arguably the best institutional option consists in an amendment to Regulation 1/2003 to extend the powers of DG Competition to conduct market investigations similar to the power of the CMA. Such a legislative change would ensure that the Authority responsible for administering the new rules is found at European level and acts horizontally across sectors.
The possibility of such an option being effective at EU level has risen recently with the increase in the portfolio of the existing Competition Commissioner to include Digital Agenda matters. In such circumstances, irrespective of any residual concerns about regulatory 'capture' of the Competition Commissioner, the broadening of her portfolio to include matters of regulatory policy arguably places her in a unique position to administer both complementary or supplementary strands of ex post and ex ante policies affecting digital platforms.
Given that any ex ante legislation that is ultimately adopted will be drafted from a competition law perspective and will be designed to complement existing competition principles, an extension of the Competition Commissioner's powers in this way seems to be proportionate and most likely to yield positive results. 198 On the other hand, competition law 'purists' concerned that the free flow of regulatory principles into the discipline of antitrust might strain the enforcement ethos of a competition body 199 might take the view that the remit for a fused competition law/regulatory mandate should be relatively narrow (e.g., digital advertising) and administered by a specialist body or a discrete operations of that body.

Relying on regulatory agencies
An alternative approach would be to entrust a separate (or an existing) Regulatory Authority to apply any complementary ex ante regulation. This Authority should ideally have horizontal competence and EU territorial reach or, if these Authorities are national, they should be strongly coordinated within an EU network.
For such EU coordination to be effective, different integration models are possible: 198 It was recently announced by Commissioner Vestager that, in order to avoid any perceived conflicts of interest between the exercise of the ex post and ex ante functions, a screening mechanism would be introduced. See https://www.mlex.com/GlobalAntitrust/DetailView.aspx?cid=1148698&siteid=190&rdir=1. 199 Particularly as regards the pressures of ongoing surveillance which might arguably arise from the greater use of behavioural remedies.

Designing an EU Intervention Standard for Digital Platforms
European University Institute 47

(i) Example of Single Supervisory Mechanism
The most integrated model for regulation is the Single Supervisory Mechanism which was established in the aftermath of the financial crisis to supervise the most significant banks in Europe. 200 Currently, the European Central Bank ECB directly supervises more than 100 banks in the Eurozone which hold more than 80% of banking assets in the Eurozone area.
Each bank is supervised by a dedicated Joint Supervision Team (JST) composed of staff of the ECB and the National Financial Supervisory Authority, including the competent authorities of the countries in which credit institutions, banking subsidiaries or significant cross-border branches of a given banking group have been established. The size, overall composition and organisation of a Joint Supervision Team is tailored to the size, business model and risk profile of the bank it supervises. However, any decision is taken at the EU level by a Supervisory Board composed of a Chair appointed for a non-renewable term of five years and a Vice-Chair chosen from among the members of the ECB's Executive Board, four ECB representatives and the representatives of national supervisors.

(ii) Example of the internal market procedures for electronic communications
With respect to the regulation of electronic communications networks and services, regulatory decisions are adopted by the National Regulatory Authorities, although acting in close cooperation with BEREC (i.e., the pan-European body representing National Regulatory Authorities) 201 and the Commission. Indeed, while the Commission may veto the market definition and the SMP designation conducted by an NRA, it may only recommend that the NRA adopts different remedies than those originally proposed. 202 Remedies adopted by NRAs under this model can in effect only be challenged before national courts on the basis of a full review of the merits of the NRA Decision.
The recent announcement of cooperation by the Austrian Competition Authority and the Austrian NRA in the electronic communications sector is also an indicator of potential future cooperation in this regard in a bid to find common solutions in relation to digital markets. 203 It should also be remembered that the Italian authorities responsible for competition policy (AGCM), electronic communications (AGCOM) and data protection matters (DPA) also envisage working together in the future to address public policy issues raised in digital markets. 204 An even greater level of convergence can be identified in those Member States where multiple regulatory functions and competition law powers are merged into the same body (e.g., Spain, the Netherlands). 205

Conclusions
The weight of opinion around the world in a series of studies and reports on digital platforms suggests that momentum is building for the view that antitrust policy requires an overhaul, possibly both at the substantive level and at the procedural level. This view is driven by the belief that, as regards digital markets, the risk of making "Type 2" errors (i.e., under-enforcement) may be greater than in other its rationale for the various theories of harm which can be associated with the exercise of market power in digital markets, we take the view that a reversal of the burden of proof under competition law will be unnecessary, as the anti-competitive practices in the case precedents have a sound economic basis, which can usually also be supported by precedents drawn from normative rules found in the Merger Regulation, Article 102 TFEU and Article 106 TFEU. The precedents drawn from ex ante regulation may also be relevant. What is required, however, is the clarification of the Commission's rationale for antitrust intervention in the form of updated guidance.
(vi) Given the particular dynamics of digital markets, it may be the case that certain market distortions are due to instances of market failure, rather than being examples of abusive market power being exercised strategically by a dominant firm. In these situations, a restoration of the status quo ante, as is usual in the case of antitrust interventions, may be insufficient to address the underlying competition concerns. Accordingly, specifically targeted ex ante intervention may be necessary, predominantly on an asymmetric basis, with a view to ensuring that markets are rendered contestable. Such remedies would usually consist of a range of interoperability and data-related measures, whether to ensure that end users are able to port their personal data freely, competitors can gain access to a dominant firm's data that is judged to be indispensable, or where competitors can gain access to a dominant firm's key innovation capabilities in order to ensure interoperability (broadly understood). In turn, where the breach of normative regulatory obligations is systematic, these breaches might be able to serve as the basis for an antitrust theory of harm where the nexus between the regulatory breach and any adverse impact on competition can be identified.
(vii) We propose the introduction of a so-called "three criteria test", where the satisfaction of the first two criteria is consistent with the application of ex post competition rules to address anti-competitive actions by dominant firms. The satisfaction of the third criterion will justify the application of targeted ex ante measures designed to render the affected relevant market(s) contestable. The first of these criteria consists of an analysis of whether the market structure in question is concentrated and non-contestable. The second of these criteria determines whether the dominant firm acts as a digital gatekeeper or an unavoidable trading partner. The third criterion appraises whether the characteristics of the identified competition problem are such that competition rules are unable to address that problem in a timely and effective manner.
Although we are agnostic about which institutions should have responsibility over the exercise of both ex post and ex ante functions, we see much merit in the view that the European Commission's DG Competition is well placed to perform these dual functions. To this end, we believe that a limited regulatory mandate for DG Competition would benefit from its powers under Regulation 1/2003 being extended to perform the function of market investigations, comparable to the powers already held by Competition Authorities in Australia and the United Kingdom. However, in order to ensure that behavioural remedies are implemented effectively and monitored over time, it may be necessary to envisage procedural changes to Regulation 1/2003 that would allow DG Competition to liaise with other national and pan-European competition or regulatory bodies to ensure effective enforcement.