REMEDIES IN EU ANTITRUST LAW

Abstract

Remedies are central in contemporary EU antitrust enforcement. However, they remain relatively misunderstood. Against this background, this article has three main objectives. It seeks, first, to shed light on the nature and purpose of remedial action under Articles 101 and 102 TFEU, with a focus on the European Commission’s activity. The point of intervention, the case law shows, is to bring an infringement effectively to an end. As a matter of positive law, it is unclear that there is room for restorative remedies (that is, remedies that seek to recreate the conditions of competition as they would have existed in the absence of the practice). Second, the article takes a critical perspective on the practice of the past decade. It appears, in particular, that a ‘principles-based approach’ to the administration of remedies is likely to lead to suboptimal outcomes. Finally, some recommendations are outlined so that the letter of the law matches the demands and ambitions of the sort of regulatory-like intervention that is necessary in digital markets and other industries presenting similar features.

I. INTRODUCTION

The centre of gravity of EU antitrust law¹ is shifting towards the design and administration of remedies. Effective enforcement now depends on how the infringement is brought to an end as much as it does on detecting and proving it to the requisite legal standard. This phenomenon is a relatively recent one. During the formative years of the discipline, remedies were, generally speaking, an afterthought that rarely ever demanded the attention and energy of courts and authorities. Absent exceptional circumstances, crafting them was not a resource-intensive, let alone challenging, exercise. As a rule, intervention would take the form of a negative obligation, which was to be administered on a one-off basis. For instance, the remedy in the typical exclusive dealing case would entail little more than demanding the supplier to cease imposing single-branding obligations on its customers.² Remedial measures in a run-of-the-mill cartel investigation are not any more sophisticated. They involve asking the participants to cease the conduct—and, if anything, refrain from engaging in it in the future.³

The conception and implementation of remedies have become considerably more complex in recent years. In the contemporary landscape, a cease-and-desist obligation will often be insufficient, in and of itself, to bring infringements to an effective end. In many instances, it will be necessary for agencies to define, in a positive manner, how the undertaking subject to an obligation is to behave (inter alia by dealing with third parties on regulated terms and conditions,⁴ respecting a certain price level,⁵ redesigning its product⁶ or altering its business model⁷). These remedies are not easy to craft. They require courts and competition authorities to become, in effect, regulatory agencies, even though they do not necessarily (or not always) have the powers, expertise, and resources to reconfigure markets—let alone monitor compliance over time. It is not by chance, for instance, that the controversy around the implementation of the remedy in cases like Microsoft⁸ and Google Shopping⁹ lingered long after the European Commission (hereinafter, the ‘Commission’) adopted its decision. As a result of this reality, tensions might emerge between the demands of effective enforcement and the institutional limits that are inherent to the system.

The reasons behind the growing complexity of remedies are well understood. The trend is explained, above all, by the prioritisation by authorities (including the Commission) of both network and digital industries.¹⁰ These two sectors are characterised by the fact that (at least) some activities (bottlenecks) do not allow for competition, in the sense that they tend to be dominated by an overwhelmingly large player. One can think, in this sense, of search engines, app stores,¹¹ and online marketplaces.¹² Leveraging conduct by a bottleneck operator cannot always be tackled by means of one-off negative obligations. Sometimes, intervention demands the administration of regulatory-like measures like the ones described above. Where, for instance, an undertaking controls an input that determines competitive outcomes on neighbouring segments, simply requiring that the infringement be brought to an end does not address the underlying concerns in any meaningful way. Specifying the terms and conditions under which the firm is to deal with rivals in adjacent markets becomes a necessity in such circumstances.

Whereas complex, regulatory-like remedies, have become a central part of the enforcement landscape, the legal order has adjusted to the new reality only partially and imperfectly. As is often the case, the law lags behind the transformation of policy-making, thereby giving rise to uncertainties and frictions. The latter have only been solved on a piecemeal, trial-and-error basis. Against this background, one of the ambitions of this paper is to show that the framework built around Regulation 1/2003¹³ was not conceived to deal with the administration of regulatory-like remedies on a frequent basis. It is explained, in this sense, why the lengthy and uncertain aftermath of Microsoft and Google Shopping is not an exceptional occurrence, but a predictable consequence of the enforcement regime. One can reasonably expect, accordingly, that the institutional flaws that these cases exposed are likely to lead to similar outcomes in the future.

The relative neglection of remedies in EU antitrust law and policy is also reflected in the commentary. For a long time, the shift in the centre of gravity of enforcement was not given the attention it deserved. With some exceptions,¹⁴ only recently have remedies featured in the literature.¹⁵ As a result, key questions—pertaining to their role in the system and their effectiveness—have been explored, if at all, relatively superficially. For instance, the question of whether Articles 101 and 102 TFEU allow for the administration of so-called restorative remedies (which would go beyond merely bringing an infringement to an end and involve a duty to inject or enhance competition within the markets affected by the anticompetitive practice) has been discussed¹⁶ but remains unanswered. This paper examines these matters systematically. In particular, it purports to explore the limits of what remedial action allows, as the law stands. In addition, it explores the ways in which the institutional regime could be adjusted so that enforcement remains effective.

The points that this paper makes are presented as follows. Section 2 discusses the nature and purpose of remedies in EU antitrust enforcement, with a specific focus on the Commission’s activity and its powers to adopt decisions under Regulation 1/2003.¹⁷ It emphasises, in this sense, that the role of these measures is to ensure that unlawful conduct is brought to an end (and is not repeated in the future). One should bear in mind, in this regard, that Articles 101 and 102 TFEU are drafted as prohibitions against certain practices the object or effect of which is the restriction of competition. There is, accordingly, no obvious basis to see them as instruments for market restructuring. In the same vein, the effectiveness of intervention must be assessed by reference to their ability to address the concerns identified in the decision. Thus, whether or not rivals thrive following intervention, or whether the relevant market becomes more competitive after the decision, are immaterial considerations as the law stands.

Section 3, in turn, discusses the various categorises of remedies from two angles. In the first place, this section proposes a taxonomy of measures that goes beyond the basic divide between the structural and the behavioural. It is important to distinguish, in this sense, between regulatory-like obligations and traditional measures, if only because of how much the former and the latter differ from one another in theory and practice. In the second place, Section 3 deals with the approaches that competition authorities have taken with regard to the design and definition of remedies. These approaches are considered from two perspectives. One perspective explores the differences between a ‘principles-based approach’, which leaves the design of the measures to the undertakings, and one that specifies in detail the obligations with which they must comply. A second perspective distinguishes between the obligations that are be imposed in the context of a prohibition decision (which are remedies in the strict sense of the word) and commitments in lieu of such remedies.

Section 4 critically assesses the reality of remedies under Regulation 1/2003. It focuses on two salient aspects. The first one is the rise in prominence of ‘principles-based’ intervention in the Commission’s practice. It is submitted, in this regard, that there is nothing in the case law indicating or implying that competition authorities are under a duty to follow such an approach to remedial intervention, or, similarly, that they must refrain from specifying the way in which undertakings must bring the infringement to an end. The second point relates to the current legal landscape, and hints at normative changes. An overview of the most salient cases shows that the regime, in its current incarnation, could be improved, not just because it does not provide for an explicit framework for the evaluation and the testing of remedies, but because it does not allow the Commission to formally and positively state that the measures put in place by undertakings bring the infringement effectively to an end.

II. THE NATURE AND PURPOSE OF REMEDIES IN EU ANTITRUST LAW

A. Bringing the Infringement Effectively to an End

The concept of remedy, which does not necessarily (or not always) have clear boundaries in other legal disciplines,¹⁸ has a very specific scope and meaning in EU antitrust law. It refers to a measure that is imposed following the formal declaration of an infringement by means of a decision and which is aimed at bringing it effectively to an end.¹⁹ An infringement, in turn, is nothing other than a practice that runs counter to one or both of the prohibitions laid down in Articles 101 and 102 TFEU. Remedies are a necessity whenever unlawful conduct continues to be implemented at the time of the adoption of a decision. Just to mention an example, a dominant firm may be required to cease its ongoing discriminatory practices vis-à-vis its rivals.²⁰ In addition, and as part of remedial action, undertakings may be ordered not to engage in it in the future.²¹

Competition authorities, including the Commission, typically focus on substance, rather than form, when designing and implementing remedies. In order words, an infringement decision, generally speaking, does not simply order undertakings to cease the specific behaviour concerned by the investigation. Typically, firms are also asked to refrain from engaging in conduct having a similar object or effect, even if superficially different.²² One should bear in mind, in this regard, that unlawful behaviour can be formally manifested in more ways than one. Participants in a cartel may achieve their aims by coordinating their pricing strategy or, alternatively, by allocating production quotas. Accordingly, bringing the infringement effectively to an end will mean not just ordering the undertakings to refrain from participating in one manifestation of the unlawful behaviour (say, price-fixing) but to avoid other forms of cartelisation (by means, for instance, of shared quotas). A focus on substance, rather than form, ensures that intervention is meaningful and effective.

The case law provides additional clarifications about the range of measures that can be required to bring an infringement effectively to an end. In the first place, the Court held, early on, that remedial action may involve not just a negative obligation (that is, an obligation not to engage, or to refrain from engaging, in certain conduct), but also a positive one. For instance, a dominant undertaking may be subject to a positive duty to resume supplies to a customer if its decision to stop doing amounts to an abuse of a dominant position.²³ Similarly, a group of undertakings that conclude an agreement within the meaning of Article 101(1) TFEU may be required to license their technologies to third parties.²⁴ In the second place, remedies may be both behavioural and structural in nature. Just to mention an early example from the case law, there may be circumstances where the infringement can only be effectively brought to an end by mandating a divestiture.²⁵

There are, on the other hand, limits on authorities’ ability to impose remedies in EU antitrust cases. Article 7 of Regulation 1/2003²⁶—as much as the general principles of EU law²⁷—demand that the chosen measure keeps proportion with the infringement. Accordingly, any (positive or negative) obligations cannot go beyond what is necessary to bring the breach to an effective end. In AKZO, for instance, the Court found that the measures were disproportionate insofar as they required the dominant firm to offer equivalent prices to firms that were not in a comparable position.²⁸ An implication of the principle of proportionality is that, whenever there are several (equally effective) options available to an authority, the said authority must choose the one that is the least onerous for the addressee of the decision.²⁹ As will be discussed in greater detail below, an agency can require the undertaking to come up with proposals relating to how it intends to comply with its duties.³⁰

Article 7 of Regulation 1/2003 introduces an additional limit on remedial action. This second constraint significantly circumscribes the Commission’s ability to impose structural obligations. This limit is best conceptualised as a presumption whereby such structural measures go beyond what is necessary to bring an infringement to an end. Altering the market structure and forcing a firm to sell some of its assets can only be imposed, according to the said provision, in two circumstances, namely ‘where there is no equally effective behavioural remedy’ and ‘where any equally effective behavioural remedy would be more burdensome for the undertaking concerned than the structural remedy’.³¹ The Preamble to the Regulation adds, in the same vein, that structural remedies are only appropriate where, as a result of the ‘very structure’ of the undertaking, ‘there is a substantial risk of a lasting or repeated infringement’.³²

B. Beyond Ceasing the Infringement: The Scope for Restorative Remedies

The immediate conclusion that follows from the preceding sub-section is that the scope and reach of remedial intervention in EU antitrust law is shaped and defined by the infringement that it is designed to address. The remedy, in other words, mirrors the relevant behaviour and cannot go beyond what is necessary and proportionate to bring it to an end.³³ As observed by authors like Ritter, there appears to be no basis to require an undertaking to recreate the conditions of competition that would otherwise have existed.³⁴ Similarly, remedial intervention cannot demand that the market share of the undertaking be decreased back to the level it had before the infringement. The more modest role of remedial intervention is to give effect to the prohibitions enshrined in Articles 101 and 102 TFEU. One should also note that the point of remedial action is not to sanction the behaviour of the undertaking or to deter firms from engaging in it.³⁵ Such a function is fulfilled by fines.³⁶

In spite of the limits described above, discussions about the feasibility and desirability of restorative remedies have become relatively frequent in the literature.³⁷ Since the scope and meaning of the notion is not always defined with clarity, it makes sense to mention, at the outset, that restorative intervention comes in two flavours: a narrow and a broad one. According to the narrow understanding, these are remedies that bring the market back to the status quo ante (that is, the conditions of competition that existed before the infringement). An example in this sense is Baltic Rail.³⁸ The Commission found that the incumbent operator in Lithuania had abused its dominant position by removing part of the railway infrastructure in a key corridor. The authority identified several remedies, one of which sought, in effect, to ‘restor[e] the competitive situation that existed’ before the behaviour—that is, the reconstruction of the removed rail track.³⁹ The firm challenged the Commission decision at first instance, but not on appeal.⁴⁰

Under the broad understanding of the notion, restorative intervention would not simply bring the market back to the status quo ante, in the sense that it would, in addition, alter its operation to reverse the structural consequences of the infringement. Pursuant to this interpretation, remedial action could provide for prospective measures giving rivals the means to compete more effectively. An example in this sense can be found in the Special Advisers’ Report on digital competition.⁴¹ The authors observe that discriminatory conduct such as self-preferencing may significantly advantage the subsidiary of a dominant firm in a neighbouring market. As a result, it may be necessary—they argue—to adopt restorative measures to ‘enable formerly disadvantaged competitors to regain strength’.⁴² Such measures would comprise obligations relating to the data that the dominant firm exploits.

Whether in its broad or narrow versions, it is unclear that there is a legal basis for restorative remedies in EU antitrust law (in other words, that the Commission and, indeed, national courts and authorities have the power to impose such measures). This point of law has been occasionally explored by commentators.⁴³ Positive arguments in support of restorative intervention tend to rely on a single passage in Ufex.⁴⁴ The Court held, in that case, that, where the ‘anti-competitive effects continue after the practices which caused them have ceased’, the Commission ‘remains competent’ to ‘to act with a view to eliminating or neutralising’ such effects.⁴⁵ According to a view favourable to restorative intervention, ‘eliminating or neutralising’ the effects of anticompetitive conduct comprises remedial action recreating the conditions of competition that would have existed had the practice not been implemented.

This single passage, however, does not provide unequivocal support for such an interpretation. It can be construed in more ways than one. The reading that is consistent with restorative intervention is not, in fact, the most plausible or straightforward one—let alone the most faithful to the body of case law as a whole. The best way to make sense of the ambiguous passage in Ufex is to read it in light of the ruling—Continental Can⁴⁶—to which the Court refers in support of its position. The latter concerned the acquisition a rival in breach of Article 102 TFEU.⁴⁷ Bringing this particular infringement effectively to an end (or, if one prefers, ‘eliminating or neutralising’ its effects on competition) necessitates a divestiture. To the extent that it does, the remedy in Continental Can did nothing other than adjust to the specificities of the abusive strategy. Just like it is sometimes necessary to impose a positive duty to deal on an undertaking, bringing the infringement to an end may sometimes require a court or competition authority to order a divestiture. This fact, however, does not mean that intervention changes to become restorative in nature. What changes is merely the technique by which the infringement ceases.

AKZO, which used virtually identical vocabulary, provides support for this restrictive interpretation of the ambiguous passage in Ufex (and, by extension, of the role and limits of remedial intervention in EU antitrust law). The Court held, in AKZO, that the obligation imposed by the Commission in that case—a ban on selective price-cutting—was ‘intended to prevent the repetition of the infringement and to eliminate its consequences’.⁴⁸ It ensured, in other words, that the dominant undertaking’s conduct would cease and that it would not be repeated in the future. As noted by Ritter, at no point did the Court suggest that ‘eliminating [the] consequences’ of the abusive conduct demanded or involved restorative intervention, whether in the narrow or the broad sense.⁴⁹ Nothing in the judgment, in fact, suggests that remedial action can involve bringing the market back to the status quo ante. If anything, the opposite is true. The latter point is developed in detail below.

One can think of a second argument which, from a positive standpoint, casts doubt about the scope for restorative intervention in EU antitrust law. The letter of Articles 101 and 102 TFEU prohibits undertakings from engaging in certain practices insofar as they amount to a restriction of competition and/or an abuse of a dominant position. One could reasonably argue, against this background, that restorative remedies are at odds with the boundaries of intervention that derive from the scope of Articles 101 and 102 TFEU. Remedial action aimed at recreating the conditions of competition that would otherwise have existed assumes that undertakings have a duty not just to refrain from engaging in certain conduct, but an obligation to preserve effective competition. It is not immediately obvious that Articles 101 and 102 TFEU can be interpreted as supporting such a far-reaching reading.

A careful look at Regulation 1/2003 provides support for a relatively restrictive understanding of the role of remedial intervention. Pursuant to Article 7(1), the Commission may ‘find that an infringement has been committed in the past’, provided that there is a ‘legitimate interest in doing so’.⁵⁰ The point of retrospective intervention in EU antitrust law, in other words, is not to restore the conditions of competition that existed before the infringement (or to restructure the market prospectively), but to allow for intervention if there is a legitimate interest in doing so. This may be the case, first, where the Commission intends to signal that an infringement is serious. Even if a fine is not imposed,⁵¹ second, a retrospective decision may give the Commission the chance to clarify the scope and meaning of Articles 101 and/or 102 TFEU.⁵² Finally, as in Continental Can, the fact that the practice has been committed does not mean that intervention is not required with a view to ‘eliminating or neutralising’ its effects on competition.⁵³ For instance, a refusal to deal may have been committed in the past, but addressing its consequences (that is, the fact that would-be rivals do not have access to the relevant input or infrastructure) may be deemed legitimate.

Even if one were to take a normative perspective, it is unclear that the administration of restorative remedies is feasible or desirable. As far as the feasibility of restorative intervention is concerned, one cannot ignore the fact that it would introduce an additional layer of complexity to the design and administration of remedies. It would no longer be enough to ensure that the infringement is brought to an effective end. In addition, courts and authorities would need to the engage in what amounts, in effect, to market restructuring. They would be required to refashion sectors and/or revive competitive dynamics so that they conform to the conditions that would have prevailed in the absence of the practice (or to the desired outcomes, if a prospective view of intervention is taken). It is unclear whether (and, if so, how often) courts and authorities—in particular the former—would be able to engage in intervention along these lines. One must bear in mind, in this sense, that even the most modest form of regulatory-like action has already been shown to test the institutional limits of the EU antitrust system. Restorative remedies may stretch it beyond what it can realistically achieve on a systematic basis.

One could argue, in response to the concerns, that these remedies would only apply in carefully selected instances, namely where the deterioration of the conditions of competition is particularly acute, or where harm is likely to result from the implementation of a narrow set of well-defined categories of conduct. It has been suggested, for instance, that restorative intervention might be required in digital markets, the features of which often favour concentration and/or exclusion.⁵⁴ One could try and make the point, in the same vein, that only administrative authorities, which can be assumed to have the necessary capacity and expertise, would venture in this form of remedial action. Such a solution would relieve national courts from the need (or, indeed, the obligation) to engage with the complexities that it necessarily entails.

The idea that restorative remedies can be imposed selectively (or, similarly, that their administration can be left to the discretion of the enforcement body) raises a number of questions. The first and overarching one is whether selective or discretionary enforcement is in line with the general principle of equal treatment.⁵⁵ One should bear in mind that the decisions in both Langnese-Iglo and Schöller were annulled insofar as the remedies imposed could not be reconciled with this EU law principle.⁵⁶ The tension with equal treatment would be particularly pronounced if the decision to impose restorative remedies were left to the discretion of the decision-making body. If such an approach were followed, like infringements would not necessarily (or not always) be treated alike. The same would be true if the extent and depth of remedial action varied depending on whether Articles 101 and/or 102 TFEU is enforced by an administrative authority or by a national court instead. If the nature and reach of intervention fluctuated based on whether enforcement is public or private, restorative measures would not just be at odds with the principle of equal treatment. They would also negatively affect the uniform application of EU law.

It is possible to argue, in response to these arguments, that restorative intervention could be made to revolve around some objective criteria, instead of discretion. These objective criteria could include factors such as the nature of the sector or the features of the relevant market (such as the existence of network effects). If this route were to be followed, it would be necessary to determine in advance a number of issues, none of which is straightforward. A first question would relate to the definition of the very factors that justify the administration of restorative remedies. A second, to the legal basis for the introduction of these criteria. It is particularly unclear, in this sense, whether it would be possible for a competition authority (including the Commission) to provide for them informally, by means of a soft law instrument, as opposed to formal legislation.

An alternative argument in support of restorative intervention is that it could lead to more effective enforcement (and, indeed, the only form of effective enforcement in certain scenarios). It has been argued, in this vein, that competition can only be meaningfully protected if authorities are in a position to adopt, where necessary, measures that go beyond ordering that infringements be brought to an end.⁵⁷ This is particularly true, according to this view, in industries that have a natural tendency towards concentration.⁵⁸ Once markets ‘tip’ in favour of one player due to the operation of network effects, for instance, bringing the infringement to an end may fail to deliver any meaningful change. Therefore, additional intervention recreating a competitive landscape would be all but indispensable. If restorative intervention were not allowed in instances where competition has been damaged beyond repair (or risks doing so), in fact, enforcement would be confined to sanctioning prohibited conduct, as opposed to ensuring that markets are effectively competitive.

While compelling, this argument in support of restorative remedies is, at its heart, one that pleads in favour of altering market structures to inject competition. Returning to the status quo before the infringement (that is, the narrow and more modest version of restorative intervention) necessarily involves—at the very least—reallocating customers and market shares. It is, for that very reason, a position which inevitably ventures beyond what Articles 101 and 102 TFEU were designed to do, both from a legal and an institutional standpoint. As already suggested above, these two provisions do not prohibit market power as such, but the strengthening thereof by means of anticompetitive conduct. Market restructuring for restorative purposes requires either the adoption a formal regulatory regime or, alternatively, the introduction of a new competition tool along the lines of the models have already been put in place in some EU jurisdictions in recent years.⁵⁹ The argument in favour of restorative intervention, in other words, is above all a plea in favour of new powers complementing what can be realistically achieved, on a systematic basis, under Articles 101 and 102 TFEU.

C. Measuring the Effectiveness of Remedies

A remedy worthy of the name must be effective. This objective will be achieved, first, where the measures are an appropriate means to bring the infringement to an end. In some instances, the breach can be adequately addressed by means of a self-executing obligation. For instance, a vertically integrated firm can be ordered not to squeeze its downstream rivals by leaving an insufficient margin between the wholesale and the retail prices.⁶⁰ In other instances, an adequate remedy will need to be spelled out in detail—and may sometimes require the implementation of a monitoring mechanism. This is so, just to mention an example, where a dominant firm is ordered to deal with rivals.⁶¹ A remedy will be effective, second, where it ensures that the undertaking subject to it does not circumvent the decision by engaging in other conduct that has an equivalent object or effect.

It has occasionally been suggested that the effectiveness of intervention depends (or should depend) not just on whether the remedy is an adequate response to the infringement, but on whether it improves the conditions of competition on the relevant market(s) affected by the practice. According to this view, a remedy mandating an undertaking to cease tying two products, for instance, would only be effective where rivals on the market for the tied product are able to maintain or increase their market share vis-à-vis the dominant firm (and, in the same vein, where the market share of the latter decreases over time). Similarly, remedial action requiring an undertaking not to impose exclusivity obligations on its customers would be deemed effective, pursuant to this interpretation, where new entrants are able to improve their position at the expense of the former’s.

This sui generis understanding of what amounts to an effective remedy has occasionally been raised by agency officials, albeit in an informal manner. In the aftermath of the Commission decision in Microsoft, for instance, then Commissioner Neelie Kroes expressed the view that ‘a significant drop in market share is what we would like to see’ following a finding of infringement.⁶² A virtually identical view has been advanced in relation to the remedies implemented to bring Google’s self-preferencing practices at stake in Google Shopping. In an extensive report, Thomas Höppner argued that ‘[o]ne would expect that once the abusive favouring has been removed and [rivals] are free to compete on the merits, the logical consequence would be that market share of Google’s [service] falls while that of [rivals] rises’.⁶³

There is little support in the case law for the idea that remedial action is only effective where it leads to an improvement of the conditions of competition. To some extent, this discussion echoes the points made above in relation to restorative measures. The judgments that deal with effectiveness directly and at greater length suggest that the point of any obligation is to ensure that intervention creates the conditions within which effective competition can be preserved and sustained over time, not to change market structures. Put differently, the point of remedial action is to ensure that rivals have the opportunity to thrive on the merits, not to engineer market outcomes. In AKZO, as already pointed out, the dominant firm was required to refrain from engaging in selective price cuts targeted at its rival’s customers (ECS).⁶⁴ The Court explained that the rationale behind the remedy was to allow ECS to endeavour to win back the customers lost during the abusive campaign, not to re-allocate them back to it.⁶⁵

The same conclusion follows if one pays attention to what is implicit in other rulings. Consider, for instance, the case law on refusals to deal. Nothing in the analysis and aftermath of the key cases suggests that the effectiveness of a duty to deal must be measured against the ability of new entrants to thrive and expand their activities once access to an input or infrastructure is provided. An obligation to deal is effective, instead, where it allows rivals to compete on the merits. The non-compliance dispute that followed the Commission decision in Microsoft, for instance, did not focus on whether rivals had been able to increase their market share on the relevant downstream market, but on whether the dominant undertaking had given access to interoperability information on fair, reasonable and non-discriminatory (hereinafter, ‘FRAND’) terms and conditions, as required in the infringement decision.⁶⁶

More generally, evaluating the effectiveness of remedies by reference to changes in the market structure is at odds with the very logic of the EU antitrust system. An idea that transpires clearly and consistently from the case law is that Articles 101 and 102 TFEU protect the competitive process. These provisions are not designed to craft or attain a specific vision of how competition should be organised in a given sector or market. It is not for courts, authorities and, indeed, undertakings⁶⁷ to allocate shares to individual participants based on their particular preferences.⁶⁸ In has long been established that competition on the merits, by definition, may lead to the exclusion of undertakings that are less attractive in terms of, inter alia, price, quality or innovation.⁶⁹ Whether or not the market becomes more competitive following remedial intervention, therefore, depends on the operation of the competitive process, not on the efforts of a court or authority to artificially preserve rivalry.

One should bear in mind, in this vein, that the medium to long-term evolution of markets typically hinges on factors other than the remedy itself. The fact that effective competition is not achieved or preserved following intervention is not necessarily attributable to the failure of remedial action. It may be attributable to other factors, including the fact that some firms perform better than others once the infringement has been brought to an end. Making the effectiveness of enforcement depend on the evolution of market shares over time may therefore lead to penalising efforts to compete on the merits. It would also create perverse incentives that are by definition at odds with the rationale and logic of EU antitrust law. It would signal that, irrespective of their performance, firms benefitting from intervention can expect their position to improve.

III. CATEGORISING REMEDIES IN EU ANTITRUST LAW

A. A Taxonomy of Remedies

Article 7 of Regulation 1/2003 distinguishes between behavioural and structural remedies. However familiar as a divide, it is insufficiently specific, in the sense that it fails to capture the many forms intervention can take depending on the circumstances and needs of each case. The point of this section is to categorise remedies in a more granular way. In line with the discussion above, the analysis that follows focuses on substance, not form. The taxonomy is not based on what an authority formally orders, but on what intervention demands in effect. Suppose that the Commission finds that an undertaking, by refusing to deal with its downstream rivals, has abused its dominant position. Focusing on substance, in such a case, means placing an emphasis on what remedial action actually requires from the firm (that is, ordering the firm to deal with third parties), irrespective of the manner in which the remedy is presented to it (as discussed below, authorities sometimes follow a ‘principles-based approach’ to the design and administration of remedies and leave the specifics to the firm).

Remedial action in EU antitrust law, understood in this substantive sense, can be categorised in accordance with two main criteria. One can distinguish, first, between measures that are administered on a one-off basis and those which, by contrast, require continuous implementation (and, by extension, monitoring). This divide is captured in the horizontal axis depicted in Figure 1. It makes sense to elaborate on the two categories and discuss what they encompass in theory and practice. The typical remedy is a one-off: the authority imposes an obligation with which undertakings can comply by changing their behaviour once. Consider the example of predatory pricing. An infringement decision would simply entail ordering the dominant undertaking to alter its pricing so that it does not fall below cost.⁷⁰ Once the prices are modified as mandated, the authority does not need to monitor compliance (if at all, it may need to ensure that the undertaking refrains from repeating the infringement). As far as traditional forms of tying are concerned, bringing the infringement to an end involves nothing other than requiring the undertaking to cease conditioning, whether contractually or otherwise, the sale of the one product to the acquisition of the other. This outcome can be achieved on a one-off basis.⁷¹

The examples discussed so far concern instances where intervention is negative and behavioural in nature. It is not difficult to think of other examples of one-off duties that are, respectively, positive and structural. For instance, the participants in an airline alliance may be required to cede some airport slots (that is, a positive duty of a behavioural nature implemented as a one-off) to third parties.⁷² Undertakings cooperating in the context of a standard development agreement, similarly, may be required to introduce a procedure for the adoption of the standard is transparent and that allows any willing stakeholder to take part in the process.⁷³ Concerning structural measures administered on a one-off basis, in turn, Continental Can (where the dominant firm was subject to a divestiture obligation) is an apt example. Similar measures may also be introduced to address the anticompetitive effects of a network sharing agreement among telecommunications operators.⁷⁴

Other remedies, whether behavioural or structural, cannot be administered on a one-off basis. Their implementation takes places on a continuous basis. The sort of remedies imposed in refusal to deal cases under Article 102 TFEU come to mind immediately as examples of behavioural duties requiring implementation for as long as they remain in place.⁷⁵ The effective administration of these obligations makes it indispensable to define the terms and conditions under which access is to be provided⁷⁶ and design mechanisms to monitor and ensure compliance. Behavioural obligations that require continuous implementation are not confined to access duties. For instance, a firm may be required to change the design of a product. In Google Shopping, the firm was ordered, in effect, to alter the operation of its search engine.⁷⁷ In many respects, such form of intervention mimics the sort of obligations applied by regulatory agencies in the telecommunications and energy industries.⁷⁸ Other behavioural remedies that may require continuous implementation include reconfiguration of the terms and conditions under which firms supply or purchase goods and services,⁷⁹ as well as the alteration of the operation of the relevant market.⁸⁰

Structural obligations are not always capable of being implemented on a one-off basis. In some instances, the remedy will need to be complemented with behavioural duties, in which case compliance will require continuous administration and monitoring. Such an outcome could be a side-effect of structural intervention in instances where an undertaking is horizontally or vertically integrated and controls a bottleneck (that is, a segment of the value chain that is a monopoly or a quasi-monopoly). In such circumstances, a breakup separating the undertaking’s various activities may demand additional measures providing for the re-regulation of the bottleneck segment moving forward. After a finding of a leveraging abuse, for instance, an authority may order a vertically integrated firm to sell the monopolistic segment to a third party.⁸¹ Following the divestiture, the activities of the firm running the bottleneck may need to be monitored to prevent other forms of anticompetitive conduct further down the line (such as the exploitation of market power).

Figure 1 presents systematically the taxonomy of remedies along the two axes described above. One can identify, on the bottom left corner, traditional antitrust remedies, which are characterised by the fact that they are both behavioural in nature and administered on a one-off basis. Enforcement tends to be instinctively associated with this manifestation of antitrust intervention, which is generally adequate and effective in oligopolistic markets.⁸² As will be discussed at length below, Regulation 1/2003 was crafted with this form of remedial intervention in mind. The upper left corner, in turn, seeks to capture merger-like remedies. These mimic the preferred way to address the concerns raised by concentrations under Regulation 139/2004—that is, a one-off structural measure that immediately addresses the competition concern and does not require subsequent monitoring.⁸³

The right hand-side of the figure captures the sort of regulatory-like remedies that may be necessary in markets revolving around a bottleneck. These are the measures with which Regulation 1/2003 is not particularly well equipped to deal. The consequences of the mismatch between the legal framework and the needs of intervention are addressed in the following section. One can distinguish, just as one does in regulation, between behavioural and structural intervention. The former refers to conduct-based approaches that seek to prevent the leveraging of market power by means of, inter alia, the functional separation of the bottleneck and/or other changes to the operation of markets (including the reconfiguration of products and business models). The latter, in turn, can be likened to measures such as ownership unbundling in the electricity and gas industries⁸⁴ (whereby transmission activities are separated from competitive segments and subsequently re-regulated) and structural separation in telecommunications.⁸⁵

B. Approaches to the Definition of Remedies

The case law and Regulation 1/2003 allow competition authorities—including the Commission—to follow two distinct approaches to the definition of remedies in EU antitrust law. Under one of the approaches (which may be called the default method), the authority is responsible for the design and specification of the remedy in the infringement decision. It is virtually automatic for a competition authority to follow the default method whenever intervention takes the form of a traditional (one-off and negative) obligation. A duty not to require exclusivity from customers, as much as one to refrain from conditioning the sale of one product to the acquisition of another one, are a by-product of a finding of infringement and do not demand elaboration. Where the obligation is a complex one (typically, when it is positive in nature and requires implementation on a continuous basis), by contrast, the definition of the obligations is no longer a straightforward exercise. In Commercial Solvents, for instance, the Commission had to specify the amounts of the raw material that the dominant undertaking had to supply to its competitor in the immediate aftermath of the decision.⁸⁶

Whenever the remedy is a complex one, there could very well be alternative routes for compliance. If the core of a business model is found to be anticompetitive, for instance, there may be a range of options from which the undertaking can choose to bring its activity in line with the authority’s demands. Similarly, where leveraging conduct is found to be anticompetitive, it is in principle possible for the firm to comply with the decision in more ways than one: under one approach, it could decide to divest the bottleneck from the rest of its activities; alternatively, it may decide to abide by a strict non-discrimination obligation when dealing with third parties competing with it. Where there are several routes to compliance, the authority may decide to specify in its decision the different options that the undertaking has to bring the infringement effectively to an end. In Baltic Rail, for example, the Commission presented the various avenues by which the abusive conduct could be ceased.⁸⁷

As an alternative to the default method, the competition authority may decide to follow a ‘principles-based approach’ to remedial intervention. It is more likely that the latter approach will feature where the necessary remedies are of a complex, regulatory-like nature. Instead of specifying the obligations in its decision (by setting out, for instance, the terms and conditions under which an undertaking is to deal with third parties) an authority may choose to require the addressee of the decision to alter its behaviour so that it conforms with an overarching principle, against which compliance will be evaluated. In Google Shopping, just to mention a salient example, the Commission ordered the dominant undertaking to abide by a principle of equal treatment when displaying search results.⁸⁸ The precise specification of the remedy, described above, was left to Google. The authority followed the same approach in another high-profile case, Android.⁸⁹

The legality of the ‘principles-based approach’ to remedial intervention rests on two legal doctrines. In the first place, it is clear from the case law that the Commission is entitled to require the addressee to present proposals about the way in which it intends to comply with the obligations imposed in the decision. In Commercial Solvents, the ECJ found that mandating a firm to advance suggestions about how it sought to bring the infringement to an end flows from the remedial powers with which the authority is vested under secondary law.⁹⁰ Accordingly, the authority was found to be entitled to ask the dominant firm to specify the ways in which, moving forward, it intended to comply with the decision.⁹¹ The invitation to submit proposals will typically include a deadline to meet the obligation.⁹²

The second legal doctrine that provides support for a ‘principles-based approach’ to intervention is the line of case law pursuant to which an undertaking cannot be compelled to comply with the decision in a particular manner where there are other, less onerous routes to achieve the same outcome.⁹³ Consider again the factual scenario at stake in Baltic Rail to illustrate this point. If there are two ways to bring the infringement to an end (say, rebuilding an infrastructure and, alternatively, improving a different route), the Commission cannot demand that the firm adopt its preferred solution. Against this background, the authority can claim (and has claimed in some major cases⁹⁴) that it is not its duty to specify how the remedy is to be brought to an end, and thus that it is enough for it to require that it be ceased.⁹⁵ It could also argue, in a similar vein, that a ‘principles-based approach’ ensures that intervention is consistent with the proportionality principle.

C. Remedies and Commitments in Lieu of Remedies

1) The Administration of Remedies Stricto Sensu

It has already been mentioned that a remedy, strictly speaking, is a measure imposed following a formal finding of infringement within the meaning of Article 7 of Regulation 1/2003. When the Commission assorts a prohibition decision with (positive or negative) obligations, it is empowered not only just set a deadline for compliance⁹⁶ but also to require the addressee to report on the concrete measures it has put in place to bring the infringement to an effective end and, more importantly, to introduce mechanisms to ensure that the terms of the decision are respected.⁹⁷ Such mechanisms may vary from one case to another. In some instances, the addressee may be required to introduce transparency measures allowing third parties to benefit from the remedies.⁹⁸

The most visible of these mechanisms, in any event, is the appointment of an independent trustee in charge of overseeing the implementation of the remedy.⁹⁹ The first-instance ruling in Microsoft defined some principles around this figure. The Commission was found to have erred in law in its decision when designing this aspect remedy. This is so for two main reasons, which now define the boundaries of the role. The first error related to the conferral to the trustee of investigative duties that the authority was not empowered to delegate to a third party.¹⁰⁰ Insofar as it did, the Commission was found to have exceeded the limits of its jurisdiction to impose remedies. The second error had to do with the imposition on the firm of a duty to bear the costs associated with ‘the appointment of the trustee’—including those concerning ‘remuneration and the expenditure’ that comes with the exercise of their monitoring functions.¹⁰¹

There are three features of the institutional framework that are worth noting for the purposes of the discussion that follows. Pursuant to Article 24 of Regulation 1/2003, first, the Commission has the power to impose periodic penalty payments on undertakings so as to compel them to bring the infringement to an effective end. The periodic penalty payment is imposed by means of a formal decision. As the aftermath of the Microsoft case shows, the Commission may accompany the fine for non-compliance with a deadline to meet the terms of the original decision.¹⁰² If not respected, these penalties may be subsequently increased.¹⁰³ The second point to note about the compliance framework is the lack of third-party involvement. There are no adequate means for the potential beneficiaries of the remedy to provide meaningful input on the adequacy of the measure. Following the infringement decision, moreover, there is no formal mechanism allowing them to make their views known about implementation (as defined by the authority or the undertaking).

Third, and finally, the Commission does not have the power to adopt a decision positively declaring that the measures implemented by an undertaking bring the anticompetitive behaviour to an effective end. As a result of this gap in the institutional framework, there is no formal route to respond to disputes around compliance and, similarly, no way to address the uncertainty that might arise following the adoption of an infringement decision. Under the current framework, the Commission can only signal its positive assessment of compliance by means of unofficial, de facto routes, including by means of speeches¹⁰⁴ and policy briefs.¹⁰⁵ By way of consequence, the addressee may never be fully certain that it is complying with the terms of the decision.¹⁰⁶ Third parties potentially benefitting from the remedies, in turn, may be kept in the dark in relation to the matter (a reality that is compounded by the fact that they are not involved at the stage of the design and implementation of the measures).

2) Commitments in Lieu of Remedies

When the Commission has concerns about the compatibility of a practice with Articles 101 and/or 102 TFEU and intends to adopt an infringement decision, it may choose to close the investigation by accepting the commitments offered by the undertakings concerned. These commitments can be made formally binding upon them by means of a decision adopted in accordance with Article 9 of Regulation 1/2003.¹⁰⁷ This sub-section is not the space to discuss at length this instrument, which has been the subject of abundant commentary in the literature.¹⁰⁸ Its more modest purpose is to highlight the substantive and procedural differences with infringement decisions when it comes to the imposition of obligations on undertakings. Unlike remedies within the meaning of Article 7, commitments are made binding in accordance with a structured framework that allows third parties to express their views and that provides, by design, clarity and certainty to stakeholders.

As far as the first point is concerned, the Commission is required to conduct—pursuant to Article 27(4) of Regulation 1/2003—what is known as a ‘market test’. The underlying aim is to ensure that interested third parties are given the chance, over a specified period, to submit their observations about the measures before the adoption of the decision.¹⁰⁹ Commitment decisions, second, provide clarity and certainty for two related reasons. One of the reasons has to do with the fact that the Commission formally declares that, following the behavioural or structural changes proposed by the parties, there are no longer grounds for action. Even though it is not a statement of the law, but an expression of the authority’s administrative discretion,¹¹⁰ the Court has ruled that commitment decisions cannot be simply ‘overlooked’ by national courts, which must consider them, at the very least, as prima facie evidence of the legal status of the behaviour at issue.¹¹¹

A second reason that explains why commitment decisions provide clarity and certainty is that they preclude, by design, a ‘principles-based approach’ to the definition of the measures addressing the potential infringement. Accordingly, the undertakings subject to the investigation will have to spell out at length how they propose to address the Commission’s concerns. What is more, these measures will have to be evaluated in detail and published either as an annex to the decision or alongside it. For instance, Amazon Marketplace and Amazon Buy Box laid down a detailed regulatory regime setting out how the undertaking would change the operation of its business and ensure compliance.¹¹² Similarly (and to mention an example of a case involving a divestiture), the decision in German Electricity Balancing Market included an annex explaining how E.On would proceed to the sale of its transmission system business to a third party.¹¹³

IV. REMEDIAL INTERVENTION IN PRACTICE: POSITIVE AND NORMATIVE PERSPECTIVES

A. The Case against ‘Principles-Based’ Remedies

The Commission has followed a ‘principles-based approach’ to remedial action in some of the cases involving the most complex, regulatory-like interventions. Microsoft and the Google saga (Google Shopping and Android) are examples that have been mentioned above to make this point. It is submitted, in light of the experience acquired in these cases, that ‘principles-based’ intervention, as an enforcement technique, is likely to be suboptimal both from a legal and a policy-making perspective. The antitrust system would be more effective and would provide greater legal certainty if this approach were abandoned in favour of one that consistently specified in detail and with clarity (that is, in the same way commitment decisions do) the obligations with which undertakings have to comply. It is submitted, in the same vein, that ‘principles-based’ intervention lacks the necessary transparency at the crucial stages of the design and implementation of the remedy, thereby preventing third parties from becoming involved in the decision-making process in any meaningful way.

The first, overarching point to make in support of the case against a ‘principles-based approach’ is that the Commission is not required by law to adopt it. It is true that, as an authority, it cannot impose a particular obligation on an undertaking if there are other, less onerous measures to bring the infringement to an end. It does not follow from this doctrine, contrary to what is sometimes suggested or implied,¹¹⁴ that the Commission must refrain, always and everywhere, from specifying the obligations by which the undertaking must abide. The doctrine merely means that the authority cannot impose its preferences on the undertaking and that it is always open to the latter to propose alternative techniques to comply with its duties. One can conclude, accordingly, that the case law gives significant leeway to the Commission, and that any obligations it imposes are presumptively adequate and proportionate. The undertaking would bear the burden of showing that its preferred approach brings the infringement to an effective end.

One may retort that, even though the Commission is under no legal obligation to follow a ‘principles-based approach’ to remedial action, it is useful¹¹⁵ for it to do so and, to the extent that it is sometimes even desirable. The approach would be particularly useful, it may be claimed, whenever bringing the infringement to an end demands technical expertise and complex assessments, as was true of Microsoft (in which case it was necessary to determine the terms and conditions for the licensing of interoperability information) and Google Shopping (where the dominant undertaking was asked, in essence, to redesign its search engine). Since the relevant technical information and know-how lies with the undertaking, the argument goes, it would make sense for the authority to delegate the specification of the remedy. This perspective is consistent with a more general principle, often invoked in EU antitrust law, whereby the party with the relevant information bears the burden of producing it.¹¹⁶

While superficially sensible, this argument comes across as ultimately unpersuasive. One of the reasons why delegating the design and implementation of the remedy to the undertaking is problematic has to do with the fact that it may obscure the complexity of intervention. In other words, this approach may convey the false impression that bringing the infringement to an effective end is more straightforward than it actually is. This is particularly true where intervention demands the adoption of regulatory-like measures. Positive action requiring continuous implementation, in particular in innovative and rapidly moving industries, can be expected to be time and resource-consuming, and may not necessarily deliver the desired outcomes (at least, it is likely to involve some degree of trial and error). It is submitted that it would be beneficial for the credibility of the system (as well as for the management of the expectations of stakeholders) if the Commission (and, indeed, any other competition authority) were open about the complexities associated with the design and implementation of particular remedies, instead of concealing them.

One can think of a second reason why the delegation of the remedies to the undertaking may not work as intended. The firm that has the relevant information does not necessarily have an incentive to fully abide by the obligations deriving from the decision. Accordingly, it may seek to exploit the asymmetry of information to delay or avoid compliance. Where there is uncertainty about the exact meaning and scope of the obligations, in the same vein, the undertaking may be naturally inclined to favour the interpretation that most favours its own interests and that is less likely to address the concerns underpinning the infringement decision. The lengthy aftermath of the Microsoft decision, which revolved around the appropriate understanding of what amounts to a FRAND price in the circumstances of the case, illustrates the idea well.

The Microsoft saga also provides an eloquent illustration of a third reason why the delegation of remedies to the undertaking may not be desirable. It shows, arguably better than any other case, that a ‘principles-based approach’ does not necessarily allow competition authorities to circumvent the sort of complex, technical assessments that it is designed to avoid. ‘Principles-based’ intervention may, at best, give the authority the chance to delay engaging with such complex assessments, but it is unlikely to allow it to bypass them altogether. The non-compliance decisions issued by the Commission in Microsoft ultimately had to engage with regulatory-like issues such as the appropriate level of remuneration for the interoperability information¹¹⁷ and the level of innovation involved in the development of the said information.¹¹⁸ The experience acquired in this saga suggests that, more than delegating the technical expertise to the undertaking, a ‘principles-based approach’, if it is to be as effective as the default one, may only save enforcement resources on a temporary basis.

The discussion in this sub-section has so far focused on the fact that the ‘principles-based approach’ is not mandated by law and that it is, moreover, likely to affect compliance (in the sense that it can be expected to delay effective enforcement and conceal the scale and complexity of intervention). There are other reasons why this approach to enforcement may be problematic and may underdeliver in practice. One must note, to begin with, that the procedure for the design and implementation of the remedy is opaque. As pointed out above, the undertaking must inform the Commission about its plans to bring the infringement to an end (and, if necessary, to monitor compliance). However, the choices in terms of design, and the negotiations with the authority about any necessary adjustments to the remedy take place in the absence of a formal framework and without any information about the process followed to adopt (and, if necessary, adjust or amend) the remedy.

The aftermath of the Google Shopping case captures well the opacity and informality that accompanies the ‘principles-based approach’ to remedy design and compliance. It has already been mentioned that, among the options that it had to abide by the principle of equality of opportunity, the undertaking chose a solution that would give third parties the chance to bid, on a non-discriminatory basis, for premium space within the search engine (that is, space within the so-called ‘shopping box’). The implementation of its obligation required further detail and specification on a number of fronts. Among other questions, the firm had to define how the functional separation of Google Shopping from the rest of its activities was to be guaranteed in practice; how the auction process would be organised and how the results would be displayed on the search engine (for instance, whether the ‘shopping unit’ would display products or, instead, links to price comparison websites).

The public information about how the dominant undertaking went about the design and implementation of the remedy is, at best, informal and piecemeal. Google announced via its own media outlets how it intended to give effect to the Commission decision.¹¹⁹ As is apparent from a reading of this information—and was pointed out by some advisers involved in the case¹²⁰—the design and implementation of the remedy went through a number of iterations, and some of the changes were introduced following negotiations with the Commission.¹²¹ Vice-President Vestager publicly suggested, during that time, that modifications to the original design were necessary to adequately address all concerns.¹²² However (and this is the crucial factor to consider), the discussions between the firm and the authority, and the rationale behind the changes introduced, were never subject to formal public scrutiny, and did not take place in the context of a structured procedure. The aftermath of the Commission decision in Android (where the remedy went, again, through several iterations) further illustrates this idea.¹²³

Concerning, finally, the evaluation of complianceunder a ‘principles-based approach’, it makes sense to mention that it is not any less informal. The current regulatory framework is incapable of providing sufficient legal certainty, either to the addressee or to third parties. It has already been pointed out, in this sense, that the authority does not have the power to formally declare, by means of a positive decision, that the design and implementation of the remedy by the undertaking bring the infringement to an end. Accordingly, and unless the Commission adopts a non-compliance decision, a ‘principles-based approach’ will never provide full clarity about whether the firm has effectively met the obligations to which it is subject. This conclusion is inescapable if one considers the aftermath of Google Shopping. Several years after the finding of infringement, third-party rivals, and their advisers, continued to claim that the dominant undertaking had failed to abide by the principle of equal treatment and that additional modifications to the search engine’s operation were necessary.¹²⁴

The state of limbo in which undertakings (both complainants and defendants) find themselves is a source of legal uncertainty for another related reason. Precisely because there is no definite point at which the Commission formally declares that all of its concerns have been fully and adequately addressed, the question of whether the undertaking complies (or has complied) with its obligations has the potential to become a question subject to recurrent controversy and forever open to discussion and (re-)negotiation. For the same reason, nothing (other than the general principles of EU law, that is) prevents the authority from demanding changes to the original remedy package, or from moving, over time, the benchmark against which compliance is assessed. The legal uncertainty that could result from the absence of a formal, positive declaration of compliance can be exacerbated by other factors.

In particular, lack of clarity about the role and limits of remedial action is likely to fuel uncertainty in this sense and leave addressees and third parties in a state of permanent limbo. As explained above, it is occasionally suggested that measures are only effective where the market structure changes to become more competitive. In fact, this interpretation of remedies appears to have featured, albeit informally, in the aftermath of some infringement decisions that relied on a ‘principles-based approach’.¹²⁵ If the effectiveness of intervention were indeed measured against the evolution of markets in the direction that is deemed necessary or desirable (if, for instance, effectiveness were assessed in light of whether the firm’s share falls below 50 percent), the remedy would have to be constantly calibrated and recalibrated, thereby turning compliance into a perpetually live issue with no clear end in sight.

B. Adjusting the Legal Framework to Contemporary Remedies

The conclusion one can draw from the preceding sub-section is that Regulation 1/2003 was never designed—and is inadequately equipped—to deal with regulatory-like remedies. As mentioned in the introduction, this legal and institutional reality is not surprising, given that such forms of intervention were a rarity under Regulation 17 (that is, before the liberalisation of network industries and the rise of digital markets).¹²⁶ It is submitted, against this background, that there are powerful reasons to bring the framework for the administration of remedies in line with the realities to which it applies and with the demands of the industries that European competition authorities have chosen to prioritise over the past two decades. Rethinking Regulation 1/2003 along the suggested lines appears to be all the more appropriate at a time when the enforcement regime is being reviewed.¹²⁷

The review of the framework for the administration of remedies would ideally draw inspiration from two main sources. The first relates to the procedural mechanism enshrined in Article 9 of Regulation 1/2003 for the adoption of commitment decisions, and which has been discussed above. The second source of inspiration is the Digital Markets Act (hereinafter, the ‘DMA’),¹²⁸ which overlaps, ratione materiae, with some of antitrust cases investigated or contemplated by European competition authorities. This sector-specific regime is, at least in part, a response to the institutional limits encountered by the Commission in such cases.¹²⁹ Unlike Regulation 1/2003, the DMA acknowledges that regulatory-like measures (such as those administered in Google Shopping and Android) are not ‘self-executing’ and typically require further specification from the authority.¹³⁰ In this vein, it provides for several routes to provide legal certainty and ensure compliance by the ‘gatekeepers’ subject to the regime.

A combined reading of the relevant provisions—in particular, Article 9 of Regulation 1/2003 and Article 8 DMA—suggests a way forward for a reformed, more structured approach to remedial intervention. As far as the substance of intervention is concerned, the new framework would require that the (positive and/or negative) obligations imposed on the undertaking be spelled out with clarity and at length in the decision itself. It would be equally important to ensure and monitor compliance with the terms of the decision. In this sense, the DMA codifies a number of mechanisms that have featured in the Commission’s administrative practice and that could provide inspiration for a revised enforcement framework. For instance, Article 11 DMA addresses the reporting obligations to which ‘gatekeepers’ are subject. Article 28, in turns, deals with the ‘compliance function’ that ‘gatekeepers’ must set up following designation. This obligation is intended to ensure that the officers in charge of monitoring compliance have the necessary resources, expertise and independence to perform their function.¹³¹

From a procedural standpoint, this new framework would ideally lay down a mechanism for the specification of remedies that guarantees the involvement of the addressee and of third parties. In accordance with the letter and spirit of Article 7 of Regulation 1/2003, it would be for the Commission to define how the infringement is to be brought to an effective end. The specification of the remedy would follow two stages. First, it would be for the Commission to outline an initial proposal, to be ‘market tested’ with third parties (in the same way commitments are pursuant to Article 27(4) of Regulation 1/2003).¹³² The second stage, following the ‘market test’, would involve the addressee of the decision. In line with the case law, it would be for the undertaking to accept the authority’s (revised) proposal or to come up with a less restrictive alternative. The Commission would then evaluate the counterproposal and decide on its adequacy and effectiveness (the authority’s assessment and conclusions in this sense would be in the body of the decision and could therefore be challenged pursuant to Article 263 TFEU). These procedural steps are summarised in Figure 2.

V. CONCLUSIONS

The Commission Antitrust Manual of Procedures devotes a mere two pages—out of a total of 258—to remedies.¹³³ The relative neglection of the question in the ‘law in the books’ reflects the extent to which the formal legal landscape has proved unable to capture the contemporary reality, where the design, implementation and monitoring of these measures are at least as important as detecting and establishing the breach in the first place. The growing gap between the ‘law in the books’ and the demands of ‘law in action’ has exposed the many unresolved issues around the nature and limits of remedial intervention in the EU antitrust system. An overview of the growing literature shows that questions such as whether there is scope for restorative measures or how the effectiveness of remedies is assessed have not been fully resolved.

Against this background, this paper has shown that the point of remedial intervention in the EU legal order is to bring the infringement to an end. One should bear in mind, in this regard, that Articles 101 and 102 TFEU are drafted as prohibitions, not as positive mandates to preserve or promote competitive market structures. In fact, there is precious little support in the case law—to the extent that there is at all—for the idea that undertakings can be required to restore the levels of competition as they would have existed in the absence of the infringement. In the same vein, the effectiveness of remedies cannot be measured by reference to whether the measures deliver changes in the relevant markets. Again, the EU antitrust system was conceived to protect the competitive process and ensure that firms can rival one another on the merits, not to engineer market structures to conform to an authority’s particular vision or preference.

This understanding of the role and purpose of remedies in the EU legal order may legitimately come across as unambitious. It may even be challenged, as some commentators regularly do, on grounds that it would be ineffective in certain industries, and particularly so in digital markets. Once competition is affected beyond repair, it is argued, bringing the infringement to an end may fail to achieve any meaningful change. It is submitted that these arguments, while reasonable, do not account for the substantive and institutional limits of Articles 101 and 102 TFEU. These two provisions do not prohibit market power as such, only the exercise thereof. It is therefore difficult to see how remedial action can be used as a mechanism to circumvent these substantive limits and inject competition within an industry. From an institutional standpoint, moreover, any objection to the current boundaries of the EU antitrust regime fails to acknowledge the implications, in terms of resources, that the design, implementation and monitoring of restorative remedies would have.

It seems inevitable to conclude, in light of this analysis, that restorative intervention (and, by extension, intervention aimed at modifying market structures in the name of effective competition) is best achieved by means of ad hoc regulatory measures, such as the DMA, as opposed to competition law enforcement. Rather than stretching the substantive and institutional limits of the EU antitrust system to attain outcomes that are typically beyond its reach (whether from a positive or a normative standpoint), it is submitted that efforts should be directed at ensuring that Articles 101 and 102 TFEU can fulfil their aims in the most effective manner. This paper advances two ideas to improve the operation of the system. The first relates to authorities’ administrative practice (and more precisely to the approach followed by the Commission when crafting remedies); the second, to the procedural framework.

As far as the first aspect is concerned, there are reasons to question whether the ‘principles-based approach’ to remedial intervention is appropriate, let alone the optimal one. This is so, in particular, in the sort of cases—such as Google Shopping and Android—that require complex, regulatory-like intervention. In theory, delegating the design and implementation of remedies to the undertaking could present some advantages. After all, firms are in a better position to determine how best the infringement can be brought to an end. It appears, upon closer inspection, that a ‘principles-based approach’ is not only a source of legal uncertainty but is built around informal and opaque procedures. As such, it is likely to delay, rather than ease, the effective implementation of remedies. In addition, the aftermath of sagas like Microsoft suggests that the approach is likely to fail to deliver on its main promise. Eventually, the authority may not be able to avoid engaging with the sort of complex technical considerations which the ‘principles-based’ approach is supposed to spare.

As far as the procedural framework is concerned, the gap between Regulation 1/2003 and the practical needs of contemporary enforcement suggests that there are compelling reasons to redesign the legal regime that applies to the administration of remedies in the EU antitrust system. The review of the framework comes across as particularly appropriate at a time when the Commission has undertaken a review of the implementation of Articles 101 and 102 TFEU. From a substantive perspective, it seems necessary to ensure that any remedies (including any accompanying measures to guarantee compliance) be spelled out, with clarity and in detail, in the infringement decision. From a procedural standpoint, it would be appropriate to implement mechanisms allowing third parties to express their views about the adequacy of the measures envisaged by the Commission, on the one hand; and giving the addressee of the decision the chance to show that there are alternative, less onerous measures to bring the infringement to an end, on the other.

Footnotes

References

Author notes