How to Fix a Failing Art. 102 TFEU: Substantive Interpretation, Evidentiary Requirements, and the Commission’s Future Guidelines on Exclusionary Abuses (2024)

Key Points

• The duration of adversarial Art. 102 proceedings before the Commission has significantly increased in recent years, such that an effective deterrence and protection competition is no longer ensured.

• The case law on Art. 102 TFEU is partially unclear and incoherent; future guidelines should offer a conceptually coherent framework and address the dysfunctionalities of public enforcement.

• In support of the endeavour, we propose an interpretation of Art. 102 TFEU that we consider to be in line with the case law and with a ‘workable effects-based approach’.

• A speedier enforcement presupposes a clarification of the law of evidence, in particular with a view to proving potential anticompetitive effects; in this vein, we identify settings in which potential anticompetitive effects may be inferred, or in which a ‘quick look’ may suffice.

I. The ‘Article 102 package’—aims of future guidelines on exclusionary abuses

On 27 March 2023, the Commission has published its ‘Article 102 package’: in an ‘Amending Communication’,1 it has modified its so-called ‘2008 Priorities Guidance’2 and officially distanced itself from some of the core principles that previously characterised the ‘more economic approach’ to the enforcement of Art. 102 TFEU.

No less importantly, the Commission has announced its intention to adopt guidelines on exclusionary abuses in the course of 2025. A ‘Call for Evidence’3 has initiated the consultation which shall lead up to the publication of a first draft in mid-2024. According to the ‘Call for Evidence’, future guidelines on exclusionary conduct shall consolidate the Union Courts’ case law on Art. 102 TFEU, including the 32 judgments on exclusionary conduct that have been passed since 2008; provide companies with greater legal certainty; and enable a consistent enforcement by the Commission, national competition authorities, and national Courts.

In fact, the project is even more ambitious. A Policy Brief4 that DG Competition has published alongside the ‘Article 102 package’ reveals the broader agenda: the initiative strives to take a new look at exclusionary abuses in the light of 15 years of experience gained in the enforcement of Art. 102 TFEU based on a ‘Priorities Guidance’ that has proven to be—at least partially—dysfunctional. In its future guidelines, the Commission aims to develop ‘A dynamic and workable effects-based approach to abuse of dominance’, i.e. an interpretation of Art. 102 TFEU that allows for a speedier enforcement with a realistic expenditure of resources. The outstanding importance of an effective enforcement ‘in economically difficult times and in view of the increasing market concentration in various sectors’ is also emphasised in the ‘Amending Communication’.

Indeed, rethinking Art. 102 TFEU in the light of the principle of effectiveness is urgently needed. In part II of this paper, we show how the duration of Art. 102 proceedings before the Commission has significantly increased over time, such that Art. 102 enforcement ultimately risks to become ineffective in protecting undistorted competition. In part III, we briefly summarise important cornerstones of the substantive interpretation of Art. 102 TFEU. Where the case law seems unclear or incoherent, we propose a reading that we consider to be in line with a ‘workable effects-based approach’. In part IV, we focus on the law of evidence: if the current handling of the effects analysis is (partially) responsible for the prolongation of Art. 102 proceedings, settings should be explored in which potential anticompetitive effects may be inferred. Part V briefly concludes.

II. Dysfunctionalities in enforcing Article 102 TFEU

In its Policy Brief, DG Comp rightly highlights the need for a handling of Art. 102 TFEU that allows for an effective, manageable enforcement. This suggests a brief analysis of whether, and if so how and why, an effective enforcement is currently failing. In the EU, Art. 102 TFEU has been enforced persistently over the years. Nonetheless, the development of the duration of Art. 102 proceedings raises concerns. Furthermore, public enforcement of Art. 102 TFEU at the EU level is currently at a low, but it is too early to tell whether this is an actual trend or just a short-term anomaly (A.). Regarding the duration of Art. 102 proceedings, various factors have contributed to this development, including, importantly, particularities and complexities of the enforcement procedure. In our analysis, we inquire whether and how the interpretation of Art. 102 TFEU and the evolution of the law on evidence may have contributed to prolonging abuse proceedings, and how future guidelines may help in promoting a more ‘workable effects-based approach’ (B.).

A. Current shortcomings in numbers

We examine existing dysfunctionalities in the Commission’s enforcement of Art. 102 TFEU by empirically analysing the Commission’s decisions relating to proceedings under Art. 102 TFEU (or the previous Treaty provisions on abuse of dominance) since 1992. The dataset we use consists of a list of the published decisions relating to a proceeding under Art. 102 TFEU between 1992 and 20235 and features self-collected data that were taken from these decisions. We are recording for each decision: the name of the decision, the year the decision was issued, the duration of the proceeding in months, the length of the decision in pages, the type of the decision (e.g. commitment decision), and what triggered the proceeding (e.g. a complaint lodged by a competitor).

1. The overall enforcement activity

Since 1992, our dataset includes 67 Commission decisions relating to a proceeding under Art. 102 TFEU. For 32years, that amounts to a mean of 2.1 decisions per year. The following figure shows the mean number of decisions issued by the Commission per year, sorted by eight periods of four years each. It also shows the difference in the type of decision between prohibition decisions in adversarial proceedings and all types of decisions in cooperative proceedings (i.e. commitment decisions or prohibition decisions following a cooperative proceeding).

As illustrated by Fig. 1, the overall level of enforcement of Art. 102 TFEU remains relatively constant between 1996 and 2019. In this period, the annual number of decisions roughly fluctuates around 2.5. At a closer look, two trends are noticeable: first, after the entry into force, in May 2004, of Regulation 1/2003,6 the number of decisions in adversarial proceedings has dropped significantly, while the number of decisions in cooperative proceedings has risen. This reflects a change in the Commission’s enforcement practice in immediate reaction to the introduction of commitment decisions in Art. 9 of Regulation 1/20037 as a formalised cooperative tool and thus an alternative to the ‘normal’ enforcement procedure under Art. 7 of Regulation 1/2003.8

Secondly, the most recent period from 2020 to 2023 is particularly interesting. After an enforcement high from 2008 to 2019 (between 2.5 and 3 decisions per year compared with between 2 and 2.5 decisions per year from 1996 to 2007), the numbers have dropped significantly: from 2020 to 2023, the Commission only adopted one decision per year, on average. Further, in this period, the Commission did not adopt a single prohibition decision but commitment decisions only. This comes after a period from 2008 to 2019 in which the number of decisions in adversarial proceedings had risen.

2. The duration of proceedings

The duration of Art. 102 proceedings is an important parameter for assessing the public enforcement’s success. Where enforcement proceedings become excessively long, abusive conduct may not be ceased in a timely fashion, but continue to unleash its exclusionary potential until the competitive structure is harmed for good. As a result, and apart from purely allocational inefficiencies, Art. 102 TFEU may lose its deterrent effect—and thereby an important part of its preventive effectiveness.9 Public enforcement notwithstanding, dominant positions may be reinforced or even expand, and competing innovation paths may be permanently altered.

2.1. The development of the duration of proceedings over time

Fig. 2 shows the median duration of the Commission’s Art. 102 proceedings sorted by the same eight periods as above.10 We measure the duration of the proceeding in months.11 The end of the proceeding is the date of the decision. The starting point of the proceeding is the earliest (formal or informal) procedural step mentioned in the decision. This may be a formal step such as the Commission’s decision to initiate proceedings. However, it may also be the filing of a complaint with the Commission or an inspection at the office of an undertaking under investigation. In defining the starting point of the proceeding this way, we want to ensure that our measurement of the duration does not depend on the Commission’s discretion to formally open proceedings in a given case.12

Fig. 2 does not suggest any particular trend regarding the duration of proceedings, but rather a remarkable degree of stability: in the first half of the period under review, the median duration of proceeding was 50months; in the second half, it was 46months.

In the periods from 2008 to 2011 and 2020 to 2023, with a particularly high share in cooperative enforcement, the duration of proceedings was below average. However, given that cooperative proceedings have very much dominated Art. 102 enforcement since 2005, the fact that the duration of proceedings across all decisions only decreases by four months after 2004 is remarkable.

Taking a closer look at the median duration of adversarial proceedings versus cooperative proceedings in the period between 2004 and 201913 (Fig. 3), two findings strike the eye:

First, and hardly surprising, the median duration of adversarial proceedings is longer than the median duration of cooperative proceedings—at first sight, commitment decisions seem to deliver on the promise. The size of the margin varies, however. While in two periods, the margin is considerable (36months (2008–2011) and 53months (2016–2019), respectively), it seems rather small in the other two periods (five months (2004–2007) and one month (2012–2015), respectively).14

Secondly, while the duration of cooperative proceedings has decreased from 2004–2007 onwards,15 adversarial proceedings are getting longer: from 2004 to 2007, their median duration was 61months; from 2016 to 2019, it was 88months.16 The period from 2012 to 2015, where the median duration of adversarial proceedings was rather short, appears to be just an exceptional outlier that may be attributed to the particularities of a very small number of cases: between 2012 and 2015, our dataset includes only four decisions in adversarial proceedings.

2.2. The occurrence and distribution of exceptionally long proceedings

The development of the median duration of Art. 102 proceedings over time is not the only factor to be analysed. Additionally, a look at the spread in the duration of proceedings and the distribution of particularly long proceedings over time is called for. While there may always be outliers, a recurrence of excessively long proceedings may indicate problems in Art. 102 enforcement in and of itself. For example, looking at our dataset, 16 decisions have followed proceedings that were longer than 72months (or six years). Out of these 16 decisions, only six have been issued until 2007,17 while ten have been issued since 2008.18 The two last decisions in adversarial proceedings in our period under review, Google Search (AdSense) and Qualcomm (predation), are also those with the longest proceedings of all Commission decisions reviewed, with 111months (or 9.3years) and 122months (or 10.2years), respectively.

B. Possible reasons and the role of guidelines

The increase in the duration of adversarial proceedings and of the increasing number of excessively long proceedings calls for a reaction: where Art. 102 proceedings last six, eight, or even ten years, the Commission’s decision will typically be unable to remedy the resulting harm to the competitive structure, and the enforcement loses any deterrent effect. In highly dynamic sectors—like the digital sector—the dominant undertakings may, by their conduct, have created long-lasting barriers to market entry, and may have influenced for good the direction of investment and innovation. Even if one were to presume—which is not at all clear—that longer proceedings lead to more accurate outcomes and thereby reduce error costs in individual cases, the enforcement regime’s ‘systemic’ error costs increase as decisions that come too late to effectively remedy an abuse are ultimately analogous to ‘false negatives’. Also, if we assume a positive correlation between long proceedings and administrative costs, excessively long proceedings may result in a lower level of overall competition law enforcement.

1. Reasons for exceptionally long proceedings

An adequate policy reaction to the trend towards longer Art. 102 enforcement proceedings presupposes a sound understanding of the causes. Obviously, these causes can be manifold—both across all cases and in each individual case. First, complexities of procedural law and/or practice may play a role,19 just like procedural strategies—both by the Commission and by the undertaking under investigation.20 Regulation 1/2003 is currently being evaluated.21 A future reform may try to address some of the procedural causes.

Secondly, the type of abuse cases may have changed over time. For some time, Art. 102 TFEU cases primarily addressed strategies by former state monopolies to defend their monopolistic position in liberalised markets. More recently, an increasing number of Art. 102 TFEU cases deals with the conduct of firms that have gained their dominant position based on superior performance and innovative strategies. In these cases, Art. 102 enforcement must carefully consider the effects that competition law intervention may have on undertakings’ incentives to invest and innovate, which may complicate the distinction between pro-competitive and anticompetitive conduct. In this regard, it does not come as a surprise that out of the ten Art. 102 proceedings the duration of which stretched over six years and more, two proceedings concerned Google,22 and three proceedings concerned technology and Chip markets.23 Nonetheless, another three proceedings dealt with old style natural monopolies.24

Thirdly—and this is where our focus lies from now on—the increase in the duration of Art. 102 TFEU proceedings from 2004 to 2007 coincided with the Commission’s endeavour to develop a ‘more economic approach’ to Art. 102 TFEU which ultimately resulted in the publication of the 2008 Priorities Guidance. With the 2008 Priorities Guidance, the Commission committed to engage in a detailed, case- and context-specific effects analysis of likely foreclosure in essentially all cases.25 Although the Commission quickly gave up the attempt to offer direct proof of consumer harm and simply deduced it from a finding of an otherwise anticompetitive foreclosure instead,26 the Commission’s effects analysis shifted towards a significantly more extensive and often more quantitative assessment compared with the Commission’s decisions in the 1990s.27 While a quantitative effects analysis will not necessarily prolong proceedings,28 it may become a complex and time-consuming exercise at times. The Commission’s self-commitment to conduct an as-efficient-competitor (AEC) test even in rebate and exclusive dealing cases29—for as long as it lasted—was an additional factor that contributed to prolongations. But even the qualitative effects analysis has become much more elaborate and complex,30 not least in light of the Union Courts’ request for a concrete and context-specific analysis.

2. The role and limits of guidelines in addressing the inefficiencies?

With its 2023 Amending Communication, the Commission has partially withdrawn the 2008 Priorities Guidance’s costly policy choices, in particular the need to show direct consumer harm and the need to conduct an AEC test when analysing exclusivity rebates. However, over the last years the Union Courts have developed their own variety of an ‘effects-based approach’. Hence, one may question whether and to what extent there remains room for future guidelines to contribute to a speedier, and thereby more effective enforcement of Art. 102 TFEU. While Commission guidelines can redefine competition policy, they cannot change competition law.31

At the same time, one should not underestimate the influence guidelines may have on the evolution of competition law.32 In competition law, the Union Courts rule on the legality of Commission decisions in individual cases (Art. 263 TFEU), or they rule on the interpretation of the competition rules in the context of referral proceedings (Art. 267 TFEU). Even in referral proceedings, the questions raised by national Courts must be accompanied by a summary of the context in which the question arises. Whatever the type of proceeding, the Courts’ judgments must therefore be contextualised. This is also true for the seemingly broad and general statements on the interpretation of Art. 101 or Art. 102 TFEU that judgments often entail: In any given case, they are meant to provide reasoning for the concrete decision or the answer to a question referred. By implication, they, too, must be contextualised.

Against this background, guidelines are a useful instrument to propose a coherent and comprehensive interpretation of Art. 102 TFEU, to identify the generalisable substantive and evidentiary principles, to separate them from more context-specific findings and to fill conceptual gaps. The effort to systematise and conceptualise the Courts’ jurisprudence in this manner, and also to adapt the interpretation of Art. 102 TFEU to changing market realities,33 may be considered a core part of the Commission’s responsibility for the implementation and orientation of competition policy.34 In the following text, we try to identify fundamental substantive (III.) and evidentiary principles (IV.) of the law on abuse.

III. Abuse of dominance—conceptual cornerstones of the substantive law on abuse

A. The goals of Article 102 TFEU

Obviously, future guidelines on exclusionary abuses will have to start with a restatement of the goals of Art. 102 TFEU. In doing so, they will very likely not dive into old battles. The drafting of the 2008 Priorities Guidance had been accompanied by an intense discussion about the objectives of European competition law in general and Art. 102 TFEU in particular.35 Proponents of a ‘more economic approach’ aimed to settle on the consumer welfare goal as a clear and measurable reference point for EU competition law.36 According to this approach’s opponents, EU competition law was to hold on to the broader goal of protecting open markets and the competitive process37 and to be guided by the overarching objective of the EU to establish and protect a well-functioning internal market with undistorted competition.38

The 2008 Priorities Guidance did not address the dispute on a theoretical level, but it let the Commission’s enforcement agenda revolve around a welfare maximisation metric: (only) those cases were to be prioritised in which ‘cogent and convincing evidence’39 suggested likely anticompetitive foreclosure with ‘an adverse impact on consumer welfare’ to be identified based on ‘qualitative and, where possible and appropriate, quantitative evidence’.40

At first sight, the ECJ’s determination in Servizio Elettrico Nazionale that the ‘ultimate objective’ of Art. 102 TFEU is the ‘well-being of consumers—both intermediate and final consumers’41 seems to follow that logic. However and importantly, the sentence is embedded in a broader explanation of the objectives and the structure of Art. 102 TFEU: Art. 102 TFEU is characterised as one of a series of rules ‘intended to prevent competition from being distorted contrary to the public interest and to the detriment of individual undertakings and consumers, and thus to contribute to economic welfare in the European Union’. Moreover, the ECJ refers to the established case law that Art. 102 TFEU ‘seeks to sanction not only practices likely to cause direct harm to consumers but also those which cause them harm indirectly by undermining an effective structure of competition’.42 From this passage, it follows that harm to consumer welfare—uncontroversially broadly construed43—is an objective, but not a precondition for finding an abuse. The ECJ seems to have rephrased what Richard Posner has observed for US antitrust law quite some time ago: ‘Efficiency is the ultimate goal of antitrust, but competition a mediate goal that will often be close enough to the ultimate goal to allow the Courts to look no further’.44 It is only at the stage of the ‘efficiency defence’ that the dominant undertaking may invoke an absence of consumer harm;45 and in fact, it is (only) the efficiency defence that requires a formal acknowledgment of consumer welfare within the system of EU competition law’s goals.

It follows that, Servizio Elettrico Nazionale notwithstanding, conduct may be considered to be prima facie abusive even where proof of consumer harm may be difficult: Servizio Elettrico Nazionale was hardly intended to overrule the case law on a dominant undertaking’s refusals to supply wholesalers engaging in parallel exports, for example.46 Conduct that clashes with the EU’s overarching market integration goal47 will arguably continue to be outlawed, with particularly high demands on an efficiency defence in view of the normative weight of the EU’s internal market policy.

It is a separate question whether EU competition law additionally pursues broader political and societal aims. According to some,48 this is what the Commission’s Amending Communication suggests when it states that the competition rules can contribute to ‘achieving objectives that go beyond consumer welfare, such as plurality in a democratic society’.49 The Policy Brief further refers to the Union Courts’ case law—in particular to Google Android50—for its claim that ‘ensuring consumer choice is a means to ultimately guarantee plurality in a democratic society’. In addition, Commissioner Vestager’s frequent references to goals such as ‘fairness and level-playing field, market integration, preserving competitive process, consumer welfare, efficiency and innovation, and ultimately plurality and democracy’ are quoted.51

As Massimiliano Kadar and Johannes Holzwarth have recently clarified,52 these references should not be read as an attempt to expand the objectives of Art. 102 TFEU. Rather, they amount to a recognition of interdependencies in a unitary legal order: an effective enforcement of competition law can—for example and of course—function as a protective barrier against excessive concentration and the redistributive effects associated with it. To this day, the European debate has not joined the Neo-Brandeisian chorus across the Atlantic that aims for a formal expansion of goals,53 nor should it. Even more clearly, it cannot be the task of future guidelines on exclusionary abuses to redefine the objectives of competition law.

B. The concept of abuse—which criteria?

1. An ‘effects-based approach’— a concept in need of specification

According to a broad consensus, any approach to exclusionary abuses will need to be ‘effects-based’ of some sort. The requisite effects relate to a given practice’s capability to restrict competition54—not to consumer harm. The ECJ’s jurisprudence has been consistent in this respect from the beginning.55 What has changed over the last 15years or so are the requirements for showing such capability. In its earlier jurisprudence, the ECJ was satisfied with a rather abstract, rule-based approach—at least with regard to some types of conduct: the use of loyalty rebates or exclusive dealing arrangements by a dominant undertaking sufficed for presuming an abuse. In defence, the dominant undertaking could rely on an objective justification and, from British Airways56 onwards, on an efficiency defence. In its more recent jurisprudence the ECJ has significantly tightened the evidentiary requirements: a capability to restrict competition must be shown in concreto, i.e. in the circ*mstances of the case at issue,57 based on an analysis of the market(s) in question and an understanding of the functioning of competition on that or those market(s).58 It must not merely be hypothetical,59 and it cannot be deduced from an ‘analytical template’ based on the type of conduct alone.60 With a view to loyalty rebates and exclusive dealing, the Commission or a competition authority may still, in a first step, rely on a relatively abstract analytical template to provisionally infer an abuse. But if, based on supporting evidence, the dominant undertaking submits that its conduct was not capable of restricting competition, the competition authority must ‘examine whether, in the particular circ*mstances, the conduct in question was indeed capable of doing so’.61 It is this requirement to show potential anticompetitive effects in each single case and with a substantial degree of context-specificity (at least where they are contested during the administrative proceeding with a sufficient degree of substantiation—on this see below, IV. B. 2.) which is at the heart of the Union Courts’ version of an effects-based approach: the judicial authorisation to presume anticompetitive effects based on certain types of conduct alone has been withdrawn.

From an enforcement perspective, it is difficult to overstate the relevance of this shift. Nonetheless, the case law is open to different interpretations. On the one hand, it might imply that, at least in adversarial settings, an abuse analysis will always—or almost always62—entail a full-fledged economic analysis of any and all potentially relevant facts,63 i.e. of all facts that may potentially have some impact on the ‘competitive structure’. In that case, the main function of the law is to provide for a procedure within which both competition authorities and the undertaking under investigation can produce and present economic evidence, and that ensures access to the file and the right to be heard. If—and only if—a competition authority has duly considered all facts and evidence submitted, it may, in the final assessment of the effects, benefit from a limited margin of appreciation. Almost by necessity, such an approach will tend to prolong proceedings in all settings where dominant undertakings may benefit from such prolongations. Large amounts of complex evidence will then be produced that enforcers will feel compelled to engage with in order not to risk losing the case in Court. Based on this sort of an effects-based approach, competition authorities will be able to handle less and less cases. Economic complexity and the cost of enforcement are maximised. Legal certainty—a major building block of an effective protection against distortions of competition—is reduced.

A different reading would suggest that the case law on exclusionary abuses has not outright abandoned a reliance on simplifying rules that implicate an inference of anticompetitive effects, but that these rules have become more differentiated and sensitive to context within a structured proceeding in which the burden to produce evidence and the burden of proof may shift to and fro.64 An important challenge in ensuring an effective enforcement of Art. 102 TFEU would then lie in redefining and specifying the preconditions under which anticompetitive effects can be initially inferred or even presumed,65 what kind of evidence must be produced to counter the inference or presumption, and what additional evidence may then revive the case. There is no basis for a general argument that inferences and presumptions should be construed restrictively, such as to minimise ‘false positives’ in any given case. Where the beneficial settings are difficult to separate from the harmful ones and harmful effects are frequent, a legal requirement of economic precision case by case can come at the cost of systemic underenforcement.66 Broader inferences may lower overall error costs. Consequently, the challenge for an effects analysis within a legal proceeding is not to burden enforcement institutions with the task to fully replicate economic complexity. Rather, account must be taken of the way legal rules function: if their aim is to minimise error cost at the systemic level, the calculus must include their effect on norm addressees and enforcement institutions.67 This is the interpretation we endorse.

2. Anti-competitive effects—preconditions for establishing an abuse

The concept of exclusionary abuses refers to conduct that is capable of producing anticompetitive effects.68 The 2008 Priorities Guidance has equated anticompetitive effects with a conduct’s capability of foreclosing AECs. The Union Courts’ case law suggests a broader approach: in order to qualify as an abuse, the relevant conduct must, first, have the (actual or potential) effect of ‘hindering the maintenance of the degree of competition still existing in the market or the growth of that competition’. However, a lessening of competition, namely a foreclosure of less efficient competitors, may result from the very process of competition that competition law aims to protect—an old and fundamental insight that has recently been dubbed the ‘AEC principle’ (see below, III. B. 4.). A second criterion is therefore needed to distinguish between pro-competitive and anticompetitive practices. Since Hoffmann-La Roche, the Union Courts require that adverse effects on competition must result from ‘recourse to methods different from those which condition normal competition in products or services’.69 Unfortunately, this second criterion is sometimes difficult to handle. At a relatively abstract level, almost any type of ‘normal’ commercial conduct can be turned into an instrument of exclusion by a dominant undertaking, depending on the circ*mstances and the precise form the conduct takes. This is true, for example, for price competition, refusals to deal, or product design. More recently, the Union Courts tend to ask whether the relevant practice lies within or outside the scope of ‘competition on the merits’.70 But that concept, while arguably of increasing importance,71 is in need of explanation itself.

We will discuss these two criteria (restrictive effects and no competition on the merits) separately below. While it is true that some tests (in particular the AEC test) let the two criteria collapse72 (see below, III. B. 4. 1.) and that, ultimately, ‘anticompetitive effects’73 must be established—a term that conflates both criteria, there are settings where the ‘no competition on the merits’ criterion gains its own weight.

When the two criteria are met, there is a prima facie case for an exclusionary abuse. Nonetheless, the dominant firm can still come forward with an objective justification or efficiency defence.

3. Conduct capable of hindering competition—the ‘effects’-criterion in Art. 102

An abuse of dominance presupposes that a dominant undertaking engages in conduct that is capable of hindering competition. While the Union Courts’ case law has fleshed out the concept in a series of judgments, a relevant degree of fuzziness persists. Part of this fuzziness may be inevitable. Nonetheless, some clarification is called for.

3.1. Adverse effects on an effective competition structure

In Hoffmann-La Roche, the concept of abuse was defined as ‘an objective concept relating to the behaviour of an undertaking in a dominant position which is such as to influence the structure of the market’ and which ‘through recourse to methods different from those which condition normal competition’ has the effect of ‘hindering the maintenance of the degree of competition existing in the market or the growth of that competition’74—a formula much repeated since then. More recently, the ECJ also refers to practices ‘capable of adversely affecting, by way of resources other than those which govern normal competition, an effective competition structure’.75 By implication, this effects-criterion does not relate to consumer harm (see above, III. B. 1.). Rather, what matters are the conduct’s potential effects on barriers to market entry or expansion,76 e.g. by hampering actual or potential competitors’ effective access to resources, data, or customers, by removing or restricting the buyers’ freedom as regards choice of sources of supply,77 or by otherwise creating competitive disadvantages for actual or potential rivals, thereby shielding the dominant undertaking from competition, reducing competitors’ incentives to innovate78 or enabling the dominant undertaking to expand into adjacent markets irrespective of the superior attractiveness of its offer. As these examples illustrate, the concept of ‘adverse effects on the competition structure’ is much broader than the concept of (full) foreclosure or marginalisation of competition.79 The potential of a practice to cause adverse effects on the competition structure will reinforce the risk implicit in the position of dominance itself, namely the risk that the dominant undertaking will negatively influence, ‘to its own advantage and to the detriment of consumers, the various parameters of competition, such as price, production, innovation, variety, or quality of goods or services’.80 Consequently, potential adverse effects on the competitive structure come with a risk of consumer harm. But they must not be confused with consumer harm: as such, the issue of consumer harm will only be addressed in the context of the efficiency defence.

Nor must the concept of ‘competitive structure’ be confused with ‘market structure’.81 An analysis of the (actual or potential) effects of a dominant undertaking’s practice on the ‘competitive structure’ will focus on effects on the ‘structural’ preconditions that determine how competition functions in the relevant markets. Some refer to effects on the ‘competitive process’ in this regard,82 i.e. to the process that results from this structure. By focusing on the ‘competitive structure’, the Union Courts emphasise the need to focus not on short-term effects, but to adopt a dynamic perspective.

3.2. Actual or potential effects

The finding of an abuse does not presuppose proof of actual anticompetitive effects:83 anticompetitive conduct can be prohibited already before harm to competition has occurred. EU competition rules are meant to react to the risk for competition that certain conduct raises.84 This raises the question of what level of risk—or which probability of anticompetitive effects—will trigger the abuse prohibition. Interestingly—and contrary to the law on merger control85—the Union Courts have been unwilling to establish a clear substantive standard of probability, however. Instead, the case law refers to ‘potential’,86 ‘probable’,87 or ‘likely’88 effects, to the capability of the conduct to cause anticompetitive effects,89 or to conduct that ‘tends to’90 cause adverse effects seemingly interchangeably.91 In Post Danmark II, where the referring court explicitly asked how likely the anticompetitive effect of a rebate scheme must be in order to constitute an abuse, and where AG Kokott, in her Opinion, argued that, ‘on the basis of an overall assessment of all the relevant circ*mstances of the individual case, the presence of an exclusionary effect [must appear] more likely than its absence’,92 the ECJ merely repeated that the anticompetitive effect must be likely,93 thus declining to establish a clear probability threshold. Under the substantive law on Art. 102 TFEU, the ECJ holds on to a vague ‘capability standard’,94 thereby delegating genuine probability assessments to the standard of proof.

Simultaneously, the ECJ is determined in its requirement of concreteness: The alleged anticompetitive effects must not be ‘purely hypothetical’,95 i.e. potential effects must not be analysed in the abstract. The determination of effects ‘may entail the use of different analytical templates depending on the type of conduct’; but even then, it must be made ‘in the light of all the relevant factual circ*mstances’, including ‘the conduct itself, the market(s) in question [and] the functioning of competition on that or those market(s).’96 Ultimately, the standard of ‘concretely likely’ adverse effects on competition requires competition authorities and Courts to identify a context-specific97 theory of how precisely the relevant conduct is capable of triggering adverse effects on competition and verify whether the market conditions on which this theory is based are met or at least likely to arise.98 According to Elias Deutscher, a ‘realistic prospect’ of anticompetitive effects is required.99

The relatively vague ‘capability’ standard comes with the disadvantage of uncertainty. Simultaneously, it makes the prohibition of Art. 102 TFEU sensitive to the potential gravity or magnitude of harm to competition that is at issue:100 Along that line, practices that may plausibly lead to a long-time, full foreclosure of competition may qualify as abusive even where they are merely capable of, but not ‘more likely than not’ to produce such effects—in particular if these practices cannot be shown to come with significant pro-competitive benefits. While this does not amount to an attempt to quantify the expected harm (as well as the expected pro-competitive benefits),101 it opens the possibility for some degree of balancing of likely harm and likely benefits, possibly including the question whether any pro-competitive benefits could be realised by other, more pro-competitive means. Simultaneously, the requirement to analyse the likely effects in light of all relevant facts of the case underlines that Art. 102 TFEU shall not capture abstract, but concrete risks to competition.102 This does not exclude the use of inferences based on economically informed experience and/or sound economic theory. But such inferences must remain subject to the challenge that, given the circ*mstances of the case, no concrete risk to competition exists.

In light of this case law, Art. 102 TFEU’s reference to ‘potential effects’ should therefore be read to clarify two things: a dominant firm’s conduct may qualify as abusive irrespective of whether actual effects have already materialised. But the risk of anticompetitive effects must be plausible and concrete in light of the facts of the case—without there being a quantified level of probability.

The requirement of concreteness marks the Union Courts’ departure from Hoffmann-La Roche103 and its shift towards an ‘effects-based approach’: In Hoffmann-La Roche, the ECJ was willing to accept, for certain types of conduct, analytical templates that inferred potential anticompetitive effects from certain forms of conduct alone, at least in typical settings.104 According to the ECJ’s more recent jurisprudence, ‘the abusive nature of a practice does not depend on the form it takes or took, but presupposes that that practice is or was capable of restricting competition and, more specifically, of producing, on implementation, the alleged exclusionary effects’, where this condition ‘must be assessed having regard to all the relevant facts’.105

3.3. Causality of conduct for adverse effects

In order to find an exclusionary abuse, some sort of causal link between the dominant undertaking’s conduct and the (actual or potential) anticompetitive effects must be established: the effects must be attributable to the dominant undertaking’s conduct.106 Typically, causality is proven by reference to a ‘counterfactual’: a competition authority will compare the likely state of competition with the allegedly abusive conduct with an alternative scenario without that conduct. In the 2008 Priorities Guidance, the Commission committed to identifying a suitable ‘counterfactual’ in each Art. 102 case.107 Obviously, all that can be required is a plausible causality,108 i.e. a showing that the conduct was capable of producing or likely to produce actual or potential effects.109

In their recent Policy Brief, DG Comp officials have questioned the commitment to establish a clear counterfactual, at least where the market realities as they might have evolved without the allegedly abusive practice become speculative: in such a setting, requiring the Commission to establish a ‘nexus of full causality’ between conduct and anticompetitive effect would risk to render Art. 102 enforcement ‘unduly burdensome or even impossible’.110 For this position, the Policy Brief relied on findings by the GC in Google Shopping and in Google Android: in Google Shopping, the GC highlighted that in some settings, trying to determine the dominant undertaking’s conduct absent the alleged abuse can become highly speculative.111 Consequently, it was satisfied with a showing of a correlation between the relevant conduct and a ‘modification of competition’ on the relevant markets and ‘additional information, which may include … the assessments of market participants’.112 In Google Android, the GC confirmed that, where a restriction of competition and its anticompetitive effects were otherwise shown, the Commission was not required to carry out a counterfactual analysis.113

By contrast, the GC quashed the Commission’s Qualcomm decision, both on procedural and substantive grounds. According to the GC, the Commission had not established that the competitive situation would have been better without Qualcomm offering exclusivity payments to Apple conditional upon Apple’s commitment to purchase its entire requirement of LTE chipsets for iPhones from Qualcomm as compared with the situation with these exclusivity payments, because Apple had no technical alternative to Qualcomm’s LTE chipsets.114

Given this tension in the GC’s more recent case law, a debate has ensued whether a counterfactual analysis is indeed required in Art. 102 cases,115 and if so, in what form, or whether the causality between conduct and anticompetitive effects can also be established otherwise. In a careful enquiry, Elias Deutscher has set out why requiring a counterfactual analysis in the strict sense may at times be an overly demanding standard that is bound to lead to type II errors when applying Art. 102 TFEU:116 inter alia, it will not take account of potential concurrent causes of anticompetitive foreclosure, i.e. of settings, where high barriers to entry and high switching costs may themselves be a relevant and possibly sufficient cause for the absence of competition, but where additional anticompetitive strategies may further decrease contestability and establish an additional layer of protection for an entrenched position of dominance; and—if wrongly conceptualised, as arguably happened in Qualcomm—a counterfactual analysis can run counter to the ‘capability standard’ of anticompetitive effects, i.e. to the fact that all that is required is that the conduct was ‘capable of’ producing actual or potential effects. This implies that the Commission need not settle on the most likely counterfactual but may simultaneously consider ‘but for’ scenarios with a probability of less than 50 per cent—provided they are realistic.117 For example, it may consider that, absent the allegedly abusive conduct, market entry would have occurred, even if not under a ‘more likely than not’ standard, due to high barriers to entry. By contrast, proponents of a counterfactual analysis will typically require the Commission to make the most likely ‘but for’ scenario its reference point—which will frequently be the status quo ante. Often enough, this will be a situation of entrenched dominance where entry cannot be expected with a 51 per cent probability. The analysis will then tend to neglect the adverse impact of the dominant undertaking’s conduct on dynamic competition and contestability: implicitly, the ‘capability standard’ of anticompetitive effects would turn into a ‘likely actual effects’ standard.118

All this cannot justify abandoning the need to establish a causal link on the basis of a ‘but for’-test, i.e. the need for some sort of counterfactual analysis.119 However, in the context of a potential effects analysis, particular caution is required: a showing that the relevant conduct was plausibly capable to contribute to anticompetitive effects in the specific context of the case must suffice; and/or that it contributed to making market entry or an expansion of competitors more difficult.120 Along these lines, it may be enough for the Commission to show a positive correlation between the allegedly anticompetitive conduct and anticompetitive effects that is not easily explained by other factors, or to show that, absent the conduct, a realistic prospect of market entry would have existed.121

4. No ‘competition on the merits’

Establishing adverse effects on the competitive structure does not suffice for establishing an abuse (see above, III. B. 2.). In addition, any actual or potential exclusionary effects must not result from ‘normal’ competition, i.e. from ‘competition on the merits’122 or ‘performance-based’ competition.123

Despite Hoffmann-La Roche’s reference to ‘methods different from … normal competition’, ‘normal’ competition and practices that fall outside the scope of ‘competition on the merits’ cannot be identified based on their form alone: the concept of abuse does not (only) refer to practices that are inherently improper,124 or that are always improper when applied by a dominant undertaking. Rather, depending on the circ*mstances of a case, almost any type of otherwise normal commercial conduct can be turned into an instrument of anticompetitive exclusion by a dominant undertaking, including price competition, product design, or innovation. Consequently, ‘competition on the merits’ must not be read as referring to a list of practices that dominant undertakings must not use. Instead, it stands for the general idea that, first, dominant undertakings shall compete on account of their skills, abilities, and superior performance,125 striving to offer consumers the best value for money or most attractive product or deal instead of competing by obstructing their competitors’ effective access to markets, resources, data, or customers or by otherwise creating competitive disadvantages for actual or potential rivals; and, secondly, that they must consider the effects of their conduct, due to their ‘special responsibility’ not to allow their conduct ‘to impair genuine undistorted competition on the common market’.126

In some settings, these principles alone will suffice to identify abusive conduct. As the ECJ has put it in Servizio Elettrico Nazionale: ‘conduct which has the effect of broadening consumer choice by putting new goods on the market or by increasing the quantity or quality of the goods already on offer must, inter alia, be considered to come within the scope of competition on the merits’.127 By contrast, practices which cannot be plausibly explained as performance-based competition, but only by the dominant undertaking’s interest to eliminate competitors—i.e. practices that, in the context of a given case, neither contribute to productive efficiencies nor to enhanced consumer attraction, but tend to foreclose competitors—will constitute an abuse under Art. 102 TFEU. The inquiry into the underlying object of a dominant firm’s strategy is frequently referred to as the ‘no economic sense’-test—a test that does not define competition that is not on the merits128 but helps to identify clear cases of so-called ‘naked exclusion’. In such cases, competition authorities may infer anticompetitive effects (see below, IV. C. 2.). The two criteria for establishing an abuse then collapse.

Frequently, a dominant undertaking’s potentially exclusionary practices or strategies will come with some potential efficiencies. In those cases, additional tests and criteria for distinguishing between pro-competitive and anticompetitive conduct will be needed:129 the concept of ‘competition on the merits’ merely serves as a general umbrella for these tests which may look different in different settings. In the Union Courts’ more recent jurisprudence, the so-called ‘AEC principle’ has gained prominence (4.1.). But given a dominant undertaking’s special responsibility not to undermine effective and undistorted competition in the internal market, even the AEC principle comes with limitations (4.2.). In other words, ‘competition on the merits’ is a concept with different facets, some of them more, some of them less normatively loaded. Practices that do qualify as ‘competition on the merits’ exclude a finding of abuse. On the other hand, practices that cannot be explained as a legitimate part of performance-based competition and result in the foreclosure of competitors, or where the pro-competitive justification is clearly out of proportion to the anticompetitive effects, will typically qualify as an abuse. What is more, a lower evidentiary threshold for proving anticompetitive effects may apply to practices that are clearly no part of performance-based competition (more on this, see below, IV. C.).

4.1. The AEC principle and the AEC test

Competition ‘may, by definition, lead to the departure from the market or the marginalisation of competitors that are less efficient and so less attractive to consumers from the point of view of, among other things, price, choice, quality, or innovation’.130 Consequently, a practice’s exclusionary effects alone cannot constitute an abuse. In Servizio Elettrico Nazionale, the ECJ has recently restated this well-established tenet—lately referred to as the ‘as-efficient competitor’-principle (AEC principle): it is ‘in no way the purpose of Art. 102 TFEU to prevent an undertaking from acquiring, on its own merits—on account of its skills and abilities in particular—a dominant position on a market, or to ensure that competitors less efficient than an undertaking in such a position should remain on the market.’131 This insight applies to both price-based and non-price-based practices.132

The AEC test—or rather the different varieties of AEC tests that have been developed for different types of conduct133—is an attempt to make the AEC principle operational by taking cost as the metric for efficiency.134 The 2008 Priorities Guidance has committed to regularly conducting an AEC test in cases of price-based conduct.135 In predatory pricing136 and margin squeeze cases,137 an AEC test, i.e. a cost–price-comparison based on the cost structure of the dominant firm, typically is the most straightforward way to distinguish between efficiency-based competition on the merits and anticompetitive exclusion. According to a consistent line of cases, a foreclosure of AECs qualifies as sufficient condition for an abuse,138 absent an objective justification or efficiency defence.139 The implementation of the AEC test will normally be manageable for the Commission.140 Simultaneously, all relevant information is known to the norm addressee.141 Consequently, the AEC test has become, for these types of cases, the standard analytical template to identify an exclusionary abuse.142 One of its advantages is that it lets the two legs of the abuse analysis collapse: the identification of ‘anticompetitive foreclosure’ is based on one single criterion. Prices below some measure of the dominant undertaking’s own cost do not qualify as ‘competition on the merits’. Simultaneously, they are capable of adversely affecting the competition structure.143

Attempts to adapt the AEC test to other types of conduct come with significant complications, however.144 First, the AEC test quickly loses its objectivity when applied to non-price-related practices, given that a hypothetical AEC would need to be comparable with the dominant firm not only in terms of cost structure, but also in terms of capacity to innovate, quality, and other respects.145 Secondly, one of the great advantages of the AEC test in price-based cases—the ability of the dominant undertaking to conduct the test itself—is no longer present when the test is applied to non-price-based conduct. Rather, additional information will be needed that lies outside the dominant undertaking’s sphere. In the case of conditional rebates, for example, the ‘contestable share’ must be determined, i.e. the share of a customer’s purchase requirements that can realistically be switched to a new competitor in a given time.146 Consequently, an AEC test becomes much more complex, time-consuming, uncertain, and controversial. Given the GC’s unwillingness to grant the Commission a margin of appreciation, there is a significant risk of losing a case in Court.147 The complex version of the AEC test proposed for conditional rebates in the 2008 Priorities Guidance and the difficulties the Commission has experienced in putting it into practice testify to the limited practicability of the AEC test in these cases.148 The Commission’s Intel decision is illustrative: the AEC test spans 151 pages149 or 29.2 per cent of the total decision length; respectively, 68.3 per cent of the pages that deal with the finding of an abuse. It has been estimated that conducting the AEC test alone took up two years.150 Furthermore, with the growing complexity, the risk of error increases,151 and with regard to any error or uncertainty, the GC grants the defendant the benefit of the doubt.152

Against this background, the 2008 Priorities Guidance’s proposition to make the AEC test a central and preferred analytical methodology even beyond purely price-based conduct153 has long been abandoned.154 After Intel and Google Android, the Commission has refrained from conducting AEC tests outside the realm of predation and margin squeeze cases.155 In doing so, it can rely on a consistent jurisprudence that acknowledges the Commission’s discretion in choosing the most suitable methodology for identifying anticompetitive conduct.156 Simultaneously, the more cautious assessment of what the AEC test can achieve is confirmed by a literature that increasingly acknowledges its limits.157 Importantly, the AEC test does not inquire into the pro-competitive or anticompetitive incentives that may have informed the dominant undertaking’s conduct, and does not provide any theory of harm:158 in the context of some theories of harm (like ‘competition softening’), the AEC principle and test are not informative and may lead to type II errors.159 In the context of other theories of harm, they may lead to type I errors.160 Consequently, as emphasised by Barbera, Fajardo Acosta, and Klein, ‘there is no one-size-fits-all answer to the question of whether and to what extent the AEC principle and tests remain appropriate and useful in discerning anticompetitive conduct.’161

4.2. Limits of the AEC principle (and of AEC tests)

Beyond the AEC test, the AEC principle serves as an important reminder of the aim and function of Art. 102 TFEU. But like the AEC test, it is not conclusive:162 while—at least according to the Union Courts’ case law163—a foreclosure of AECs is a sufficient condition for finding an abuse,164 competition may also be harmed by the exclusion of less efficient competitors.165 As recently reiterated by the ECJ in European Superleague,166 conduct may be categorised as an abuse ‘not only where it has the actual or potential effect of restricting competition on the merits by excluding equally efficient competing undertakings from the market(s) concerned, but also where it has been proven to have the actual or potential effect—or even the object—of impeding potentially competing undertakings at an earlier stage … from even entering that or those market(s) and, in so doing, preventing the growth of competition therein to the detriment of consumers, by limiting production, product, or alternative service development or innovation’.167 Consequently, where the structure of the market—in particular high barriers to entry—makes it particularly difficult for an entrant to achieve a comparable level of efficiency,168 including due to strong positive network effects or economies of scale, where, as a consequence, the presence of the dominant undertaking weakens competition such that as-efficient competitors cannot emerge, or where the dominant position results from legal privileges that the undertaking enjoyed in the past, in particular from monopoly rights,169 the AEC principle will not limit the application of Art. 102 TFEU:170 EU competition law recognises that even less efficient competitors can exert competitive pressure such that ‘the market structure and the choices available to consumers do not deteriorate further’171 and that, contrary to the position adopted in the 2008 Priorities Guidance, the AEC principle therefore does not serve as a ‘soft safe harbour’.172 Where barriers to entry are high and competitive pressure from as-efficient competitors is structurally weak, the negative effects on productive efficiency as they may result from the entry or continued presence of less efficient competitors are considered less important than their potentially positive impact on allocative and dynamic efficiency.173 As the notion of ‘competition on the merits’ encompasses the possibility of such balancing, it is broader than the AEC principle and the AEC tests that help operationalising its meaning.

IV. Effects analysis—standard of proof, assessment of evidence, and analytical shortcuts

The brief sketch of the basic legal structure of Art. 102 TFEU shows a considerable degree of consolidation in some areas of the law; significant uncertainties persist in others. Most importantly, the principles that are to guide the effects analysis remain a puzzle. While there is room for future guidelines to clarify the law and to push it into an appropriate direction, the substantive law on abuse will always remain a ‘standard’, not a ‘rule’.174

There is reason therefore for future guidelines to venture beyond the incoherencies and uncertainties of substantive law. Arguably, the uncertainties surrounding the evidentiary standards likewise contribute to exceptionally long Art. 102 proceedings. In the following text, we propose to clarify first the standard of proof (A.) and subsequently the conceptual framework for the assessment of evidence (B.). Our focus will be on analytical shortcuts for determining potential anticompetitive effects (C.). Our main claim is that a major, if not the most important, contribution of future guidelines on exclusionary abuses to a more effective approach to Art. 102 TFEU may be to further clarify and develop these analytical shortcuts. Ultimately, the evidentiary framework for the effects analysis should meet a number of requirements: in line with sufficiently reliable economic experience and knowledge, the analysis must establish a concrete risk of anticompetitive foreclosure; at the same time, it must remain manageable such as to enable an enforcement that considers the enforcing institutions’ constraints of time and resources and the need to enforce Art. 102 TFEU in a timely fashion.

A. The standard of proof

As a matter of substantive law, the finding of ‘potential anticompetitive effects’ presupposes that a concrete and real, not merely hypothetical, risk of anticompetitive effects is established. The Union Courts have rejected a specification of the level of risk in probabilistic terms (see above, III. B. 3. 2.). In principle, this substantive threshold is to be distinguished from a specification of the standard of proof, i.e. the degree of probability with which the presence of relevant preconditions of an abuse must be proven as a matter of procedural law:175 According to Art. 6 ECHR, proceedings which allow for the imposition of fines176 and which therefore qualify as quasi-criminal177 are governed by a ‘beyond a reasonable doubt’ standard.178 The Union Courts frequently refer to a ‘full conviction’ standard instead. While it may not be synonymous with a ‘beyond a reasonable doubt’ standard on a theoretical level,179 it is sufficiently flexible to accommodate that standard in its practical implementation.180

Theoretically, a flexible, yet relatively moderate substantive threshold of anticompetitive risk must therefore be proven under a rather demanding standard of proof. In practice, the Union Courts tend to conflate the substantive threshold and the standard of proof, however—in particular when they assume that, in establishing ‘likely’ or ‘probable’ anticompetitive effects, the norm addressee will benefit from a presumption of innocence. In fact, the presumption of innocence applies to the standard of proof only, but presupposes that the substantive legal threshold for ‘potential anticompetitive effects’ under Art. 102 TFEU is pre-defined.181 If the Union Courts apply the presumption of innocence to the determination of potential anticompetitive effects nonetheless, this can only mean that a demanding standard of proof will apply with regard to the facts that substantiate the existence of a relevant and sufficiently concrete risk to undistorted competition.182

The causal hypothesis meant to establish an infringement of Art. 102 TFEU is not treated as a fact to be established to that same standard of proof. It is rather treated as a specification of the substantive norm for which the Commission possesses some margin of appraisal. Consequently, a ‘manifest error of assessment’ standard applies. This is how Unilever Italia must be read when it finds that the demonstration of effects must ‘be based in tangible evidence which establishes, beyond mere hypothesis, that the practice in question is actually capable of producing such effects, since the existence of doubt in that regard must benefit the undertaking which engages in such a practice’:183 the requirement of ‘tangible evidence’ applies to the facts—but the evidence must prove no more than that the relevant conduct is capable of producing anticompetitive effects ‘beyond mere hypothesis’.184

B. Assessment of the evidence

When it comes to the assessment of the evidence, the principle of unfettered evaluation of evidence applies: generally, the Union Courts are free in assessing the probative value they assign each single piece of evidence (or a set of evidence) and, ultimately, the body of evidence as a whole, with possible interconnections and corroborations between the different pieces of evidence.185 The law on evidence does not stipulate a general prioritisation of quantitative over qualitative evidence. From an evidentiary perspective, both come with equal weight. What ultimately matters is that, based on the totality of the evidence, the Court is fully convinced that the relevant conduct is capable of producing anticompetitive effects.

Beyond these basic principles, a conceptual framework for the assessment of evidence—and in particular circ*mstantial evidence on potential anticompetitive effects—is notably absent. Nor have the Union Courts developed a clear terminology. This has led to a significant degree of uncertainty—an uncertainty that induces the Commission to engage in an ever more expansive effects analysis in order to reduce the risk of later losing the case in Court.

1. Inferences, presumptions, and other analytical shortcuts—preliminary remarks on terminology

In the majority of cases, the Commission will focus its inquiry on potential effects on the competitive structure. As potential effects cannot be proven directly, some sort of indirect evidence will be required: proof of anticompetitive effects will be based on indicia from which it can be inferred that anticompetitive effects are sufficiently likely. The selection of indicia will derive from a causal hypothesis—frequently referred to as the ‘theory of harm’—which must be plausible and sufficiently reliable. Absent an established legal presumption that would direct the Court to infer anticompetitive effects on a normative basis, the causal hypothesis, together with the proof of the facts on which it is based, must lead to the full conviction of the Court that anticompetitive effects are sufficiently likely (see above, IV. A.). The defendant can question the general plausibility and/or reliability of the theory of harm; it may argue that it does not fit the special setting of the case, or it may try to show that the facts do not support the causal hypothesis. Finally, it may try to disprove anticompetitive effects, e.g. based on an AEC test. The Courts will ultimately decide whether they are convinced by the causal hypothesis and by the facts that are presented to substantiate it.

Sometimes, an abbreviated analysis may suffice to infer potential effects—namely if, based on the economically informed experience of the Court, the combination of a small, selected number of factors (e.g. a certain type of conduct, combined with a certain degree of dominance and a certain market coverage of a practice) leads to a high probability of harm to competition. We refer to such inferences as ‘factual inferences’.186 The legitimacy of a factual inference depends on the existence of a strong empirical link between a pre-selected fact A (or a set of facts A1, A2, and A3) to be proven and a fact B—in our case: potential anticompetitive effects—which is inferred. Where such a link is known to exist, factual inferences may justify an abbreviated assessment of the evidence for a Court to reach a full conviction that a certain type of conduct will likely lead to anticompetitive effects. A defendant may try to raise doubts about the strength of the empirical link between the relevant set of facts and the fact inferred, either generally or in the context of the specific, potentially atypical, case; or may resort to challenge the plausibility of the inferred fact directly—for example by conducting an AEC test to show that, despite the factual inference, anticompetitive effects are unlikely. In such a case, the Commission will need to engage with the evidence brought forward. But provided that the existence of an empirical link and/or its applicability in the given setting is not fundamentally called into doubt, the initial inference will continue to carry (some) evidentiary weight.

Although factual inferences may reduce the scope of the factual inquiry that is required to come to a full conviction of fact B, the law of evidence sometimes goes further and recognises a lowering of the evidentiary threshold. This is where ‘real’ presumptions (or ‘legal presumptions’) come into play: they direct the Court to conclude, from the presence of fact A (or a set of facts A1, A2, and A3), that fact B (e.g. potential anticompetitive effects) is likewise proven, irrespective of the strength of the empirical link between fact A and fact B. This is justified because presumptions do not (or not primarily) follow an empirical rationale. Rather, they are based on normative grounds: generally speaking, they shall ensure the well-functioning of the law and law enforcement as an administrative system. In some cases, legal presumptions shall elicit information from the party against which the presumption works, because that party has better access to the relevant information.187 Frequently, they are meant to lower the administrative cost of proving a given fact, such as to ensure the manageability of proof, reduce overall error cost, and increase legal certainty. In that case, legal presumptions—like factual inferences—will have some empirical grounding, but the empirical link between fact A and fact B to be presumed may be weaker than what would be required in order to recognise a factual inference. In any case, it will be of the essence that fact A is readily observable by the norm addressee and the law enforcement authority at relatively low cost.

In some cases, legal presumptions are rebuttable, in others irrebuttable. Either way, the undertaking under investigation may challenge the presence of fact A, i.e. of the preconditions of the presumption. In case of a rebuttable presumption, it may also challenge fact B, but the burden of proof is reversed. Only if it fully proves the absence of fact B will the burden of proof shift back to the Commission. However, before the burden of proof is reversed, thought must be given to whether the circ*mstances to be proven are within the sphere of the party bearing the burden of proof, i.e. whether they are readily observable to that party. Typically, this will not be the case for the non-existence of potential anticompetitive effects. Consequently, the burden of proof for potential anticompetitive effects will normally not be reversed. In some instances, such effects may, however, be irrebuttably presumed.

2. Indirect proof of potential effects

The above categorisation allows us to describe more precisely how the preconditions for proving potential anticompetitive effects have shifted: in the past, the ECJ had been ready to accept irrebuttable legal presumptions of anticompetitive effects for some types of conduct (in particular exclusionary rebates and exclusive dealing/exclusivity payments) based on the type of conduct alone and had thereby lowered the evidentiary threshold from proof of a concrete likelihood to an abstract risk of anticompetitive effects. The more recent jurisprudence has rejected these ‘normative’ presumptions, however. Arguably, the former presumptions have been turned into factual inferences—which may, however, depending on the type of conduct, need to be based on a more complex set of facts (e.g. the degree of dominance or the duration and/or market coverage of the practice).

A comparison between Hoffmann-La Roche and the ECJ’s Intel decision of 2017 illustrates this shift: based on a relevant probability—i.e. a general risk—that loyalty rebates and exclusive dealing may anticompetitively foreclose competition, Hoffmann-La Roche established an irrebuttable legal presumption that the adoption of such practices by a dominant undertaking constitutes an abuse.188 The dominant undertaking could plead for an objective justification or an efficiency defence. But it could not rebut the possibility of potential anticompetitive effects. The critique of this presumption as ‘formalistic’ and ‘disconnected from modern economics’189 overlooked its normative basis: the presumption facilitated enforcement, allowed for a high degree of legal certainty, and was arguably based on implicit assumptions on what level of risk EU competition law should accept regarding adverse effects on competition. A comprehensive economic assessment would therefore need to be based on a comparative analysis of error costs.190

In its Intel judgment of 2017, the ECJ has reversed course, however. While claiming to ‘clarify’ the law on loyalty rebates and exclusive dealing,191 it has turned the irrebuttable presumption into a different type of inference—but contrary to the GC’s reading192 not into a rebuttable presumption—but into a factual inference of (concrete) potential anticompetitive effects:193 the burden of proof is not reversed, such that the dominant undertaking would need to proof the implausibility of potential anticompetitive effects—frequently an impossible task. Rather, where the dominant undertaking casts doubt on the potential for anticompetitive effects by presenting substantiated evidence—e.g. an AEC test aiming to show that as-efficient competitors are not foreclosed—the Commission has to revisit whether it can prove potential anticompetitive effects to the full conviction of the Court. Where the factual inference is not refuted, it will continue to carry evidentiary weight. If the dominant undertaking has shown special circ*mstances, the factual inference may be weakened and need to be refined. An AEC test—if conducted in a reliable and error-free manner—does not, as such, disprove the factual inference, but may constitute a factor in need of explanation and must be considered in the overall assessment of the case. Simultaneously, where errors in the AEC test are exposed,194 or where the AEC test was not applicable (see above, III. B. 4. 1.),195 the Commission may continue to fully rely on the factual inference.

With regard to other types of conduct, the Union Courts have recognised factual inferences already some time ago. Predatory pricing and margin squeeze cases may be considered to fall into this category, where an abuse will generally be inferred from an AEC test that shows that as-efficient competitors could not compete.

Whether or not factual inferences apply, recent case law is explicit that ultimately, the full conviction of the Court of the concrete potential of anticompetitive effects is required.

The Commission must verify whether the market conditions on which its theory of harm is based are met or at least likely to arise in the specific context of the case.196 In Unilever Italia, the ECJ, with a view to the showing of anticompetitive effects of exclusivity clauses, demanded an analysis of ‘factors specific to the circ*mstances of the case, such as the extent of that conduct on the market, capacity constraints on suppliers of raw materials, or the fact that the undertaking in a dominant position is, at least, for part of the demand, an inevitable partner’.197 Generally, the ECJ asks for ‘specific, tangible points of analysis and evidence’.198

Whenever the dominant undertaking submits, supported by evidence, that anticompetitive effects are unlikely or implausible in the circ*mstances of a given case, the Commission has to hear the submission and review the evidence. This is part of the Commission’s evidentiary burden—it must fully convince the Court of the likely anticompetitive effects. But it also follows from general procedural principles—in particular the right to be heard of the undertaking under investigation. On the procedural side, the Union Courts have emphasised the Commission’s obligation to examine ‘carefully and impartially all the relevant aspects of the Individual case’.199 So far, the requirement of the ‘relevance’200 of the dominant undertaking’s submission has been interpreted generously. In order to keep the evidentiary proceedings manageable, a rigorous relevance check would be called for: only if the evidence submitted by the dominant undertaking is capable of undermining the full conviction regarding the likelihood of effects should the Commission be required to engage in further inquiry and analysis.201

Regarding the relevance of an AEC test presented by the dominant undertaking,202 the Union Courts’ recent jurisprudence suggests that it will always meet the initial threshold of relevance: a showing that as-efficient competitors retain the ability to compete may cast doubt on the likelihood of anticompetitive effects.203 Nonetheless, even if conducted in a methodologically reliable and error-free manner, such an AEC test is not necessarily determinative: For example, the Commission may demonstrate that, given the setting of the case, an abuse already follows from the foreclosure or marginalisation of less efficient competitors. The AEC test would then be irrelevant for the outcome of the case.

3. Actual effects

An infringement of Art. 102 TFEU does not presuppose proof of actual anticompetitive effects (see above, III. B. 3. 2.).204 However, direct proof of actual effects may sometimes be the best evidence available, in particular where the allegedly abusive conduct has persisted over a longer period of time and where the exclusionary effects are obvious.

Frequently, the causes of any actual effects will be controversial and the attempt to offer additional evidence to resolve such controversy will tend to prolong proceedings, however. The Commission will then rather opt for showing potential effects on the basis of indirect evidence.

It is a separate question whether the absence of actual effects may qualify as indirect proof that a certain line of conduct was incapable of producing anticompetitive effects. As a general matter, the ECJ has dismissed such allegations.205 Yet, sometimes, and if backed up by additional evidence, the absence of actual anticompetitive effects may indicate that anticompetitive effects were unlikely in the circ*mstances of the case.206 If not delimited by clear and transparent criteria, the latter jurisprudence has the potential to significantly prolong proceedings: as a practical matter, the principle that the Commission is not required to inquire into actual effects would be undermined. For that reason, a dominant undertaking that offers proof of a lack of actual anticompetitive effects should be required to also refute or at least weaken the theoretical link between the relevant conduct and potential anticompetitive effects before the Commission’s obligation to further investigate is triggered.

Finally, there are exceptional cases where—contrary to the rule that no actual effects need to be shown—potential effects are difficult to prove to the full conviction of the court without a close examination of actual effects. This may be true in atypical and novel abuse cases: the potential of some practices to adversely affect competition has not yet been sufficiently explored, and relevant theories of harm may be ambiguous in their prediction. For example, potential effects may depend on typical consumer behaviour. In these cases, actual effects may provide the most reliable indication. Atypical bundling/tying is a case in point: when analysing Microsoft’s bundling of the Windows operating system with the Windows Media Player, the Commission considered that, contrary to the classical tying case, PC users could and did to a certain extent download third-party media players through the Internet free of charge. Therefore, the Commission could not assume without further analysis that a tying of the Media Player with the operating system would, by its very nature, be liable to foreclose competition.207 This approach was confirmed in Google Android:208 as it was easy for mobile users to obtain general search or browser apps competing with those that were subject to Google’s tying policy, the Commission established that that policy had the actual effect of making it ‘harder for competing search services to gain search queries and the revenues and data needed to improve their services’, that they increased ‘barriers to entry by shielding Google from competition from other search services’, and that they reduced ‘incentives for the innovations that competitors marketing specialised search services … wished to offer’.209

C. Towards novel analytical shortcuts

In the search for a ‘more workable approach’ to Art. 102 TFEU, a clarification of the evidentiary rules and requirements in future guidelines on exclusionary abuses holds great potential. While the Union Courts’ recent jurisprudence has not shown much openness or sympathy for attempts to re-establish conduct-based legal presumptions with a normative grounding,210 there seems to be ample room for developing factual inferences tailored to different types of abuse (1.). Simultaneously, the Union Courts’ dislike of conduct-based presumptions need not extend to presumptions with a different peg: in particular, presumptions based on the ‘object’ of a given type of conduct or on the identification of an exclusionary strategy have recently attracted increasing interest, partly under the term of ‘naked exclusions’. The consequence of finding an abuse ‘by object’ may be that no effects analysis is required (2.). In other settings, an effects analysis of some kind will be indispensable to establish an infringement of Art. 102 TFEU, but an abbreviated analysis may suffice. This may be the case, in particular, where the anticompetitive effects clearly outweigh any pro-competitive justification (3.). Finally, where the relevant conduct does not fall within the scope of ‘competition on the merits’ for other reasons, the evidentiary threshold for the proof of potential anticompetitive effects may legitimately be lowered to a showing of ‘plausible adverse effects’ (4.). While the analytical shortcuts discussed under 1. and 2. can be categorised as factual inferences and legal presumptions, respectively, the shortcuts developed under 3. and 4. do not qualify as inferences of either kind: in these settings, anticompetitive effects are not inferred, but their proof is otherwise facilitated.

1. Towards a novel generation of factual inferences

Legal presumptions as defined above lower the evidentiary threshold based not only on an empirical link between fact A and fact B, but also, and importantly, on some normative goal they are meant to promote.211 Factual inferences, by contrast, have a purely empirical basis. The evidentiary threshold—in our case: full conviction of the court of potential anticompetitive effects—remains unchanged. But based on a factual inference, i.e. the economically informed experience of the court that the presence of certain facts will typically lead to anticompetitive effects, the requisite scope of the factual inquiry may be narrowed.212 According to Pablo Ibáñez Colomo, these inferences213 should no longer be based on the form of the conduct alone, but include some context-specific indicia214 which would need to be further tailored to the different types of abuse. Given their empirical basis, factual inferences would need to be based on sound economic insights which would, however, need to be translated into manageable legal tests.215 Ideally, the Commission’s future guidelines on exclusionary abuses would propose an initial set of tests. But arguably, the development of manageable factual inferences would be a continuous joint endeavour of competition lawyers and economists.

2. ‘Naked exclusion’—no effects analysis in case of an anticompetitive object

2.1. ‘Naked exclusion’ and ‘abuses by object’

Beyond factual inferences, there may be room for the development of new kinds of presumptions. In particular, the idea of a presumption based on the absence of a pro-competitive rationale is gaining prominence.

A first indication can be found in the 2008 Priorities Guidance: ‘If it appears that the conduct can only raise obstacles to competition and that it creates no efficiencies, its anticompetitive effects may be inferred’.216 According to this wording, the legal treatment of the so-called ‘naked exclusions’ would amount to a quick-look effects analysis: where plausible pro-competitive effects are absent (i.e. no ‘competition on the merits’) and a general potential for anticompetitive effects is obvious, ‘it is not necessary for the Commission to carry out a detailed assessment before concluding that the conduct in question is likely to result in consumer harm’.217 ‘Naked exclusions’ might then amount to just another sub-category of factual inferences.

By contrast, parts of the literature have referred to these settings as ‘abuses by object’, likening them to ‘restrictions by object’ in Art. 101 TFEU.218 However—and irrespective of the debate whether that categorisation may usefully be transplanted to Art. 102 TFEU219—the concept of ‘restrictions by object’ is not a model of clarity even in the context of Art. 101 TFEU: in the past, one might have argued that it established an irrebuttable presumption of illegality for certain types of typically harmful agreements under Art. 101(1) TFEU—although a presumption subject to a possible exemption under Art. 101(3) TFEU. In some more recent judgments, it seems to serve as a justification for a quick-look effects analysis with the aim to determine whether, in a concrete setting, a typically harmful agreement has, or has not, plausible pro-competitive effects.220 Whenever plausible and sufficiently significant pro-competitive effects are established, a full effects analysis will be required.221

The basic analytical framework underlying the latter approach—the finding that a typically harmful practice for which no plausible pro-competitive effects can be demonstrated may be presumed to be capable of producing anticompetitive effects—is not limited to Art. 101 TFEU. It has also been applied in Art. 102 cases.222 The rationale is partially empirical, considering that conduct with an anticompetitive object comes with a high probability of potential anticompetitive effects. However, it is backed up by a normative component, given that it is hard to see why such conduct should be protected. Because of its normative foundation, this shortcut qualifies, in our categorisation, as a legal presumption.

2.2. How to identify naked exclusions

Where the conduct of a dominant undertaking qualifies as a ‘naked exclusion’, the proof of an abuse is significantly facilitated. As no effects analysis is required, the concept has the potential to contribute to a more effective enforcement of Art. 102 TFEU—provided that cases of ‘naked exclusion’ can be identified with ease and a sufficient degree of legal certainty.

Some examples of ‘naked exclusion’ are mentioned in the 2008 Priorities Guidance: no effects analysis shall be required where a dominant undertaking ‘prevents its customers from testing the products of competitors or provides financial incentives to its customers on condition that they do not test such products, or pays a distributor or a customer to delay the introduction of a competitor’s product’.223 A part of the Intel case is illustrative of the latter: according to non-contested findings of the Commission, Intel had paid three computer manufacturers (HP, Acer, Lenovo) for halting or delaying the launch of certain products containing competitors’ x86 CPUs, thereby limiting the sales channels available for these products.224 The Commission characterised this conduct as a ‘naked restriction’ and concluded that, as the commercialisation of certain AMD-based products was cancelled or delayed, customers were deprived of a choice they would otherwise have had.225

However, beyond Intel, the Commission has been cautious in relying on the category of ‘naked restrictions’ so far. For instance, in Baltic Rail, the dominant undertaking dismantled its own railway infrastructure to stop a competitor from offering services based on that infrastructure. Even though this seemed like a textbook example of a ‘naked exclusion’,226 the Commission extensively analysed the effects.227

The above discussion suggests that the ‘no economic sense’ test may be the right criterion to identify other cases of ‘naked exclusions’. The Commission would then need to determine whether a practice holds ‘no economic interest for a dominant undertaking, except that of eliminating competitors’.228 In Super Bock, the ECJ has redefined the scope of application of the concept of ‘restrictions by object’ under Art. 101(1) TFEU, however. According to that judgment, any pro-competitive benefits of an agreement that, prima facie, come with a sufficient degree of harm to competition will preclude its qualification as ‘restriction by object’ only if the pro-competitive benefits are “demonstrated, relevant, intrinsic to the agreement concerned and sufficiently significant ...”.229 If one applies the same principles to the concept of ‘naked exclusion’ under Art. 102 TFEU, unilateral conduct may also qualify as ‘naked exclusion’ if it has some pro-competitive justification, but causes harm to the competitive structure that is clearly disproportionate to its benefits.

It is less clear whether a qualification as a ‘naked exclusion’ is possible and appropriate where substantial pro-competitive effects can be demonstrated, but could also be achieved by a less exclusionary line of conduct. The principle of proportionality suggests that less exclusionary alternatives to an exclusionary practice matter. But if the category of ‘naked exclusion’ is about presuming effects, and thereby ‘radically’ facilitating enforcement, ‘naked exclusions’ may not be the right category for this analysis.

What is clear from the Union Courts’ jurisprudence is that proof of anticompetitive intent as such will not suffice to establish a case of ‘naked exclusion’. According to a long-standing line of case law, the abuse of a dominant position is an ‘objective concept’.230 While a showing of anticompetitive intent is not required,231 evidence of such intent ‘constitutes a factor that may be taken into account in order to determine that a dominant position has been abused’.232 However, it is not sufficient in itself.233 The same is true for a showing of an anticompetitive strategy.234 Judgments in which the Union Courts have found that a relevant type of conduct was capable of harming competition because it was part of a plan to harm competition235 or ‘part of a plan for eliminating a competitor’236 is no proof to the contrary. At a closer look, this case law237 collapses with a ‘no economic sense’ inquiry: neither the Commission nor the Courts have relied solely on direct (subjective) evidence of intent. Rather, they have inquired into the economic rationality of the conduct in a given market context.238

In some settings, the finding of a ‘naked exclusion’ under EU competition law may be strongly influenced by the interdependence between EU competition law and the internal market goal. In particular, a dominant firm’s conduct impeding parallel exports of goods from low-price to high-price countries in the internal market may qualify as abuse with no further effects analysis being required.239 In this case, it is the conflict of the conduct with a fundamental Treaty goal that justifies the finding of a prima facie case of illegality.

2.3. Curtailing consumer choice as a case of a form of ‘naked exclusion’?

In some Art. 102 cases, not least in the context of tying, the notion that a dominant undertaking reduced consumer choice without producing countervailing benefits for consumers has figured prominently.240 If the reduction of choice is guided by no economic rationale other than restricting competition, such conduct may amount to a ‘naked exclusion’. No further effects analysis would then be required.241 In fact, such conduct interferes with one of the most basic operating principles of performance-based competition, namely the freedom and ability of the demand side to choose between different products or services according to their individual preferences.

Nonetheless, the Commission frequently performs a full effects analysis in these cases.242 An effects analysis will be indispensable indeed where the dominant undertaking comes forward with a plausible pro-competitive justification, e.g. related to significant benefit for the demand side. In such a case, the concept of ‘naked exclusion’ is no longer applicable. However, a suggestion that tying practices in consumer-facing markets will maximise the users’ convenience of use may not suffice where the same convenience can be achieved by allowing undertakings that offer competing products or services to access a relevant API.

3. A ‘quick look’ effects balancing in the case of disproportionally harmful conduct

Where the Commission can establish a ‘naked exclusion’, no effects analysis is required. As has been argued above, the category of ‘naked exclusion’ extends to cases where prima facie harm to competition is clearly out of proportion to some relatively minor pro-competitive benefits that may flow from the exclusionary practice. To ensure that the category of ‘naked exclusion’ retains its benefit to radically simplify and speed up the enforcement of Art. 102 TFEU, it must be limited to those cases where the disproportionality is immediately obvious.

In other cases, the potential harm to competition of a given practice may still be disproportionally large compared with its alleged pro-competitive benefits, but a somewhat more comprehensive analysis may be required to establish the disproportionality. Still, the disproportionality may follow from a ‘quick look’. In particular, there may be some settings where some pro-competitive benefits are plausible, but the same pro-competitive benefits can, as a general matter and based on sufficient experience in the market, be realised by a less anticompetitive line of conduct.243 For example, some efficiencies related to conditional rebates can likewise be realised by way of quantity discounts.

The principle of proportionality is deeply engrained in the structure of Art. 102 TFEU. This is not only true when it comes to the defences that a dominant undertaking can raise: in order to prevail with an efficiency defence, the dominant undertaking’s conduct must be indispensable and proportionate to the goal pursued.244 Likewise, the notion that dominant undertakings are subject to a ‘special responsibility’, i.e. that they must ‘not […] allow their conduct to impair genuine, undistorted competition on the internal market’,245 flows from the principle of proportionality. What is more, the analytical templates that the ECJ has developed for specific types of conduct—like the test for tying, for refusal to give access to an essential facility and even for predatory pricing—are informed by that principle. Whenever the Commission can show, upon a quick look and based on established economic wisdom,246 that the potentially anticompetitive conduct is not necessary to realise the alleged efficiencies, a general showing of plausible anticompetitive effects should suffice, with no concrete in-depth inquiry being required.

Similarly, where, upon a quick look, the potential anticompetitive harm is much greater, or of much graver quality, than the expected efficiencies, no in-depth proof of the scope and size of the anticompetitive effects should be required. Rather, a rough demonstration of the disproportionality should suffice.

4. No competition on the merits—lowering of the evidentiary threshold for anticompetitive effects

Finally, evidentiary shortcuts may be justified where the allegedly anticompetitive conduct lies outside the scope of ‘competition on the merits’. The analysis of what is, and what is not, ‘competition on the merits’ is sometimes deeply intertwined with an effects analysis. The application of the AEC test is the most important example (see above, III. B. 4. 1.). These are not the cases discussed here.

Conduct may be found to fall outside the scope of ‘competition on the merits’ irrespective of its effects mainly in two types of settings: first, where competitive advantages of a dominant undertaking do not result from superior performance, but from some sort of state privilege (4.1.); secondly, in case of conduct that is illegal on non-competition grounds—for example, because it violates the GDPR (4.2.). In both cases, the conduct, in and of itself, does not indicate anticompetitive effects: given the goal of Art. 102 TFEU to protect competition, potential anticompetitive effects need to be established before an abuse of dominance can be found.247 However, an error-cost-oriented framework suggests that, with a view to such conduct, the evidentiary threshold for anticompetitive effects can be lowered: prohibiting conduct that is clearly outside the scope of ‘competition on the merits’ does not come with a risk of ‘false positives’ or a risk to undermine desirable incentives to compete. Consequently, a mere plausibility of anticompetitive effects may suffice for establishing an infringement. The lowering of the evidentiary threshold from ‘concrete likelihood’ to a more abstract ‘plausibility’ may facilitate the enforcement of Art. 102 TFEU.

4.1. Competitive advantages that result from a statutory monopoly or exclusive rights

There are some indications that the Union Courts are willing to lower the evidentiary threshold for anticompetitive effects, at least in those cases where competitive advantages follow not from an undertaking’s investments and performance, but rather from privileges conferred upon it. The ECJ’s judgment in Servizio Elettrico Nazionale is an example in point: in anticipation of the full liberalisation of the Italian electricity market for households, and with the goal to prevent a large-scale departure of customers to competing suppliers, SEN—a subsidiary of the former state monopolist ENEL—had obtained the consent of customers in the still regulated part of the market to receive commercial offers for unregulated electricity supply from companies of the ENEL group—but not from ENEL’s competitors. The ECJ found, first, that an undertaking that holds a statutory monopoly and uses resources that result from these exclusive rights and are therefore inaccessible to a hypothetical as-efficient competitor for the purpose of extending the dominant position in the protected market to another market is not competing ‘on the merits’.248 Secondly, the existence of a bias in collecting customers’ consent would establish an irrebuttable presumption that any difference in the quantity of customer data transferred would not result from ‘competition on the merits’.249 Thirdly, given that access to such customer data likely conferred a comparative competitive advantage, the ECJ concluded that the subsequent use of such data should be considered to be capable of producing an exclusionary effect on the free market.250 Consequently, if it were shown that the ENEL group, in seeking customers’ consent to receive offers for energy supply in the free market, discriminated against competitors, ‘that fact alone would be sufficient to show that the conduct [of the dominant undertaking] at the very least was capable of impairing effective, undistorted competition’.251 In short, a discriminatory use of competitively relevant resources resulting from a statutory monopoly was irrebuttably presumed to have potential anticompetitive effects.

4.2. Conduct that is illegal on non-competition grounds

There is somewhat conflicting case law on whether the illegality of a relevant line of conduct under non-competition norms should be considered when determining whether conduct is, or is not, ‘competition on the merits’ within the meaning of Art. 102 TFEU, and whether it leads to a lowering of the threshold for proving anticompetitive effects. Clearly, the legality of a practice under non-competition laws does not pre-empt a finding that such conduct constitutes an abuse under Art. 102 TFEU.252 In Servizio Elettrico Nazionale, the ECJ opted for a broader wording, thereby insinuating that the illegality of a practice under non-competition laws is likewise irrelevant for the finding of an abuse: according to Servizio Elettrico Nazionale, ‘[t]he illegality of abusive conduct under [Art. 102 TFEU] is unrelated to the characterisation of that conduct in other areas of law’.253

That statement stands in some tension to the ECJ’s preliminary ruling in Meta Platforms, however. According to that judgment, ‘the compliance or non-compliance of [the] conduct with the provisions of the GDPR may, depending on the circ*mstances, be a vital clue among the relevant circ*mstances of the case in order to establish whether that conduct entails resorting to methods governing normal competition and to assess the consequences of a certain practice in the market or for consumers’.254 Consequently, in examining an abuse, it might be necessary for a competition authority to inquire ‘whether that undertaking’s conduct complies with rules other than those relating to competition law’.255 Indeed, the relevance of the compliance or non-compliance with the GDPR for establishing whether the relevant conduct constituted ‘competition on the merits’ seems intuitive. However, it is not, as such, indicative of whether that conduct potentially harms the competitive structure. Therefore, the ECJ’s finding that the compliance or non-compliance with the GDPR may also be a ‘vital clue’ when it comes to the assessment of the ‘consequences of a certain practice in the market or for consumers’ comes as a surprise: There is neither an empirical nor a conclusive normative link between an infringement of the GDPR and anticompetitive harm.

The German Federal Court of Justice (BGH), in its Facebook decision,256 has therefore adopted a different line of argument: under the German prohibition of abuse of dominance (§ 19(1) GWB), an abbreviated analysis of anticompetitive effects may suffice in cases where a dominant undertaking adopts unfair contractual conditions (in Facebook: a clause requiring consent to the merging of personal user data obtained from different sources) and harm to the other party to the contract is clearly established. More specifically, a showing of a plausible, abstract risk to adversely affect competition will then be enough.257

As of now, it is not yet clear whether the ECJ would be willing to adopt a similar approach. However, where the relevant conduct by a dominant undertaking is illegal under non-competition rules, the error costs of prohibiting it (also) under competition law are low. As such, this would not justify activating the competition law enforcement machinery. A link to the goals of EU competition law, and hence some potential to adversely affect competition, is an indispensable precondition. But for the sake of a more effective enforcement, it may well be justified to drop the requirement to prove a concrete risk to competition and to let a plausible abstract risk suffice.

V. Tentative conclusions

The Commission’s initiative for drafting guidelines on exclusionary abuses comes at the right moment in time. In the Commission’s enforcement practice, the belief that the 2008 Priorities Guidance, with its commitment to intervene only based on concrete proof of consumer harm, has quickly given way to the recognition that the quality of competition law depends not only on the economic accuracy of the rules, but also on the administrability of the enforcement regime, including the provability of the relevant facts at reasonable cost and within a reasonable timeframe. As Joskow once observed: ‘The test of a good legal rule is not primarily whether it leads to the correct decision in a particular case, but whether it does a good job deterring anticompetitive behaviour throughout the economy given all the relevant costs, benefits, and uncertainties associated with diagnosis and remedies’.258

Consequently, the focus of the new initiative now is on establishing a ‘workable effects-based approach’.259 In order to ensure Art. 102 TFEU’s effectiveness, including its deterrent effect and the Commission’s ability to impose remedies that undo the competitive harm, the duration of Art. 102 proceedings must be significantly reduced.

While one may think of several different causes for the increase of the median duration of Art. 102 proceedings over time, a high degree of uncertainty surrounding the substantive law on abuse and tightened requirements regarding the proof of anticompetitive effects are arguably among them. Simultaneously, the evidentiary rules regarding the proof of effects are particularly uncertain. In such a setting, a risk-averse Commission tends to invest an excessive amount of time and resources to meet the highest possible standard in order not to lose a case that they have determined to be of special importance.

Guidelines on exclusionary abuses can contribute to deducing, from a partly inconsistent case law of the Union Courts, a (more) coherent legal framework. They can thereby help to clarify the substantive law on abuse of dominance—albeit within the interpretative framework defined by the Union Courts. For example, they can explain the concept of ‘competition on the merits’ and its relation to the AEC principle; and they can specify what is required, under Art. 102 TFEU, for showing ‘potential anticompetitive effects’. Although the Union Courts have explicitly refused to require a specific degree of risk, they have repeatedly found that an abstract risk of anticompetitive effects will not suffice, but that a concrete risk to the competitive structure must be established. In fact, this appears to be the core feature of the Union Courts’ ‘effects-based approach’ to Art. 102 TFEU.

Arguably, an equally important contribution of future guidelines to a more effective enforcement of Art. 102 TFEU may lie in developing a clear and coherent framework with regard to the evidentiary requirements for proving anticompetitive effects—requirements that are currently utterly confusing. This is true in particular with regard to the types of inferences and other analytical shortcuts on which the Commission can rely, and with regard to the preconditions on which these inferences or shortcuts can be based.

From the ECJ’s 2017 Intel judgment onwards, the Union Courts have rejected the use of conduct-based presumptions of anticompetitive effects. But future guidelines on exclusionary abuses could set out the preconditions for when ‘factual inferences’ are justified, i.e. when potential anticompetitive effects can be inferred, on a sound empirical basis, from a limited number of easily observable facts. Simultaneously, presumptions with a partly normative, partly empirical basis have not been banned from the law on abuse altogether: potential anticompetitive effects may be presumed in case of ‘by object’ exclusions or so-called ‘naked exclusions’—a concept that should be further developed in the guidelines. Arguably, the category of ‘naked exclusion’ extends to cases where the pro-competitive justification is obviously insignificant compared with the anticompetitive potential.

In other cases, an abbreviated effects analysis may suffice. This may be the case where the pro-competitive effects of a practice may likewise be achieved by less exclusionary alternatives, or where conduct is at issue that lies outside the scope of ‘competition on the merits’. In the latter case, the showing of an abstract risk of anticompetitive effects may suffice.

Ultimately, all this may not be enough. More may be needed to arrive at an effective enforcement of Art. 102 TFEU. The Commission’s initiative to re-evaluate Regulation 1/2003 has already been mentioned. As an increasing number of Member States is about to introduce some sort of ‘market investigation’ into their competition law toolbox,260 the debate about introducing a ‘New Competition Tool’261 that would empower the Commission to intervene independently of the finding of an abuse may see a revival also at the EU level. Possibly, a legal framework for Art. 102 proceedings without fines is required, which would allow for an enforcement under a lower standard of proof and with somewhat different evidentiary requirements. Also, a better and clearer framework for more effective remedies is called for.

Nonetheless, future guidelines that reduce the uncertainties surrounding the substantive interpretation of exclusionary abuses and establish a coherent and administrable evidentiary framework are a key element of any more comprehensive reform of Art. 102 TFEU. Compared with the alternatives—more sector-specific regulation à la DMA or a legal specification of the conditions for the existence of an abuse in a legislative act by the Council, based on Art. 103 TFEU262—guidelines on exclusionary abuses are the most attractive and realistic option to contribute to a regime within which Art. 102 TFEU can gain new vitality in protecting competition.

1

Communication from the Commission: Amendments to the Communication from the Commission—Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [2023] OJ C116/1 (‘Amending Communication’).

2

Communication from the Commission: Guidance on the Commission’s enforcement priorities in applying Article 82 of the EC Treaty to abusive exclusionary conduct by dominant undertakings [2009] OJ C 45/7 (‘2008 Priorities Guidance’).

4

Linsey McCallum and others, ‘A dynamic and workable effects-based approach to abuse of dominance’, European Commission, Competition policy brief No 1/2023 (‘Policy Brief’).

5

We exclude negative clearance decisions due to their special legal content, which hinders an effective comparison with other decisions.

6

Council Regulation (EC) No 1/2003 of 6 December 2002 on the implementation of the rules of competition laid down in Articles 81 and 82 of the Treaty [2003] OJ L1/1.

7

For an instructive overview of the commitment decision practice see Niamh Dunne, ‘Commitment Decisions in EU Competition Law’ (2014) 10 Journal of Competition Law & Economics 399, 406. The main categories of commitment decisions in Art. 102 cases are decisions in the energy sector, in technological markets and in financial markets.

8

Regulation 17/62 (First Regulation implementing Articles 85 and 86 of the Treaty [1962] OJ 13/204) did not provide for a comparable legal framework for cooperative enforcement.

9

On the importance of considering deterrence effects when choosing optimal legal standards see, for example, Yannis Katsoulacos and David Ulph, ‘On Optimal Legal Standards for Competition Policy: A General Welfare-Based Analysis’ (2009) 57 Journal of Industrial Economics 410; Jan Broulík, ‘Preventing Anticompetitive Conduct Directly and Indirectly: Accuracy Versus Predictability’ (2019) 64 Antitrust Bulletin 115, 118.

10

Given the sometimes small numbers of decisions in these respective periods, the findings have to be interpreted carefully and any conclusion can only be tentative.

11

For reasons of practicality, we count both the month in which the starting point occurs and the month in which the ending point (i.e. decision) occurs.

12

This methodology is certainly not flawless. For example, it can lead to certain distortions in comparing ex officio investigations with proceedings initiated by a complaint: only for the latter will we be able to observe the actual point in time at which the Commission first gained relevant information about the alleged infringement. By contrast, any informal procedural step by the Commission depends on the Commission’s discretion, as well. Consequently, ex officio proceedings may still seem shorter in our dataset than they actually are. However, with a view to illustrating the problem of excessively long proceedings, we consider that our methodology of measuring their duration is ‘good enough’.

13

This is the only period in which this differentiation can be made in a useful way, as can be seen in Figure 1.

14

Absent a more in-depth analysis of the reasons for this small margin, no policy conclusions can be drawn from this finding, at least not across all decisions. Of course, Figure 3 appears to confirm that cooperative enforcement does lead to shorter proceedings than adversarial enforcement. This finding is, however, subject to the above-stated incapability of our methodology to measure the duration of a proceeding reliably given that, before the formal initiation of proceedings, it lies within the discretion of the Commission whether it will mention, in a later decision, any prior informal steps, see above n 12. Many commitment decisions are silent about these informal procedural steps.

15

Cooperative proceedings lasted for a median 56months in 2004 to 2007 and from then on were considerably shorter (33months (2008–2011), 47months (2012–2015), and 35months (2016–2019), respectively).

16

A selection bias may result from the differentiation between cooperative and adversarial proceedings, if the more complex cases were typically be dealt with in an adversarial setting. Nonetheless, the trend remains problematic.

17

Van den Bergh Foods Limited (Cases IV/34.073, IV/34.395 and IV/35.436) Commission Decision of 11 March 1998; Virgin/British Airways (Case IV/D-2/34.780) Commission Decision of 14 July 1999; Deutsche Post AG (Case COMP/35.141) Commission Decision of 20 March 2001; DSD (Case COMP D3/34493) Commission Decision of 20 April 2001; AstraZeneca (Case COMP/A.37.507/F3) Commission Decision of 15 June 2005 and Coca-Cola (Case COMP/A.39.116/B2) Commission Decision of 22 June 2005.

18

Intel (Case COMP/C-3/37.990) Commission Decision of 13 May 2009; RAMBUS (Case COMP/38.636) Commission Decision of 9 December 2009; Rio Tinto Alcan (Case COMP/39.230) Commission Decision of 20 December 2012; Slovak Telekom (Case AT.39523) Commission Decision of 15 October 2014; ARA Foreclosure (Case AT.39759) Commission Decision of 20 September 2016; Google Search (Shopping) (Case AT.39740) Commission Decision of 27 June 2017; Baltic Rail (Case AT.39813) Commission Decision of 2 October 2017; BEH Gas (Case AT.39849) Commission Decision of 17 December 2018; Google Search (AdSense) (Case AT.40411) Commission Decision of 20 March 2019, and Qualcomm (predation) (Case AT.39711) Commission Decision of 18 July 2019.

19

See, for example, Johannes Laitenberger and James Kröger, ‘Towards an “even more efficient” approach?’ [2023] Zeitschrift für Europäisches Privatrecht 621, 628 ff.

20

Procedural (law) reasons for long proceedings may include priority setting by the Commission (e.g. in Qualcomm (predation) (n 18)), high procedural standards (frequently related to protecting the defendant’s rights to be heard and to access to the file) or overlong and ultimately fruitless commitment negotiations (e.g. in Google Search (Shopping) (n 18), and Google Search (AdSense) (n 18)). Procedural instruments that might shorten proceedings—interim measures in particular—are rarely used; for their particular potential, see Massimiliano Kadar, ‘The Use of Interim Measures and Commitments in the European Commission’s Broadcom Case’ (2021) 12 Journal of European Competition Law & Practice 443, 451.

21

See on this, Massimiliano Kadar, ‘Evaluating 20Years of Regulation 1/2003: Are EU Antitrust Procedures “Fit for the Digital Age”?’ (2024) 85 Antitrust Law Journal 577; Laitenberger and Kröger (n 19); and on the experience with Regulation 1/2003, Wouter PJ Wils, ‘Regulation 1/2003: An Assessment After Twenty Years’ (2023) 46 World Competition 3. The experience under national procedural law suggests that the enforcement of Art. 102 TFEU might be substantially facilitated if a special procedure were introduced for enforcing the abuse prohibition without fines.

22

Google Search (Shopping) (n 18); Google Search (AdSense) (n 18).

23

Intel (n 18); RAMBUS (n 18); Qualcomm (predation) (n 18).

24

Slovak Telekom (n 18); Baltic Rail (n 18); BEH Gas (n 18).

25

2008 Priorities Guidance, para 20.

26

Tristan Rohner, Art. 102 AEUV und die Rolle der Ökonomie (Nomos 2023) 220. For example, in Intel, after 151 pages of AEC test, the Commission finds a reduction of consumer choice on not more than six pages—see Commission Decision in Intel (n 18), pp 459–464. The argument is summarised in recital 1603: Intel’s conduct ‘excluded, limited, or delayed AMD x86 CPUs in the market. As such, Intel’s exclusionary practices had a direct and immediate negative impact on those customers who would have had a wider price and quality choice if they had also been offered the product of their favourite OEM and/or retailer with x86 CPUs from Intel’s competitors.’

27

In the 1990s, the room in the Commission decisions dedicated to the effects analysis was quite limited. Usually, the Commission showed the anticompetitive effect purely qualitatively, cf. e.g. British Midland v. Aer Lingus (Case IV/33.544) Commission Decision of 26 February 1992, recitals 26 ff; Warner-Lambert/Gillette and Others and BIC/Gillette and Others (Cases IV/33.440 and IV/33.486) Commission Decision of 10 November 1992, recitals 25 ff; Irish Sugar plc (Cases IV/34.621 and 35.059/F-3) Commission Decision of 14 May 1997, recitals 114, 125, 135; Amministrazione Autonoma dei Monopoli di Stato (Case IV/36.010-F3) Commission Decision of 17 June 1998, recitals 35, 37; Ilmailulaitos/Luftfartsverket (Case IV/35.767) Commission Decision of 10 February 1999, recitals 42 ff.

28

For example, in the Telekomunikacja Polska case, the Commission based its quantitative analysis on data that it had received from the dominant undertaking and the Polish telecommunications authority (UKE). The analysis of the conduct’s effects on the competitive structure apparently remained manageable—Cf. Telekomunikacja Polska (Case COMP/39.525) Commission Decision of 22 June 2011, recitals 820 ff.

29

See on this complex rendition of the AEC test below, III. B. 4. 1.

30

For example, in Baltic Rail, the Commission’s effects analysis spans 30 pages (almost a third of the entire length of the decision)—see Baltic Rail (n 18), pp 65–94. See on the effects analysis in this case also below, IV. C. 2. 2.

31

For the Union Courts’ conclusive competence to interpret EU law see Art. 19 TEU, 267 TFEU. Furthermore, for the non-binding nature of Commission guidelines see Case C-376/20 P Commission v CK Telecoms UK Investments, EU:C:2023:561, para 125. See also the European Parliament’s Report on institutional and legal implications of the use of ‘soft law’ instruments (<https://www.europarl.europa.eu/doceo/document/A-6-2007-0259_EN.html> accessed 31 March 2024), cautioning against implicit legislation by means of, inter alia, interpretative communications and an extrapolation of the case law of the Court of Justice ‘into unchartered territory’.

32

Which degree of discretion the Commission possesses in drafting guidelines remains controversial. In favour of a broad degree of discretion—even of the admissibility of deviations from the case law: Pınar Akman, ‘The European Commission’s Guidance on Art. 102 TFEU: From Inferno to Paradiso?’ (2010) 73 Modern Law Review 605, 626f.; Rohner (n 26) 274. For a more narrow construction: Håkon A Cosma and Richard Whish, ‘Soft Law in the Field of EU Competition Policy’ (2003) 14 European Business Law Review 25, 52; Liza Lovdahl Gormsen, A Principled Approach to Abuse of Dominance in European Competition Law (CUP 2010) 158; Liza Lovdahl Gormsen, ‘Why the European Commission’s Enforcement Priorities on Article 82 EC should be Withdrawn’ (2010) 31 European Competition Law Review 45, 50; Oana Andreea Ştefan, ‘European Competition Soft Law in European Courts: A Matter of Hard Principles?’ (2008) 14 European Law Journal 753, 764; see also Opinion of AG Kokott in Case C-95/04 P British Airways v Commission, EU:C:2006:133, para 28.

33

See Massimiliano Kadar, Johannes Holzwarth and Virgilio Pereira, ‘Abuse of dominance under Article 102 TFEU: a survey on 2023’ (2024) 15 Journal of European Competition Law & Practice (published online 6 May 2024), p 2, pointing to increasingly concentrated sectors and the role of digital markets for the EU economy, and suggesting that these changes have made false negatives in the enforcement of Art. 102 TFEU ‘both more likely and more costly’.

34

For this, see Case C-234/89 Delimitis, EU:C:1991:91, para 44.

35

For a summary see Anne C Witt, The More Economic Approach to EU Antitrust Law (Hart 2016) ch 4 with further references to the debate. See also the contributions in Daniel Zimmer (ed) The Goals of Competition Law (Edward Elgar 2012).

36

See, inter alia, Okeoghene Odudu, ‘The wider Concerns of Competition law’ (2010) 30 Oxford Journal of Legal Studies 559, 605–612; Pınar Akman, The Concept of Abuse in EU Competition Law: Law and Economic Approaches (Hart 2012) 49–106.

37

See, inter alia, Adrian Künzler, Effizienz oder Wettbewerbsfreiheit? (Mohr Siebeck 2008) 323 ff. See also: Heike Schweitzer, ‘Efficiency, political freedom and the freedom to compete’, in Daniel Zimmer (ed) The Goals of Competition Law (Edward Elgar 2012) 169 ff.

38

See Art. 3 para 1 lit b TFEU and Protocol No 27 on the internal market and competition, according to which the internal market includes a system ensuring that competition is not distorted. See also, recently, Case C-377/20 Servizio Elettrico Nazionale and Others, EU:C:2022:379, paras 42–43.

39

2008 Priorities Guidance, para 20.

40

2008 Priorities Guidance, para 19.

41

Servizio Elettrico Nazionale (n 38), para 46. Arguably, the term ‘well-being of consumers’ is used synonymously with the economic concept of consumer welfare in this passage. This is also suggested by the fact that the aim of protecting ‘consumer well-being’ is to justify the ‘efficiency defence’.

42

Servizio Elettrico Nazionale (n 38), para 44 (and 47) with further references to prior case law. See also: Fernando Castillo de la Torre, ‘Is the Effects-Based Approach Too Cumbersome?’, in Adina Claici, Assimakis Komninos and Denis Waelbroeck (eds) The Transformation of EU Competition Law (Wolters Kluwer 2023) 146.

43

The concept of consumer welfare does not refer to, and must not be measured in terms of, price only, but also encompasses choice, quality, and innovation—see Servizio Elettrico Nazionale (n 38), para 46; C-307/18 Generics (UK) and Others, EU:C:2020:52, para 165.

44

Richard A Posner, Antitrust Law (2nd edn, University of Chicago Press 2001) 29.

45

Against this background, the broader wording in the Amending Communication, para 1 appears appropriate: the competition rules contribute to ‘a level playing field where markets serve consumers’.

46

See Joined Cases C-468/06 to C-478/06 Sot. Lélos kai Sia and Others, EU:C:2008:504.

47

Even Servizio Elettrico Nazionale (n 38) explicitly acknowledges the relevance of this goal in paras 42–43.

48

See e.g. Van Bael & Bellis on Competition Law, March 2023, p 6.

49

Amending Communication, para 1.

50

Case T-604/18 Google Android, EU:T:2022:541, para 1028.

51

For references see fn 3 of the Policy Brief.

52

Massimiliano Kadar and Johannes Holzwarth, ‘Effects-based approach? Effects-based approach! The European Commission’s “Article 102 Package”’ [2023] Neue Zeitschrift für Kartellrecht 333, 334. Similarly: Kadar, Holzwarth and Pereira (n 33), p 3.

53

See e.g. Lina M Khan, ‘Amazon’s Antitrust Paradox’ (2017) 126 Yale Law Journal 710, 737 ff; Lina Khan, ‘The New Brandeis Movement: America’s Antimonopoly Debate’ (2018) 9 Journal of European Competition Law & Practice 131; Tim Wu, ‘After Consumer Welfare, Now What?’ [2018] CPI Antitrust Chronicle, April 2018.

54

Servizio Elettrico Nazionale (n 38), para 50; Case C-680/20 Unilever Italia Mkt. Operations, EU:C:2023:33, paras 41–42; Case C-333/21 European Superleague Company, EU:C:2023:1011, para 130.

55

Case 85/76 Hoffmann-La Roche v Commission, EU:C:1979:36, para 91.

56

Case C-95/04 P British Airways v Commission, EU:C:2007:166, para 86.

57

Servizio Elettrico Nazionale (n 38), para 50; Unilever Italia (n 54), paras 41–42; European Superleague (n 54), para 130.

58

European Superleague (n 54), para 130.

59

Unilever Italia (n 54), para 42.

60

European Superleague (n 54), para 130.

61

Servizio Elettrico Nazionale (n 38), para 50; Case C-413/14 P Intel v Commission, EU:C:2017:632, paras 138 and 140.

62

Most commentators accept that there are some ‘by object’ abuses (in analogy to Art. 101(1) TFEU) which allow the Commission or competition authority to infer effects; but they argue that this ‘box’ should be narrowly construed—see, for example, Pablo Ibáñez Colomo, ‘Competition on the merits’ (2024) 61 Common Market Law Review 387, 410–414.

63

Justin Lindeboom speaks of an ‘all things considered’-approach, see ‘Formalism in Competition Law’ (2022) 18 Journal of Competition Law & Economics 832, 858 ff.

64

Generally on this point: Arndt Christiansen and Wolfgang Kerber, ‘Competition policy with optimally differentiated rules instead of per se rules vs. rule of reason’ (2006) 2 Journal of Competition Law & Economics 215. See also: Pablo Ibáñez Colomo, ‘Structured Legal Tests, Effective Judicial Review and Missing the Trees for the Forest’ (2022) 13 Journal of European Competition Law & Practice 461–462.

65

On the terminology see below, IV. B. 1.

66

For a similar argument, albeit in a different legal context, see Justice Breyer, Dissenting Opinion, Leegin Creative Leather Products, Inc. v. PSK, Inc., 551U.S. 877 (2007).

67

Justice Breyer, Dissenting Opinion, Leegin Creative Leather Products, Inc. v. PSK, Inc., 551U.S. 877 (2007).

68

Servizio Elettrico Nazionale (n 38), para 50.

69

Hoffmann-La Roche (n 55), para 91. See also Case C-209/10 Post Danmark, EU:C:2012:172, para 24; Servizio Elettrico Nazionale (n 38), para 44; European Superleague (n 54), para 125.

70

See below, III. B. 4.

71

See Kadar, Holzwarth and Pereira (n 33), p 3.

72

Emphasised, for example, by Ibáñez Colomo (n 62) 415.

73

The term makes it clear that this precondition requires both the finding of restrictive effects and a determination that these do not result from pro-competitive conduct. Consequently, some argue that the ‘no competition on the merits’ criterion should not be assessed separately—see Opinion of AG Rantos in Case C-377/20 Servizio Elettrico Nazionale and Others, EU:C:2021:998, para 48; Damien J Neven, ‘The As-Efficient Competitor Test and Principle. What Role in the Proposed Guidelines?’ (2023) 14 Journal of European Competition Law & Practice 565, 569: ‘… the absence of normal competition or competition on the merits are just different paraphrases for the presence of anti-competitive effects’.

74

Hoffmann-La Roche (n 55), para 91. See also Servizio Elettrico Nazionale (n 38), para 44 with further references.

75

Servizio Elettrico Nazionale (n 38), para 68; Unilever Italia (n 54), para 36. See also the Amending Communication, Annex para 1. For a critical view regarding the criterion of negative effects ‘on the competitive structure’ see Neven (n 73) 566 and 568–569: The term would be ill-defined and the concept could be used as ‘a shortcut to avoid the analysis of anticompetitive effects’.

76

European Superleague (n 54), para 131.

77

Post Danmark (n 69), para 26; referring to Case C-280/08 P Deutsche Telekom v Commission, EU:C:2010:603, para 175. See also: Case T-604/18 Google Android (n 50), para 281.

78

See, for example, Case T-604/18 Google Android (n 50), para 294.

79

For this see also Policy Brief, p 4.

80

Case T-604/18 Google Android (n 50), para 281.

81

Erroneously suggesting that commitment to maintaining an ‘effective competition structure’ is analogous to protecting a given ‘market structure’: Pınar Akman, ‘A Critical Inquiry into “Abuse” in EU Competition Law’, (2024) 44 Oxford Journal of Legal Studies (published online 7 March 2024), p 26.

82

See e.g. Raffaele Di Giovanni Bezzi, ‘A Tale of Two Cities: Effects Analysis in Article 102 TFEU Between Competition Process and Market Outcome’ (2023) 14 Journal of Competition Law & Practice 83; Renato Nazzini, The Foundations of European Union Competition Law (OUP 2011) 14 ff. In its decisions, the Commission sometimes explicitly uses the term ‘competitive process’ in its effects analysis, see e.g. Google Android (Case AT.40099) Commission Decision of 18 July 2018, recitals 863, 971, 1142. See also 2008 Priorities Guidance, para 6: ‘The emphasis of the Commission’s enforcement activity in relation to exclusionary conduct is on safeguarding the competitive process in the internal market.’ Simultaneously, the 2008 Priorities Guidance distinguished between ‘protecting an effective competitive process’ and ‘protecting competitors’.

83

See Case C-52/09 TeliaSonera Sverige, EU:C:2011:83, para 64; Case T-336/07 Telefónica and Telefónica de España v Commission, EU:T:2012:172, para 90; Case C-457/10 P AstraZeneca v Commission, EU:C:2012:770, para 112; Case C-23/14 Post Danmark II, EU:C:2015:651. Also: Castillo de la Torre (n 42) 152.

84

See Unilever Italia (n 54), para 41. In terms of the German legal tradition, Art. 102 TFEU qualifies as a ‘Gefährdungsdelikt’—see also Opinion of AG Kokott in Case C-8/08T-Mobile Netherlands and Others, EU:C:2009:110, para 47.

85

See Case C-413/06 P Bertelsmann and Sony Corporation of America v Impala, EU:C:2008:392, para 52: in order to prohibit a merger, a significant impediment to competition must be more likely than not to occur. See also: CK Telecoms (n 31), para 84.

86

Case T-612/17 Google and Alphabet v Commission (Google Shopping), EU:T:2021:763, para 438; TeliaSonera (n 83), para 77; Post Danmark II (n 83), para 66.

87

Post Danmark II (n 83), para 74.

88

Post Danmark (n 69), para 44; Post Danmark II (n 83), para 67.

89

Case T-201/04 Microsoft v Commission, EU:T:2007:289, para 867; Case C-165/19 P Slovak Telekom v Commission, EU:C:2021:239, para 109; Case T-814/17 Lietuvos geležinkeliai v Commission, EU:T:2020:545, para 80; Telefónica (n 83), para 268; Post Danmark II (n 83), paras 31, 68.

90

Lietuvos geležinkeliai (n 89), para 80; Telefónica (n 83), para 268.

91

Policy Brief, p 3 (with fn 22): ‘Despite the varied terminology, the applicable legal standard endorsed by the Union Courts and applied by the Commission must be understood as being one and the same’.

92

Opinion of AG Kokott in Case C-23/14 Post Danmark II, EU:C:2015:343, para 82. AG Kokott further explains this standard in paras 83–84: A higher bar than ‘more likely than not’ (e.g. a requirement that exclusionary effects must be ‘very likely’ or ‘particularly likely’) would be inappropriate, due to the special responsibility of the dominant undertaking not to undermine effective and undistorted competition in the internal market. ‘That responsibility entails some obligation to exercise restraint on the market. The dominant undertaking must therefore refrain from all commercial practices which are likely to produce an exclusionary effect’. A ‘more likely than not’ standard is also advocated in the literature—see, for example, Pablo Ibáñez Colomo, ‘The (Second) Modernisation of Article 102 TFEU: Reconciling Effective Enforcement, Legal Certainty and Meaningful Judicial Review’ (2023) 14 Journal of European Competition Law & Practice 608, 618.

93

Post Danmark II (n 83), para 67.

94

See Elias Deutscher, ‘Causation and counterfactual analysis in abuse of dominance cases—lessons from the General Court’s Qualcomm ruling’ (2023) 19 European Competition Journal 481, 509 ff. From the case law see e.g. Unilever Italia (n 54), para 41 (‘that conduct had […] the ability to restrict competition on the merits’).

95

Post Danmark II (n 83), para 65; the ECJ referred to para 80 of AG Kokott’s Opinion (n 92) in this regard. For the same phrasing see: Servizio Elettrico Nazionale (n 38), para 70; with a reference to Post Danmark II (n 83), para 65; Unilever Italia (n 54), para 42 (with reference to the presumption of innocence, which suggests that the ECJ, in this passage, deals with the standard of proof).

96

European Superleague (n 54), para 130. For the need to consider the market context when determining potential effects see also: Policy Brief, p 3 (no ‘simplistic or formalistic standard’), referring to Generics (UK) (n 43), para 154; Telefónica (n 83), para 175; Deutsche Telekom (n 77), para 175; TeliaSonera (n 83), para 28; Case T-235/18 Qualcomm v Commission (Qualcomm—exclusivity payments), EU:T:2022:358, paras 396–398; Case C-165/19 P Slovak Telekom (n 89), para 42.

97

Unilever Italia (n 54), para 44.

98

See Servizio Elettrico Nazionale (n 38), para 70: If a competition authority assumes a causal link between a certain line of conduct and (potential) restrictions of competition, the specific market conditions on which the assumption is based must be met or at least likely to arise.

99

Deutscher (n 94) 510.

100

See also: Deutscher (n 94) 515.

101

In favour of an ‘expected value’ test in the context of digital mergers: Massimo Motta and Martin Peitz, ‘Big Tech Mergers’, Discussion Paper Series CRC TR 224, May 2020, pp 35 ff.

102

In the German legal tradition, Art. 102 TFEU would therefore qualify as a ‘konkretes Gefährdungsdelikt’ as opposed to an ‘abstraktes Gefährdungsdelikt’.

103

Hoffmann-La Roche (n 55), para 91.

104

In its subsequent decision-making practice, the Commission did verify whether an assumption of anticompetitive effects was justified in the circ*mstances of a given case, i.e. whether the setting was atypical and therefore required a closer effects analysis—see Case T-201/04 Microsoft (n 89), para 868.

105

Servizio Elettrico Nazionale (n 38), para 72; Generics (UK) (n 43), para 154, and Case C-165/19 P Slovak Telekom (n 89), para 42, as references.

106

Case T-612/17 Google Shopping (n 86), para 441.

107

2008 Priorities Guidance, para 21.

108

See, inter alia, Deutsche Telekom (n 77), para 257; and Telefónica (n 83), para 394–402, esp. 402.

109

Servizio Elettrico Nazionale (n 38), para 98: The competition authority had to show, supported by evidence, that the procedure to collect its customers’ consent for the transfer of data favoured a transfer of the data to its own subsidiaries.

110

Policy Brief, p 4.

111

Case T-612/17 Google Shopping (n 86), para 377 speaks of an ‘arbitrary or even impossible exercise’. See also: Castillo de la Torre (n 42) 168 with further references.

112

Case T-612/17 Google Shopping (n 86), paras 376 ff, para 382. It would then be for the dominant undertaking to come forward with relevant information that might cast doubt on the existence of a causal link (para 382).

113

See Case T-604/18 Google Android (n 50), para 893.

114

Case T-235/18 Qualcomm (n 96), paras 400–415. For an in-depth analysis see Deutscher (n 94).

115

For a defence of the counterfactual analysis see Neven (n 73) 566 and 571–572, arguing that the counterfactual is ‘one of the most fundamental disciplines in the assessment of effects’ and should not be relaxed. Similarly Daniel Higer, ‘Die jüngste Initiative der Kommission zu Art. 102 AEUV: Abkehr von einer bloßen Prioritätenmitteilung hin zu Leitlinien’ [2023] Neue Zeitschrift für Kartellrecht 385, 389. For the need to conduct a counterfactual analysis of some sort see also: Opinion of AG Kokott in Case C-48/22 P Google Shopping, EU:C:2024:14, paras 169 ff.

116

Deutscher (n 94) 484 ff.

117

Deutscher (n 94) 511–512.

118

Deutscher (n 94) 508; referring to Case T-612/17 Google Shopping (n 86), para 377.

119

See also the Opinion of AG Kokott in Case C-48/22 P Google Shopping, EU:C:2024:14, paras 169 ff.

120

For a similar line of argument, against the need for a counterfactual analysis in the strict sense, see Castillo de la Torre (n 42) 170–171.

121

Deutscher (n 94) 509.

122

For that wording see Post Danmark (n 69), para 22; Case C-413/14 P Intel (n 61), para 134; Servizio Elettrico Nazionale (n 38), para 45. See also: 2008 Priorities Guidance, para 6.

123

For that wording see Post Danmark (n 69), para 24 (‘normal competition on the basis of the performance of commercial operators’). See also Liza Lovdahl Gormsen, A Principled Approach to Abuse of Dominance in European Competition Law (CUP 2010) 45 (fn 155), who links the idea of ‘competition on the basis of performance’ to the German concept of ‘Leistungswettbewerb’ (see below n 124), as in the German version of the ECJ’s judgment in Hoffmann-La Roche (n 55), para 91.

124

Pablo Ibáñez Colomo suggests that the Union Courts’ case law in the 1970s and 1980s was informed by the idea that lawful and unlawful conduct could be told apart in the abstract—see Ibáñez Colomo (n 62) 389. It is unclear whether the ECJ’s early case law was indeed based on this assumption. See, for example, Case 6/72 Europemballage and Continental Can v Commission, EU:C:1973:22, para 27, emphasising that an abuse may exist ‘regardless of the means and procedure by which it is achieved’, rather focusing on effects (para 26). In Germany, where the concept of ‘competition on the merits’ was discussed early on (since the late 1920s, as ‘Leistungswettbewerb’), it was uncontroversial that this concept did not allow for an abstract distinction between lawful and unlawful conduct—see, for example, Burkhardt Röper, ‘Zur Verwirklichung des Leistungswettbewerbs’ in Hans-Jürgen Seraphim (ed) Zur Grundlegung wirtschaftspolitischer Konzeptionen (Duncker & Humblot 1960) 261 ff.

125

For references of the Union Courts to the concept of ‘performance-based competition’ see Post Danmark (n 69), para 24 (‘competition on the basis of performance’); Case C-62/86 AKZO v Commission, EU:C:1991:286, para 70 (‘competition on the basis of quality’); Case C-202/07 P France Télécom v Commission, EU:C:2009:214, para 106 (‘competition on the basis of quality’).

126

Case C-322/81 Michelin v Commission, EU:C:1983:313, para 57.

127

Servizio Elettrico Nazionale (n 38), para 85.

128

For the fact that the category of conduct that is not ‘on the merits’ is broader than the category of ‘naked exclusion’ as identified by the ‘no economic sense’ test see: Case T-612/17 Google Shopping (n 86), para 188.

129

This has also been pointed out by AG Rantos in his Opinion in Servizio Elettrico Nazionale (n 73), para 55, pointing out the ‘abstract’ nature of the concept of ‘competition on the merits’ and that it ‘cannot be defined in such a way as to make it possible to determine in advance whether or not particular conduct comes within the scope of such competition’. See also Neven (n 73) 570 and 572: The concept of ‘competition on the merits’ as such is not operational.

130

Servizio Elettrico Nazionale (n 38), para 73.

131

Servizio Elettrico Nazionale (n 38), para 73. According to Adriano Barbera, Nicolás Fajardo Acosta and Timo Klein, ‘The Role of the AEC Principle and Test in a Dynamic and Workable Effects-Based Approach to Abuse of Dominance’ (2023) 14 Journal of European Competition Law & Practice 582, 583, the AEC principle thereby recognises the importance of ‘appropriability’ in dynamic competition.

132

Servizio Elettrico Nazionale (n 38), para 79. See, however, Neven (n 73) 571, who doubts that the AEC principle ‘offers useful guidance for non-pricing practices’.

133

For the fact that there is not one singular AEC test, but that the way that the AEC principle is translated into a testable mathematical proposition will depend on the type of conduct, see Barbera, Fajardo Acosta and Klein (n 131) 587.

134

Barbera, Fajardo Acosta and Klein (n 131) 587.

135

Cf. 2008 Priorities Guidance, para 23 in section C ‘price-based exclusionary conduct’. For the fact that the AEC test is primarily applicable to price-based practices see Servizio Elettrico Nazionale (n 38), para 80. Very explicit on this point: Opinion of AG Kokott in Google Shopping (n 119), para 197: ‘the scope of application of the as-efficient-competitor test should not be extended to practices which bear no relation to price competition’. See, however, Barbera, Fajardo Acosta and Klein (n 131) 585, for the proposition that, in principle, AEC tests could be developed also for non-price conduct.

136

For an explanation of how the AEC test works in predation cases see Barbera, Fajardo Acosta and Klein (n 131) 587.

137

For an explanation of how the AEC test works in margin squeeze cases see Barbera, Fajardo Acosta and Klein (n 131) 588.

138

Servizio Elettrico Nazionale (n 38), para 76; Unilever Italia (n 54), para 39.

139

However, according to the 2008 Priorities Guidance, para 27, a finding that a price charged by the dominant undertaking has the potential to foreclose equally efficient competitors should be considered just one piece of evidence, to be integrated ‘in the general assessment of foreclosure …, taking into account other relevant quantitative and/or qualitative evidence’. See also Neven (n 73) 570. Similarly Barbera, Fajardo Acosta and Klein (n 131) 591–592.

140

For the manageability of the AEC test in a margin squeeze case see, for example, Commission Decision in Slovak Telekom (n 18), pp 228–262: the AEC test spans only 34 pages or 8.4 per cent of the total decision length, even though the dominant undertaking did not fully comply with the Commission’s requests for information.

141

Cf. for predation: AKZO (n 125), para 74; for margin squeeze: TeliaSonera (n 83), para 41. This advantage is also emphasised in the literature, see, for example, Barbera, Fajardo Acosta and Klein (n 131) 587.

142

AKZO (n 125), para 72; Case T-271/03 Deutsche Telekom v Commission, EU:T:2008:101, para 194.

143

Post Danmark (n 69), para 25; Servizio Elettrico Nazionale (n 38), para 72. See also: Ibáñez Colomo (n 62) 415.

144

On the functioning of the AEC test in these cases, see Nazzini (n 82) 235–242; and Barbera, Fajardo Acosta and Klein (n 131) 588–590.

145

See Unilever Italia (n 54), para 39.

146

For an illustration of the complexity of the calculation see Commission Decision in Intel (n 18), recitals 1010–1011. Cf. on that, Robert Lauer, ‘The Intel saga: what went wrong with the Commission’s AEC test (in the General Court’s view)?’ (2024) 20 European Competition Journal 45, 58 ff.

147

For critical comments on the failure of the GC to recognise ‘the margin of error that is inherent in the implementation of the test’ and to grant the Commission a ‘confidence interval around the point of estimate of the difference between cost and effective price’ see, inter alia, Neven (n 73) 566 and 575. In particular, the Court’s view that there is ‘a single correct number’ for the contestable share would misunderstand ‘the nature of the empirical exercise that the implementation of the AEC test involves’.

148

Cf. for the critique Fiona M Scott Morton and Zachary Abrahamson, ‘A Unifying Analytical Framework for Loyalty Rebates’ (2017) 81 Antitrust Law Journal 777, 796 ff; Lauer (n 146) 56 ff.

149

Commission Decision in Intel (n 18), pp 302–453.

150

Cf. Wouter PJ Wils, ‘The Judgment of the EU General Court in Intel and the So-Called “More Economic Approach” to Abuse of Dominance’ (2014) 37 World Competition 405, 431.

151

Castillo de la Torre (n 42) 191–192: ‘If the EU Courts are too demanding on how the AEC test is conducted, they may be providing arguments not to use it’.

152

See Case T-604/18 Google Android (n 50), paras 643–644; Case T-286/09 RENV Intel Corporation v Commission, EU:T:2022:19, paras 244 ff. For a critical perspective see Castillo de la Torre (n 42) 164–165. Similarly Laitenberger and Kröger (n 19) 625: ‘in complex cases, the judicial review puts a very heavy burden on enforcement’.

153

See 2008 Priorities Guidance, paras 23 and 27: If the data suggested that an equally efficient competitor could compete effectively with the dominant undertaking’s pricing conduct, the Commission would not intervene.

154

The Amending Communication merely confirms a prior shift in the Commission’s practice, see Amending Communication, Annex para 3. See also: Policy Brief, pp 6–7. For a different view on the appropriateness of an AEC test in case of non-price conduct see Barbera, Fajardo Acosta, and Klein (n 131) 585: ‘there is no reason, in principle, why [an AEC test] cannot also be developed for non-price conduct, such as self-preferencing and bundling’.

155

Recent publications personally linked to DG Comp suggest that the Commission is planning on continuing to refrain from conducting AEC tests in these cases, see especially Kadar and Holzwarth (n 52) 333, pointing to the high consumption of resources, the problems in accurately calculating the contestable share and the general ‘administrability’.

156

According to long-standing case-law, the AEC test is just one methodological option—see Post Danmark II (n 83), paras 56–61; Servizio Elettrico Nazionale (n 38), para 81; Unilever Italia (n 54), paras 57–58. Even if the Commission does not conduct an AEC test itself, it is obliged to engage with an AEC test presented by the dominant undertaking as part of its defence. From the Commission’s perspective, this may, however, be more manageable and less risky (further on this: see below, IV. B. 2.).

157

Whether these limits justify dispensing with an AEC test beyond the established categories of predatory pricing and margin squeeze, or rather calls for an AEC principle and AEC tests that are framed not as a generally applicable principle, but as contingent on a specific theory of harm, is currently being debated. In favour of a framing of the AEC principle and tests as contingent on varying theories of harm: Neven (n 73) 580.

158

This has recently been pointed out by commentators, see Barbera, Fajardo Acosta and Klein (n 131) 587 and 591; Neven (n 73) 574.

159

Neven (n 73) 579.

160

Neven (n 73) 577–578.

161

Barbera, Fajardo Acosta and Klein (n 131) 592.

162

For the fact that the concept of ‘competition on the merits’ is broader than the AEC test see Unilever Italia (n 54), para 39: an abuse is established either ‘where the conduct complained of produced exclusionary effects in respect of competitors that were as-efficient as the perpetrator of that conduct in terms of cost structure, capacity to innovate, quality’; or ‘where that conduct was based on the use of means other than those which come within the scope of “normal” competition, that is to say, competition on the merits’ (emphasis added). See also Opinion of AG Kokott in Google Shopping (n 119), para 196: ‘The ACE test is not generally applicable, letalone an essential prerequisite for determining whether the conduct of an undertaking in a dominant position is in keeping with the means of competition on the merits’. For the opposite position see Akman (n 81), p 22, suggesting that the AEC principle is conclusive.

163

See, for example, TeliaSonera (n 83), para 43.

164

As recently observed by Damien Neven, a main function of the AEC principle is that it marks ‘a red line’—see Neven (n 73) 565 and throughout. However, Neven points out that the ‘red line’ principle may lead to type I errors (at pp 576, 577–578) and suggests that this concern should be reflected in future guidelines.

165

See 2008 Priorities Guidance, para 24; Policy Brief, fn 44; Opinion of AG Kokott in Google Shopping (n 119), para 194. Also Neven (n 73) 570; Germain Gaudin and Despoina Mantzari, ‘Google Shopping and the As-Efficient-Competitor Test: Taking Stock and Looking Ahead’ (2022) 13 Journal of European Competition Law & Practice 125, 126–127; Pablo Ibáñez Colomo, ‘Anticompetitive Effects in EU Competition Law’ (2020) 17 Journal of Competition Law & Economics 309, 339.

166

European Superleague (n 54), para 131.

167

European Superleague (n 54), para 131, with reference to Generics (UK) (n 43), paras 154–157. Inter alia, the wording in European Superleague encompasses cases in which the dominant undertaking denies rivals a minimum scale—on this, see: Benjamin Klein, ‘Exclusive Dealing as Competition for Distribution “On the Merits”’ (2003) 12 George Mason Law Review 119.

168

Post Danmark II (n 83), para 59.

169

Emphasised by Ibáñez Colomo (n 62) 407.

170

The same may be true where efficiencies result from the scope of products that the dominant undertaking offers, but hypothetical competitors do not and cannot offer the same range of products—see Neven (n 73) 570–571.

171

Opinion of AG Kokott in Google Shopping (n 119), para 195.

172

See the Amending Communication, rephrasing para 23 of the 2008 Priorities Guidance; Neven (n 73) 566.

173

Barbera, Fajardo Acosta, and Klein (n 131) 592 and 593, discussing when—and when not—the goal of promoting contestability should prevail over the principle that artificially keeping less efficient firms in the market will not be in the interest of efficiency and consumers.

174

For the general distinction see: Louis Kaplow, ‘Rules Versus Standards: An Economic Analysis’ (1992) 42 Duke Law Journal 557.

175

See Andriani Kalintiri, Evidence Standards in EU Competition Enforcement (Hart 2019) 72–73; 79–80. If both the substantive threshold and the standard of proof were interpreted in probabilistic terms, this could theoretically lead to a finding that relatively low probabilities of anticompetitive effects would suffice for the finding of an abuse. For a different perspective see Deutscher (n 94) 504 ff.

176

For this power of the Commission in proceedings under Art. 7 of Regulation 1/2003 see Art. 23 of Regulation 1/2003.

177

For the general criteria for qualifying proceedings as ‘criminal’ within the meaning of Art. 6 ECHR see Engel and Others v The Netherlands, App no 5100/71 (ECtHR 8 June 1976), cf. Kalintiri (n 175) 25 ff. See for competition authorities’ proceedings Menarini Diagnostics SRL v Italy, App no 43509/08 (ECtHR 27 September 2011).

178

Minelli v Switzerland (1983) Series A no 4, para 37; Poncelet v Belgium App no 44418/07 (ECtHR 30 March 2010), para 50; Garycki v Poland App no 14348/02 (ECtHR, 6 February 2007), para 68.

179

Kalintiri (n 175) 81–85.

180

Kalintiri (n 175) 85; similarly Fernando Castillo de la Torre and Eric Gippini Fournier, Evidence, Proof and Judicial Review in EU Competition Law (Edward Elgar 2017), para 2.029: ‘generally judges see no essential difference’. In a few cases, the ‘beyond a reasonable doubt’ standard has been explicitly referred to—see Opinion of Mr Vesterdorf acting as AG in Case T-1/89 Rhône-Poulenc v Commission, EU:T:1991:38, p II-954; similarly Joined Cases 29/83 and 30/83 CRAM and Rheinzink/Commission, EU:C:1984:130, para 16.

181

See, for example, Unilever Italia (n 54), para 42 (with further references to case law). In other cases, it remains unclear whether the Courts aim to specify the level of risk required by Art. 102 TFEU or rather the standard of proof—Critical in this regard: Castillo de la Torre (n 42) 153–154.

182

For the standard of proof regarding such facts see Opinion of AG Kokott in Case C-8/08T-Mobile Netherlands and Others, EU:C:2009:110, fn 60: ‘the standard of proof determines the requirements which must be satisfied for facts to be regarded as proven.’

183

Unilever Italia (n 54), para 42; citing Case 27/76 United Brands v Commission, EU:C:1978:22, para 265, and Joined Cases C-89/85, C-104/85, C-114/85, C-116/85, C-117/85 and C-125/85 to C-129/85 Ahlström Osakeyhtiö and Others v Commission, EU:C:1993:120, para 126.

184

If read in this light, Unilever Italia does not interfere with the principle that it is for EU law to define the substantive requirements of an abuse as well as the allocation of the burden of proof, but for national procedural law to define the standard of proof (see recital 5 of Regulation 1/2003) and to decide on the rules on evidence (see also Case C-211/22 Super Bock Bebidas, EU:C:2023:529, para 55, with a view to Art. 101 TFEU). What is more, given that, at the national level, Art. 102 TFEU may be enforced in a purely administrative procedure, with no threat of fines (See, for example, § 54(1) GWB (Gesetz gegen Wettbewerbsbeschränkungen; German Act against restraints of competition)), a ‘beyond reasonable doubt’-standard cannot be generalised. National rules on standard of proof and on evidence must, however, be in line with the general principles of equivalence and effectiveness—cf. Case C-453/99 Courage and Crehan, EU:C:2001:465, para 29.

185

For an attempt to sketch the theoretical framework for the assessment of evidence see Opinion of Mr Vesterdorf acting as AG in Rhône-Poulenc v Commission (n 180), p II-954. The approach has been adapted by the Union Courts—see cf. Joined Cases T-25/95, T-26/95, T-30/95, T-31/95, T-32/95, T-34/95, T-35/95, T-36/95, T-37/95, T-38/95, T-39/95, T-42/95, T-43/95, T-44/95, T-45/95, T-46/95, T-48/95, T-50/95, T-51/95, T-52/95, T-53/95, T-54/95, T-55/95, T-56/95, T-57/95, T-58/95, T-59/95, T-60/95, T-61/95, T-62/95, T-63/95, T-64/95, T-65/95, T-68/95, T-69/95, T-70/95, T-71/95, T-87/95, T-88/95, T-103/95 and T-104/95 Cimenteries CBR and Others v Commission, EU:T:2000:77, para 1838. For the holistic approach to the assessment of evidence see Castillo de la Torre and Gippini Fournier (n 180) 78–86; see e.g. Case T-321/05 AstraZeneca v Commission, EU:T:2010:266, para 477 with further references to the case law.

186

Sometimes, these inferences are also referred to as ‘factual presumptions’. To clearly distinguish factual inferences from legal presumptions, we exclusively use the term ‘factual inferences’ in the following text. Contrary to ‘factual inferences’, legal presumptions do not have a purely empirical, but (also, sometimes primarily) a normative basis (e.g. the goal to elicit information from the defendant or to react to the cost and/or difficulties of access to evidence).

187

For this purpose of presumptions see Cyril Ritter, ‘Presumptions in EU competition law’ (2018) 6 Journal of Antitrust Enforcement 189, 206.

188

Hoffmann-La Roche (n 55), paras 89–90.

189

See, for example, Akman (n 81), p 6, with further references to the wide-spread criticism of an overly ‘formalistic’ jurisprudence on Art. 102 TFEU in the early years.

190

See, to that respect, also Wils (n 150) 428, arguing that the degree of legal certainty and the corresponding allocation of risk should be considered in interpreting Art. 102 TFEU.

191

Case C-413/14 P Intel (n 61), para 138: ‘that case-law must be further clarified in the case where the undertaking concerned submits, during the administrative procedure, on the basis of supporting evidence, that its conduct was not capable of restricting competition and, in particular, of producing the alleged foreclosure effects’.

192

See Case T-286/09 RENV Intel (n 152), paras 124, 129.

193

Case C-413/14 P Intel (n 61), paras 138–140. See also Castillo de la Torre (n 42) 180–181; Raffaele Di Giovanni Bezzi, ‘Anticompetitive Effects and Allocation of the Burden of Proof in Article 102 Cases: Lessons from the Google Shopping Case’ (2022) 13 Journal of European Competition Law & Practice 112, 122–124.

194

One may assume that verifying an AEC test presented by the dominant undertaking is much less time-consuming and resource-intensive than conducting an AEC test of its own.

195

See Opinion of AG Kokott in Google Shopping (n 119), para 196: ‘in so far as [the AEC test] is not applicable, neither the Commission nor the General Court can be compelled to consider arguments by the dominant undertaking in connection with its use’.

196

See Servizio Elettrico Nazionale (n 38), para 70: If a competition authority assumes a causal link between a certain line of conduct and (potential) restrictions of competition, the specific market conditions on which the assumption is based must be met or at least likely to arise.

197

Unilever Italia (n 54), para 44: When assessing whether an undertaking’s conduct is capable of restricting effective competition, a competition authority may rely on the guidance from economic sciences, confirmed by empirical or behavioural studies. However, factors specific to the circ*mstances of the case must be taken into account, ‘such as the extent of that conduct on the market, capacity constraints on suppliers of raw materials, or the fact that the undertaking in a dominant position is, at least, for part of the demand, an inevitable partner’. In Unilever Italia, the ECJ has required such analysis irrespective of the defendant coming forward with substantiated evidence suggesting the absence of anticompetitive effects. It is unclear whether the ECJ thereby strives to establish a qualified standard for factual inferences of anticompetitive effects.

198

European Superleague (n 54), para 130.

199

Servizio Elettrico Nazionale (n 38), para 52. See also Post Danmark II (n 83), para 68. Critically Castillo de la Torre (n 42) 174 ff.

200

See, for example, Servizio Elettrico Nazionale (n 38), para 72: Whether a practice is or was capable of restricting competition ‘must be assessed having regard to all the relevant facts (see, to that effect, judgments of 30 January 2020, Generics (UK) and Others, C-307/18, EU:C:2020:52, paragraph 154, and of 25 March 2021, Slovak Telekom v Commission, C-165/19 P, EU:C:2021:239, paragraph 42).’ See also: Unilever Italia (n 54), para 40: ‘it is for the competition authorities to demonstrate the abusive nature of conduct in the light of all the relevant factual circ*mstances surrounding the conduct in question (judgments of 19 April 2012, Tomra Systems and Others v Commission, C-549/10 P, EU:C:2012:221, paragraph 18, and of 12 May 2022, Servizio Elettrico Nazionale and Others, C-377/20, EU:C:2022:379, paragraph 72), which includes those highlighted by the evidence adduced in defence by the undertaking in a dominant position.’

201

For a critical perspective regarding the ‘all circ*mstances considered’-jurisprudence see Castillo de la Torre (n 42) 175–176.

202

See, for example, the undertaking’s presentation of an AEC test in Qualcomm (Exclusivity payments) (Case AT.40220) Commission Decision of 24 January 2018, recital 487.

203

As a matter of policy, a different approach would also seem feasible: To ensure the administrability of abuse proceedings, one may want to require the defendant to first show the inadequacy of the Commission’s methodological choice, or that the outcome of the Commission’s analysis was inconclusive, before obliging the Commission to consider a different type of test. This approach would protect the Commission’s margin of judgment in choosing an adequate methodology to establish anticompetitive effects in a given case.

204

TeliaSonera (n 83), para 64; Post Danmark II (n 83), para 66; Servizio Elettrico Nazionale (n 38), para 54, and Unilever Italia (n 54), para 41.

205

Servizio Elettrico Nazionale (n 38), para 54; Unilever Italia (n 54), para 41.

206

Servizio Elettrico Nazionale (n 38), para 56.

207

Microsoft (Case COMP/C-3/37.792) Commission Decision of 24 March 2004, recital 841. See also Case T-201/04 Microsoft (n 89), para 977.

208

Case T-604/18 Google Android (n 50), para 295.

209

Case T-604/18 Google Android (n 50), para 294.

210

See, however, Laitenberger and Kröger (n 19) 632, pointing towards suggestions for a well-dosed … ‘form based’ or—as some might prefer to call it—‘more normative approach’.

211

Often, the normative goal is accompanied by an empirical link in cases of legal presumptions, too. This link might, however, not be strong enough to justify a factual inference, see above IV. B. 1.

212

Efforts in this direction also appear at Kadar and Holzwarth (n 52) 333: ‘Setting up clear-cut principles of enforcement based on experience and sound economics is not incompatible with an effects-based approach’.

213

Pablo Ibáñez Colomo speaks of ‘proxies’. Within the terminological categorisation proposed above, these ‘proxies’ would qualify as ‘factual inferences’.

214

Like, for example, the degree of dominance, market coverage and duration of the practice etc. See Ibáñez Colomo (n 92) 619–620.

215

José Luís Cruz Vilaça, ‘The intensity of judicial review in complex economic matters—recent competition law judgments of the Court of Justice of the EU’ (2018) 6 Journal of Antitrust Enforcement 173, 187: ‘The great challenge for the jurist, and in particular for the judge, is to turn economic theories into solid legal criteria, capable of securing the clarity of the concepts and their adaptability to a complex reality, as well as to enhance legal certainty and predictability in the application of the law’.

216

2008 Priorities Guidance, para 22.

217

2008 Priorities Guidance, para 22.

218

See Ibáñez Colomo (n 62) 413–414. Others have objected, pointing to the wording of Art. 102 TFEU which does not differentiate between abuses by object or effect—see Akman (n 81), p 11.

219

See, on the one hand, Case T-612/17 Google Shopping (n 86), paras 435 ff, explicitly negating the existence of ‘abuses by object’. See, however, European Superleague (n 54), para 131, where the possibility to prove an anticompetitive object is mentioned, albeit without clarifying the legal preconditions and implications of such a finding.

220

Such a view is suggested by Super Bock (n 184), paras 34–37.

221

See Super Bock (n 184), para 36.

222

See, for example, Case T-203/01 Michelin v Commission (Michelin II), EU:T:2003:250, para 241; Case T-340/03 France Télécom v Commission, EU:T:2007:22, para 195: ‘showing an anticompetitive object and an anticompetitive effect may, in some cases, be one and the same thing. If it is shown that the object pursued by the conduct of an undertaking in a dominant position is to restrict competition, that conduct is liable to have such an effect.’ For a different view—relying on the wording of Art. 102 TFEU as compared with the wording of Art. 101 TFEU—see Akman (n 81), p 11.

223

2008 Priorities Guidance, para 22.

224

See Commission Decision in Intel (n 18), recitals 1641 ff. See also: Intel (Case AT.37990) Commission Decision of 22 September 2023 (imposing fines for the naked restriction after the GC, in Case T-286/09 RENV Intel (n 152), had fully annulled the Commission’s 2009 Decision in Intel (n 18), but had simultaneously confirmed that Intel’s payments to the three computer manufacturers amounted to an abuse of dominance).

225

Commission Decision in Intel (n 18), recital 1672.

226

See Michele Giannino, ‘Lithuanian Railways: The Court of Justice Narrows Down the Scope of Application of the Doctrine of Essential Facilities’ (2023) 7 European Competition and Regulatory Law Review 260, 263; Kadar, Holzwarth and Pereira (n 33), p 4; Pablo Ibáñez Colomo, ‘GC Judgment in Case T-814/17, Lithuanian Railways—Part I: object and indispensability’ (Chillin’Competition, 1 December 2020)<www.chillingcompetition.com/2020/12/01/gc-judgment-in-case-t%E2%80%91,814–17-lithuanian-railways-part-i-object-and-indispensability/> accessed 31 March 2024.

227

See Baltic Rail (n 18), pp 65–94: Roughly one-third of the total length of the decision relates to the effects analysis. On this see Kadar, Holzwarth and Pereira (n 33), p 4, insisting that the Commission was not required to engage in a detailed effects analysis in this case, considering the seriousness of the restriction and the lack of plausible efficiency justifications.

228

See Servizio Elettrico Nazionale (n 38), para 77—which uses this definition to identify practices that do not come within the scope of ‘competition on the merits’. For a discussion of the ‘no economic sense’ test, see U.S. Department of Justice, ‘Competition and Monopoly: Single-Firm Conduct under Section 2 of the Sherman Act’ (2008) 39.

229

Super Bock (n 184), para 36.

230

Servizio Elettrico Nazionale (n 38), para 60 (with further references).

231

Servizio Elettrico Nazionale (n 38), para 62 (with further references).

232

Servizio Elettrico Nazionale (n 38), para 63 (with further references). See also: 2008 Priorities Guidance, para 20, where the Commission considers direct evidence as one of a number of factors that the Commission will consider in finding an abuse.

233

Servizio Elettrico Nazionale (n 38), para 63 (with further references). Further on the role of intent: Mariateresa Maggiolino, ‘The role of intent in abuse of dominance and monopolization’, in Pınar Akman, Or Brook and Konstantinos Stylianou (eds) Research Handbook on Abuse of Dominance and Monopolization (Edward Elgar 2023) ch 13; Nazzini (n 82) ch 6: ‘The Test of Intent’.

234

See 2008 Priorities Guidance, para 20.

235

Case T-111/96 ITT Promedia v Commission, EU:T:1998:183, para 55; AKZO (n 125), para 72.

236

AKZO (n 125), para 72.

237

Michelin II (n 222), paras 241–242; Case T-340/03 France Télécom (n 222), para 195.

238

Nazzini (n 82) 206.

239

See Sot. Lélos kai Sia (n 46), para 66.

240

See, for example, Case T-201/04 Microsoft (n 89), paras 650–652, 1046, 1069; also: Commission Decision in Intel (n 18), recitals 1679, 1747; also: Castillo de la Torre (n 42) 181–182.

241

See Castillo de la Torre (n 42) 181, pointing to Unilever Italia (n 54), para 57.

242

For the effects analysis with a view to the closing of the interface for competing work group server operating systems to interoperate with Windows, see: Case T-201/04 Microsoft (n 89), paras 565 ff. For the effects analysis with regard to the pre-installation of the Windows Media Player see Case T-201/04 Microsoft (n 89), paras 976 ff.

243

Where a competition authority relies on the existence of less restrictive alternative practices, the defendant may challenge the viability of this alternative or show that such an intrusion into the freedom to conduct one’s business is disproportionate.

244

British Airways (n 56), para 86; TeliaSonera (n 83), para 76; Post Danmark (n 69), paras 41–42. See also: 2008 Priorities Guidance, paras 28 ff.

245

See Michelin (n 126), para 57; Case C-413/14 P Intel (n 61), para 135; Servizio Elettrico Nazionale (n 38), para 74.

246

This analytical shortcut does not amount to an empowerment of the Commission to engage in creative market engineering. The ‘less restrictive alternative’ test should be limited to some well-established standard settings.

247

For an apparently further-reaching position see Ibáñez Colomo (n 62) 403–404: Conduct that is, by its very nature, at odds with competition on the merits shall be considered presumptively abusive, with no need to analyse effects. However, the dominant undertaking can show that the conduct is incapable of having restrictive effects.

248

Servizio Elettrico Nazionale (n 38), para 91.

249

Servizio Elettrico Nazionale (n 38), para 99.

250

Servizio Elettrico Nazionale (n 38), para 100.

251

Servizio Elettrico Nazionale (n 38), para 102.

252

Case C-457/10 P AstraZeneca (n 83), para 132. See also para 133: The fact that the objectives of those other rules are complied with does not suggest that the application of Art. 102 TFEU is no longer required to ensure effective and undistorted competition within the internal market.

253

Servizio Elettrico Nazionale (n 38), para 67.

254

Case C-252/21 Meta Platforms and Others, EU:C:2023:537, para 47.

255

Meta Platforms (n 254), para 48. In his Opinion in the Meta Platforms case, AG Rantos had taken a somewhat more cautious position, arguing that the compliance or non-compliance of that conduct with the GDPR, ‘not taken in isolation but considering all the circ*mstances of the case, may be a vital clue as to whether that conduct entails resorting to methods prevailing under merit-based competition’—see Opinion of AG Rantos in Case C-252/21 Meta Platforms and Other, EU:C:2022:704, para 23.

256

BGH, 23 June 2020, KVR 69/19—Facebook.

257

BGH, 23 June 2020, KVR 69/19—Facebook, para 76.

258

Paul L Joskow, ‘Transaction Cost Economics, Antitrust Rules and Remedies’ (2002) 18 Journal of Law, Economics, and Organization 95, 99f.

259

Policy Brief, title.

260

See, for example, Germany (see Nada Ina Pauer, ‘The 11th Amendment of the German Act Against Restraints of Competition’ (2023) 14 Journal of European Competition Law & Practice 354, 355–357); a novel market investigation tool is currently being considered in Sweden (see e.g. Helene Andersson, ‘Main Developments in Competition Law and Policy 2023—Sweden’ (Kluwer Competition Law Blog, 19 February 2024)<www.competitionlawblog.kluwercompetitionlaw.com/2024/02/19/main-developments-in-competition-law-and-policy-2023-sweden/> accessed 31 March 2024).

261

On this, see Massimo Motta, Martin Peitz and Heike Schweitzer (eds), Market Investigations: A New Competition Tool for Europe? (CUP 2021).

262

On this, see Laitenberger and Kröger (n 19) 633 ff.

Author notes

Professor Dr. Heike Schweitzer, LL.M. (Yale) holds a Chair for Private Law and Competition Law and Economics at Humboldt-Universität zu Berlin. Simon de Ridder is doctoral researcher at the Chair. For their comments on an earlier draft of this paper, we are grateful to Elias Deutscher, Andreas Engert, Lennart Enwaldt, Johannes Holzwarth, Philipp Hornung, Pablo Ibáñez Colomo, Thorsten Käseberg, Robert Welker, and Wouter Wils. For valuable support in conducting the empirical analysis, we thank Maximilian Wolters. Of course, all opinions and remaining mistakes are ours.

© The Author(s) 2024. Published by Oxford University Press.