Chillin'Competition

Last week’s post discussed a quintessential example of a ‘by object’, namely disparagement strategies. By the end of the entry, I pointed out that, for some reason, there is some reluctance to call a spade a spade or, more precisely, to call abuses by object by their name. A number of potential alternative labels have been floated, such as ‘naked restrictions’.

It is not immediately obvious to understand the reticence to accept this category. Abuses by object are neither an academic theory nor a policy proposal: they are a creation of the Court of Justice.

A cursory overview of the case law reveals that there are some practices (say, pricing below AVC à la AKZO) that are prohibited (i) without the need to show anticompetitive effects precisely because (ii) they have no plausible explanation other than the exclusion of a rival.

What is more, Superleague dissipated any doubts that might have existed about the existence of abuses by object in the world of positive law.

If abuses by object are very much a thing, what explains, then, the reluctance to embrace a concept which brings not just clarity and consistency but which, moreover, dispenses with the need to establish the exclusionary impact of some practices?

Reading some commentary here and elsewhere, I believe I have come to understand, the mystery behind this counterintuitive attitude.

The reluctance appears to be based on a misunderstanding of the concept and operation of ‘by object’ infringements. According to this (mis)understanding, this category of infringement would be incompatible with some features of the case law interpreting the notion of abuse.

The Court made it clear in Intel that, even when the anticompetitive effects of a practice are presumed, it is always possible for the dominant firm to produce evidence showing that the said practice is incapable of having an exclusionary impact.

This is where, the argument goes, Article 102 TFEU departs from ‘by object’ infringements. According to this view, it is not possible to rebut a finding that an agreement restricts competition by object under Article 101(1) TFEU by showing that it is incapable of having anticompetitive effects.

The only problem is that this interpretation of Article 101(1) TFEU is, as the law stands, incorrect. Not only is it possible to rebut a finding of a ‘by object’ infringement on the basis of the absence of effects, this argument has been successfully invoked in a number of cases.

The judgment of the Court of Justice in Servier made this point clear, and arguably more explicitly than preceding ones (AG Kokott’s Opinion in Generics articulated this idea very effectively too). An agreement can only infringe Article 101(1) TFEU, whether by object or effect, where there is (actual or potential) competition to restrict in the first place.

By the same token, the agreement will escape the prohibition where it appears that competition would have been impossible (which would be the case, for instance, there are regulatory barriers to entry that prevented such competition irrespective of the behaviour of the parties).

Again, this is not such a theoretical possibility. As mentioned a few times here, this very question led to the partial annulment of the Commission decision in E.On Ruhrgas. It was also raised (unsuccessfully) in Toshiba, which was a plain-vanilla cartel case.

What matter for the purposes of this discussion is that, had the members of the cartel, proved that regulatory barriers to entry made competition between European and Japanese producers impossible, there would have been no infringement. This conclusion is clear from the appeal judgment in the case.

Against this background, there is nothing in the Article 101(1) TFEU case law that is at odds with Intel. When a practice is abusive by object, the possibility to rebut the presumption of effects exists, just like it does under Article 102 TFEU.

As far as ‘by object’ conduct is concerned, however, the bar to rebut the presumption is very high, as the Commission rightly explains in its Draft Guidelines. The dominant firm would have to show that competition would have been impossible in the relevant economic and legal context. The bar may be high, but the possibility exists nonetheless.

Relying on ‘abuse by object’ as a category, as opposed to ‘naked restriction’, has an additional advantage. It seems to me that the former is broader than the latter. There is conduct that is prohibited as a ‘by object’ infringement even though it can be plausibly explained other than as a means to restrict competition.

Conduct aimed at partitioning the internal market, at issue in cases like ABInbev, is the example that comes to mind immediately. It may be true that sometimes price discrimination can be rationalised as a pro-competitive strategy.

To the extent that it goes against the overarching objective of market integration, however, it is prohibited by object under Article 101(1) TFEU. And, coming back to the theme that motivated this post, it stands to reason that it is also prohibited as such under Article 102 TFEU.

Which takes me to the conclusion: it would be immensely beneficial if the codification exercise underpinning the Guidelines streamlined the fundamental issues and called ‘abuses by object’ by their name.

I have spent the past few weeks (re-)reading in detail the case law and administrative practice on disparagement and related issues, including landmark rulings such as (the other) Hoffmann-La Roche and AstraZeneca. The Commission decision in Teva-Copaxone is a fascinating read that builds on the extensive experience acquired at the national level (see here for a discussion of the some recent developments in France, where disparagement has long been a priority).

The reasons why I became interested in these cases will be apparent in a few weeks’ time. I will explore a different set of questions for the time being. As I was reading judgments and decisions, I realised they provide valuable lessons for the interpretation of Article 102 TFEU and the forthcoming Guidelines on exclusionary abuses.

My first thought was that disparagement cases perfectly illustrate why the Court of Justice is right to have crafted a ‘by object’ category of abuses. If conduct is restrictive by object under Article 101(1) TFEU, it stands to reason that it is also prohibited under Article 102 TFEU by its very nature (that is, without the need to show that it is capable of having anticompetitive effects).

In (the other) Hoffmann-La Roche, the Court held that an agreement aimed at disseminating misleading information with a view to reducing the competitive pressure that one product puts on another is restrictive by object. Nothing would justify requiring a case-by-case analysis of effects where the same practice is unilaterally implemented by a dominant firm.

The Commission was cautious in its analysis in Teva-Copaxone, and took the care of looking at the exclusionary capability in the relevant market. While this approach is easy to rationalise from the authority’s perspective, treating disparagement conduct as a ‘by object’ infringement under both Articles 101 and 102 TFEU would be preferable.

In this sense, it seems to me that the Commission would have discharged its burden of proof after showing that the behaviour is objectively aimed at disparaging a rival. From this perspective, the analysis of the anticompetitive effects would be superfluous as a matter of law – but might make sense as a safeguard from a policy-making perspective.

Since the decision in Teva-Copaxone has been challenged before the General Court, we may get an answer on this crucial point of law.

Consistency also demands that abuses by object are called by name, as opposed to say, ‘naked restrictions’ or ‘per se abuses’ (remember there is no such thing as a per se infringement in EU competition law).

The latest iteration of the Draft Guidelines on exclusionary abuses resists the idea of calling a spade a spade, thereby adding unnecessary clutter to law and policy. It would be preferable if the final version streamlined the issue by embracing the Court’s case law (which is, after all, the goal of the project).

My next post will explain why reluctance to accept the case law as it stands is unwarranted. In the meantime, I look forward to your comments on these questions.

The new edition of the LSE Short Course on State Aid and Subsidies Regulation will be organised, again, in February of this year. I really look forward to it, as it comes at a time when major developments are reshaping the discipline in fundamental ways (so fundamental, in fact, that they inspired me to write a whole book on them).

The Short Course will cover all these developments, including the changes brought about by the Clean Industrial Deal framework, the emerging administrative practice under the EU Foreign Subsidies Regulation and the growing case law interpreting the UK Subsidy Control Act 2022.

It always seemed to me that State aid and subsidies are learnt ‘along the way’ by many lawyers. The point of this Short Course is precisely to provide what many might miss: the necessary basics and conceptual framework to navigate the discipline and remain on top of contemporary trends.

As usual, the Short Course will take place online and it is designed with full-time professionals in mind. Attendance will be capped at around 25 participants to maximise interaction (always one of the big pluses of this format: every edition is shaped by the discussions that take place).

The sessions will be on four consecutive Thursdays: 6th, 13th, 20th and 27th February (at the usual time: 2pm to 6pm London time).

An LSE Certificate will of course be available upon completion, along with CPD points for practitioners.

If you have any questions about the organisational aspects of the two courses, do not hesitate to contact my wonderful colleague Mandy Tinnams: A.Tinnams@lse.ac.uk.

In competition law terms, last year was marked by Android Auto. Following the judgment, the applicability of the Magill and Bronner doctrines depends on two questions: (i) whether the dominant firm has developed the assets ‘solely for the needs of its own business‘ and (ii) whether a duty to deal would ‘fundamentally alter the economic model‘ on which the development of the assets relied.

The refusal to deal doctrines are only relevant where these two cumulative conditions are met. Where either of these conditions fails, the potentially abusive nature of the refusal will be evaluated in light of the standard effects analysis that applies as a default under Article 102 TFEU.

A central question that was never tackled by the Court in Android Auto is whether Enel and Google are competitors in the relevant adjacent market (and, similarly, whether Enel’s app and Google Maps were rival products).

The whole analysis is carried out on the assumption that Enel and Google are indeed competitors in the said adjacent market, even if the latter is merely hypothetical or has shifting boundaries.

There were good reasons for the Court to make this assumption. After all, the Magill and Bronner doctrines only provide support for intervention where the dominant firm and the one requesting access are actual or potential competitors.

As the law stands, there is no basis in the case law to support the idea that Article 102 TFEU can be relied upon to compel a dominant firm to deal with a non-competitor – what I call an exploitative refusal to deal.

It would therefore be necessary to develop an ad hoc doctrine extending the range of instances where an undertaking can be order to share an asset with a firm with which it has chosen not to deal. In the alternative, the Court could choose to expand the reach of Android Auto.

In a sense, Android Auto lends itself quite naturally to an extension along these lines. The fundamental idea behind the judgment, after all, is that a dominant firm having developed a (partially) open platform must accept that firms operating in and around the platform become involved in its design.

In such circumstances, the only twist that would be needed relates to the assessment of anticompetitive effects. Instead of exclusionary, such effects would be exploitative.

One could argue, for instance, that, by failing to feature a particular application (or category thereof), end-consumers would be deprived of more attractive functionalities without an objective justification.

Where the dominant firm is already giving access to another provider in the relevant category, one could argue, in addition, that refusing to deal with another provider amounts to exploitative discrimination within the meaning of Article 102(c) TFEU.

Such an expansion would not be wholly uncontroversial. A potential argument against it is that it would probably amount to compelling a firm to deal in virtually every scenario (so long as there are no technical reasons justifying the refusal, that is). In the same vein, the exploitation route would allow claimants and authorities to circumvent the need to establish the exclusionary effects of the refusal.

It is difficult to say whether the scope of Article 102 TFEU will expand in this direction. What one can certainly say is that it would not be surprising if the expansion occurred. The legal shift would bear all the hallmarks of the new EU competition law, and in particular the trend towards regulatory-like intervention and the blurring of lines between exploitation and exclusion.

As the year comes to a close, I thought I would share some of the research that got published this year. It was an unusual exercise, as every one of the pieces reminded me of what is yet to come — and which I hope I will be sharing in the coming months. The exercise also delivered a resolution: increase the output on the blog (easily achievable given this year’s meagre production).

This year saw the publication of two articles (frustratingly, other pieces got stuck in the peer review process much longer than anticipated):

• Remedies in EU Antitrust Law (see here). published in Open Access in the Journal of Competition Law and Economics.
• Judicial review in EU merger control: towards deference on issues of law? (see here), publised, also in Open Access, in European Law Open.

In addition, I am delighted to announce that ‘How Android Auto Reshapes the Law of Refusal to Deal (and What it Means in Practice)‘ (see here), which received the AdC Competition Policy Award, has been accepted for publication and is forthcoming in World Competition.

I am also delighted that two book chapters were published this year:

• State aid control beyond the EU: legal convergence and unilateral expansion (see here), jointly authored with Damien Neven, was published in the long-awaited second edition of the treatise and brainchild of Philipp Werner and Vincent Verouden, EU State Aid Control: Law and Economics.
• Hoffmann-La Roche and the Notion of Abuse, published in Paul Craig and Robert Schütze (eds), published in Landmark Cases in EU Law, Volume 2. This piece was an interesting exercise in legal history, which got me looking back at the early days of competition law enforcement in Europe. The volume as a whole is really interesting. I will be sharing it on ssrn as soon as the embargo is lifted.

As mentioned the other days, morever, Luca Prete and Leila Rezki have put together a Liber Amicorum in Honour of Nils Wahl, to which I proudly contributed and which is available here.

Here’s wishing you a wonderful 2026!

I have just uploaded on ssrn (see here) a paper entitled ‘The Categorisation of Practices in EU Competition Law‘. It will be coming out in the forthcoming Liber Amicorum in Honour of Nils Wahl, jointly edited by Luca Prete and Leila Rezki.

As I explain, the definition of legal categories is one of the core functions that the EU courts are asked perform in the area of competition law (and, dare I add, in EU law at large). Wouter Wils explained a while ago that ‘all human thinking‘ involves the use of categories. It is therefore an illusion to think that the law can be administered without resorting to them. What matters, Wouter rightly added, is not whether categories exist, but whether those used are sound.

The paper explains the fundamental role of categories in a field, like EU competition law, that revolves around open-textured concepts such as ‘restriction of competition’ or ‘abuse’. It also discusses, second, the main approaches that might be followed to the characterisation of conduct. Finally, it provides an overview of the techniques that the Court of Justice has followed over the years to define and refine categories.

It would be great to get your thoughts on the piece. And keep an eye on the launch of the book!

The meaning and scope of Article 101(1) TFEU is much clearer than it was just a decade ago (see here for a discussion). This relieving reality is in no small part a consequence of the steady stream of cases dealing with pay-for-delay practices in the pharma sector.

Rulings addressing pay-for-delay issues (and variations thereof) have provided the core test to establish whether a practice amounts to a restriction by object (since Generics, the issue is whether the only plausible explanation for it is the restriction of competition) and emphasise that, as far as Article 101 TFEU is concerned, anticompetitive effects will be assessed against the relevant counterfactual (a point on which I elaborated here).

More generally, the pay-for-delay saga illustrates that context is everything: as recently explained by Advocate General Emiliou in Tondela (here), a restriction of competition can never be established in the abstract. Whether it is a by object or by effect infringement, it is necessary to consider the circumstances surrounding the agreement.

One could have thought that these cases could not provide more clarity on the analytical framework to assess restrictions of competition. The Court judgment in Teva, however, has showed that the saga still has much potential to shed light on Article 101(1) TFEU.

Assessing the object of a practice in its context is not the same as evaluating its effects

The ruling in Teva is particularly helpful in that it engages with a line of reasoning that has become commonplace in recent years. According to some commentators, the contextual assessment of restrictions by object is problematic insofar as it blurs the line between object and effect.

I have never been persuaded by that criticism, which does not reflect the reality of the case law. Assessing the object (the rationale, that is) of a practice in the relevant economic and legal context is fundamentally different from establishing its actual or potential effects in the market (or markets) in which it is implemented.

It is certainly true that, once the anticompetitive object of an agreement is established, it is not necessary to ascertain its effects (something that the Court briefly reminds in Teva, paras 121-124, when dealing with the second plea raised by the appellants). It does not follow from this fact, however, that a contextual analysis is not required.

More generally, nowhere is it written that proving the object of an agreement should ignore the surrounding circumstances, or that it should be easy or straightforward to do so. The idea that the ‘object’ category is something of a ‘shortcut’ is foreign to EU law. It is borrowed from US antitrust law, which is fundamentally different from its European counterparts in more ways than one.

The appellants in Teva sought to advance a variation on this argument. They claimed, more precisely, that the General Court had conflated object and effect by, inter alia, introducing a counterfactual assessment, thereby making it impossible for the parties to the agreement to discharge their burden of proof.

The Court summarily dismissed these claims. When ascertaining the rationale, or object, of an agreement, it makes sense to consider the set of incentives it creates (paras 77 and 78). I would say more: how a practice influences the incentives of the parties will often provide a vital clue about what it seeks, objectively speaking, to achieve. That assessment does not entail an evaluation of the effects of the agreement in any way, contrary to what the parties claimed.

In the same vein, the Court concludes, the General Court did not require the parties to show what would have happened in the absent of the agreement. The relevant question remained that of whether the contentious transactions could be plausibly explained other than as a means to restrict competition (para 100). To the extent that it did, the General Court’s assessment remained firmly within the boundaries defined by Generics.

The unit of analysis is the clause, but context matters

The preceding case law had been clear in stating that the relevant unit of analysis is the clause, not the agreement as a whole (and this includes the crucial issue of severability). Examples abound. Arguably the canonical one is Pronuptia, where the Court distinguishes between the clauses that are ancillary to the operation of the franchise and those that are not (which, moreover, were found to restrict competition by their very nature).

The Court’s long-standing position is the only reasonable one. It would be too easy for the parties to avoid a finding of a ‘by object’ infringement if they were allowed to invoke the non-restrictive nature of the agreement taken as a whole. In effect, it would empty the ‘object’ category of its substance. For instance, a standard-setting agreement could be used as a vehicle to conceal a cartel agreement.

What Servier pointed out, and the Court mentions again in Teva (para 68) is that the analysis of the clauses cannot be undertaken in isolation. It may be the case that the clauses, taken together, are different aspects of a common strategy. If that is true, it is only reasonable that they are analysed together. By the same token, it is futile for the parties to try and argue that a given clause, examined in isolation, is not restrictive by its very nature. Such an interpretation would, again, entail a fundamental misunderstanding of the nature of the analysis under Article 101(1) TFEU.

At the end of the day, the Court’s clarification in Servier and Teva is another way of saying that context is everything at the ‘by object’ stage. The contextual analysis must by definition commence by looking at the agreement of which the clauses are a part. It is the first starting point, before shifting to factors such as the nature of the products or the dynamics of competition within the sector. The first step of the contextual assessment is likely to provide valuable hints about their nature.

Consider the example of price-fixing. There is a difference between a ‘naked’ price-fixing clause and one that is inserted in an agreement that pursues a non-restrictive aim. For instance, the Commission has long held (and rightly so, since the Court itself held just as much in Gottrup-Klim) that price-fixing clauses in the context of a joint purchasing agreement are not necessarily restrictive by object (see here). It has recently provided a good example that applies to SEP licensing.

The Commission Decision in Google Ad Tech exemplifies, better than any other, the shift in the centre of gravity of competition law enforcement towards remedies (see here for a more extensive discussion of this point). It also shows that remedy design is probably the single most important aspect to address in the ongoing reform of Regulation 1/2003.

In the current economic and technological landscape, a finding of infringement no longer marks the end of a case. This milestone is, as they say, just the end of the beginning. The effectiveness of enforcement, in digital and others markets, depends less on establishing an abuse of dominance than it does on ensuring that the infringement is brought to an end by means of a workable and well-crafted remedy.

Towards a structured framework for the design and implementation of remedies

The approach that the Commission is following in Google AdTech is a (welcome) innovation against this background. It has implemented what is, in effect, an ad hoc procedure to define the obligations with which the firm must comply. As this post is being finished, the Commission and Google are in the midst of conversations in the context of this informal procedure (see here for a recent update on the matter).

The aftermath of the decision suggests that the Commission acknowledges that merely ordering a firm to bring the infringement to an effective end no longer does the trick, and neither does a ‘principles-based’ approach that requires an undertaking to abide by an open-textured standard (such as non-discrimination). As I have written elsewhere, the complexity of remedies cannot be wished away by authorities.

Even if only de facto for the time being, Google AdTech suggests that the design and implementation of the remedy will, from now on, follow a structured framework that is independent from the finding of infringement.

Pursuant to the Automec doctrine, this ad hoc procedure gives the firm the chance to come up with a proposal to bring the infringement to an effective end. On the other hand, it allows the Commission to evaluate whether the undertaking’s proposal indeed addresses the concerns identified in the decision.

Where necessary (and this is the crucial third step), the Commission will specify the obligations with which the firm must comply (just like it does in the context of a commitments procedure). On this point, it has clearly signalled that it may adopt structural measures breaking up Google’s adtech business.

What matters, irrespective of the nature of the remedy, is that the duties imposed upon the firm are spelled out clearly and in sufficient detail to ensure immediate and full compliance.

The need for a structured framework: the experience of Google Shopping

The experience of the past few years shows that a framework for the specification of remedies is indispensable for the effective operation of the competition law regime. Delegating the design and/or implementation of remedies to firms is in nobody’s interest, neither that of the competition authority, nor the addressees of the decision nor of third parties potentially benefitting from intervention.

More importantly, it is now difficult to dispute that the ‘principles-based’ approach to remedial action has failed. The uncertainty to which it gives rise has major practical consequences, which were dramatically exposed last week (see here).

Google Shopping is a case where controversy around compliance with the remedy never really went away. Whether or not the measures implemented by the firm back in 2017 brought the infringement effectively to an end has always been (and remains) contentious.

It is against this background of perpetual limbo that a German court awarded damages to two of Google’s rivals on the market for price comparison last week. Crucially, the damages award extends to the period following the adoption of the Commission decision, and is therefore based on the assumption that the firm never actually complied with the principles-based remedy imposed.

Even though the Commission never opened non-compliance proceedings against Google, the absence of a decision formally and positively declaring that the infringement had been brought to an effective end paved the way for the award of damages beyond 2017.

This judgment suggests that any remedy implemented following a ‘principles-based’ approach is potentially vulnerable to challenge and, as such, a source of legal and non-legal uncertainty.

Moving forward: codifying the framework

The Commission’s introduction of an ad hoc remedial framework in Google AdTech is therefore the right way forward. Acknowledging the realities and demands of contemporary industries is not just wise but indispensable. It would be desirable if the ad hoc remedies procedure were improved along three dimensions.

First, transparency. The exchanges between the firms and the authority should benefit from similar levels of openness as those observed in the context of commitment procedures. What the firm proposes as per Automec and the authority’s assessment of the proposal should not happen behind closed doors.

Second, third-party involvement. One of the advantages of the commitments procedure within the meaning of Article 9 of Regulation 1/2003 is that obligations are ‘market-tested‘ with stakeholders. A structured procedure would give these stakeholders a meaningful chance to provide input about the proposals and to express their concerns, if any.

Third, codification. Ideally, this procedure would be enshrined in a reformed Regulation 1/2003. As a second best, it could at least be codified at the internal level (the 2019 version of the Manual of Procedure was, say, succinct about the issue or remedies and would definitely benefit from a revamp and further detail).

An editorial of mine has recently come out with the Journal of European Competition Law & Practice (JECLAP). It is available here. The point it makes is simple: the first instinct of illiberal regimes is arguably to disregard the law and ignore expertise (whether it relates to climate change, vaccine safety or any other issue).

I argue, against this background, that, if competition law is to play a role in the protection of democracy and its institutions, it must resist the temptation to abandon law and expertise in the name of expediency or effectiveness. Such a move is not only likely to weaken policy-making and affect its quality, but would inevitably backfire over the long run.

I look forward to your thoughts on it. I explored the same ideas in the keynote lecture delivered in honour of Heike Schweitzer a few months ago (see here).

I have uploaded on ssrn a new paper (see here) a paper that discusses the judgment of the Court of Justice in Android Auto as well as its substantive and institutional implications. I am pleased that the paper received yesterday the AdC Competition Policy Award, organised every year by the Portuguese Competition Authority.

The way in which Android Auto has changed the law of refusal to deal, it seems to me, may not have been fully appreciated. To make this point, it is sufficient to apply the Court’s reasoning in Android Auto to the facts at stake in Bronner. The latter would have been decided differently; evidence of indispensability would not have been required to establish an abuse.

As I explain in the paper, this difference is a function of the way in which Android Auto (re)interprets the indispensability condition. In Bronner and Magill, whether or not the dominant firm had kept the assets ‘for the needs of its own business‘ was assessed by reference to the relevant market concerned by the refusal. It was therefore irrelevant that the TV channels in Magill were licensing their copyright to newspapers (but not weekly magazines) or that Mediaprint was printing and distributing another publication in Bronner at the time of the facts.

The analytical approach followed in Android Auto construes the indispensability condition differently. According to the new doctrine, where the dominant firm is dealing with a third party in market A, it can no longer invoke indispensability in relation to a refusal concerning market B. For the same reason, the judgment has significantly reduced the scope of the refusal to deal case law.

This transformation has obvious implications for digital markets. Dominant players in these markets often run systems that are partially open and partially closed. This said, the criteria followed by the Court suggest that the ruling is likely to have an impact in other industries and markets.

In any event, the implications of the judgment go beyond the shrinking of the refusal to deal doctrines. Android Auto allows for intervention that is not confined to a mere duty to deal. A dominant firm may indeed be required not just to share an input or infrastructure with third parties, but to redesign its assets by taking into account the demands of the said third parties. In this sense, the operation of the infrastructure becomes a cooperative venture.

I look forward to your comments on the paper. And thanks again to the AdC!