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.