Chillin'Competition

The Commission usually takes some time to publish its decisions. This is not necessarily a bad thing, at least for the purposes of this blog. We now have the chance to discuss a few decisions that have (finally) come out, and complement (or complete) our first thoughts on them.

I will start with ISU, about which I wrote earlier this year. The non-confidential version of the decision can be found here.

In my previous post, I asked myself whether the practice at stake was really a restriction by object. You will remember that the Commission took issue with a set of rules adopted by the International Skating Union. These rules sought to constrain athletes’ ability to take part in competitions run by rival organisations.

The Commission concluded that the rules, which are (in essence) a vertical restraint providing for a non-compete obligation, amount to a ‘by object’ infringement. Now we know the reasoning behind this conclusion.

There is something remarkable about the decision. If you take a look at it, you will realise that the Commission does not seem to incorporate the lessons of the most recent – and directly relevant – Court rulings, namely Cartes Bancaires and Maxima Latvija.

To prove my point, I propose a simple exercise: apply to the facts underpinning Cartes Bancaires and Maxima Latvija the reasoning found in ISU. You will come (inevitably) to the conclusion that the practices at stake in these two cases were restrictive by object.

In ISU, the Commission concludes that the eligibility rules are a ‘by object’ infringement for three main reasons:

• The objective and subjective purpose of the rules was to preclude other organisations from running competing events.
• The International Skating Union sought to protect its own economic interest through the rules.
• The eligibility rules were not related to a legitimate sporting objective.

Remember Maxima Latvija? In that case, the ‘anchor tenant’ of a shopping centre was given the power to veto the renting of other premises in the mall to third parties. The objective purpose of such a rule is clearly to avoid competition – and the Court ruling is based on this premise. In this sense, the case is no different from ISU.

And the anchor tenant, by restricting competition to itself and limiting the shopping centre’s freedom of action, was certainly trying to protect its own commercial interests (by the way: which firm doesn’t?).

In spite of the above, the Court ruled in Maxima Latvija that the practice was not restrictive by object.

Remember Cartes Bancaires? Essentially, the contentious rules sought to penalise one category of competitors. The objective purpose of these rules was to hinder these firms’ ability to compete. What is more, there was direct evidence in the case suggesting that the subjective intent of the rules was indeed to restrict competition and thus to protect the economic interest of another category of firms.

If one were to follow the reasoning in ISU, these two factors would be sufficient to conclude that the practice was restrictive by object. But you all know that the Court came to the opposite conclusion in Cartes Bancaires.

Cartes Bancaires is important for another major reason. The Court made it quite clear that a practice that seeks to address a genuine free-riding concern is not a ‘by object’ infringement. In other words: the fact that some firms seek to protect their economic interests is not in itself an issue (again, which firm does not seek to advance its economic interest by means of an agreement within the meaning of Article 101 TFEU?). The issue is instead whether the firms seek to address a market failure or simply extract rents (as in a cartel agreement).

The Court concluded that the measure was plausibly pro-competitive (and thus not ‘by object’) in Cartes Bancaires. And the free riding argument is also compelling in ISU, as I mentioned last time (but the issue does not seem to be given the relevance it deserves in the decision).

Against this background, the question that I find intriguing is why the Commission has not followed the case law on restrictions by object in ISU.

This question is intriguing in this particular case because establishing the restricting effects of the eligibility rules was a ‘home run’ for the Commission. Given the position of the International Skating Union, it could not have been much easier to conduct an effects analysis (as the Commission does in the case).

Had the Commission followed the ‘by effect’ route alone, the case would have been entirely uncontroversial. I would say more: the analysis of the Commission in the ‘by effect’ section shows that the case makes enormous sense from a prioritisation perspective too.

Why, if it was not at all necessary (and was in fact potentially counterproductive), did the Commission insist on qualifying the rules as a ‘by object’ infringement in ISU?

The most convincing explanation is that the Commission, as a repeat player, is interested not only in reaching the desired outcome in individual cases but in shaping the law in a particular way. In this sense, ISU provided an excellent opportunity to advance its interpretation of the notion of restriction by object in the wake of Cartes Bancaires.

I have to say I am not particularly surprised by this. I have spent the past couple of years reading pretty much every Commission decision, and this is a consistent pattern of behaviour across the board (Article 101 TFEU, Article 102 TFEU and merger control).

Examples? Just think of how the Commission interpreted Delimitis in Scholler and Langnese-Iglo (Valentine Korah, in her unique style, wrote at the time – mid-1990s – that Delimitis appeared ‘not to have been read’).

More examples? The Court emphasised, from the outset, that ‘competition’ for the purposes of Articles 101 and 102 TFEU means ‘competition that would have existed in the absence of the practice’. However, the Commission failed to consider the counterfactual in several landmark decisions that were annulled as a result.

In fact, if you read ISU, you will identify several controversial statements in this regard. For instance, the Commission asserts that the members of the International Skating Union – that is, the national associations – are potential competitors. However, this remarkable statement is not substantiated – as if the lessons from European Night Services, CISAC or E.On Ruhrgas (‘real, concrete possibilities’) had not been learnt.

Conclusions?

What do we make of the failure to incorporate some of the crucial insights from Cartes Bancaires? It is neither good nor bad. It is a reality and a feature of the EU competition law system we have to acknowledge and with which we have to live.

We have acquired sufficient experience over the years to know that the Commission is likely to behave in this way – across provisions and over the years. I would say that, first and foremost, it is useful for the authority to be aware of this reality, so as to anticipate when and why administrative decisions are more likely to be annulled.

It has taken me a while to realise the significance of the recent Court judgment in Maxima Latvija. On its face, it does not seem to add much to what we know about agreements that restrict competition by object. This impression is probably due to the fact that the Court is not very explicit about why it ruled the way it did. If one takes into account what was not said, but is implicit, together with what the Court did actually say, it is possible to draw some valuable lessons.

What the Court said in Maxima Latvija

The necessary and sufficient factors to establish a ‘by object’ restriction

Paragraphs 22 and 23 are very much in line with previous case law. It is clear from the latter that the question of whether an agreement restricts competition by object is established in light of the content of the agreement and the context of which it is part. This is nothing new, but it is valuable that the Court confirms that these are the necessary and sufficient factors to evaluate whether a set of restraints is incompatible, by its very nature, with Article 101(1) TFEU.

The ‘by object’ category is not a presumption of anticompetitive effects

I repeat myself a lot, but it cannot be emphasised enough that the ‘by object’ category does not encapsulate a presumption of anticompetitive effects. When commenting on Bananas, I mentioned that an agreement such as an exchange of information can be found to restrict competition by object irrespective of whether there is evidence of its impact on competition. As Bananas itself shows, an exchange of information may be prohibited by its very nature even if it is not particularly likely to have a significant impact on prices.

Maxima Latvija is the mirror example. The case is about an obligation included in an agreement between the ‘anchor tenant’ of a shopping mall and the lessor. I understand from the ruling that the ‘anchor tenant’ had to give its consent to the letting of other premises to third parties. This restraint would work as an exclusivity obligation, in the sense that the tenant is given the right to oppose the letting of premises to competing supermarket chains.

The Court concedes that this contractual obligation is capable of having (‘could potentially have’) an anticompetitive effect. However, this fact alone is insufficient to establish that it is restrictive by its very nature. In other words, a restriction by object does not exist merely because an agreement can be presumed to have anticompetitive effects (and no, the Court never said the opposite in T-Mobile).

This is something that the Court has always held, but tends to be forgotten. Selective distribution agreements reduce substantially the ability and incentive of retailers to engage in price competition. It is in fact safe to presume that selective distribution softens price competition. However, the Court has always ruled, from Metro I to Pierre Fabre, that this fact alone is insufficient to establish a restriction of competition. As is well known, selective distribution networks fall outside the scope of Article 101(1) TFEU altogether in some circumstances.

Maxima Latvija is particularly interesting because there is recent empirical evidence suggesting that exclusivity obligations included in agreements between shopping malls and tenants have anticompetitive effects. Itai Ater comes to this conclusion in an article published earlier this year in the Journal of Economics & Management Strategy. So there it is: an agreement that is known to lead to higher prices has not been found to restrict competition by object.

What the Court did not say in Maxima Latvija

Why is an agreement not found to restrict competition by object if it is known – and can be expected – to have negative effects on some parameters of competition? The Court did not say much about it in Maxima Latvija. Actually, the Court does not really explain why the agreement is not restrictive by object. The good news is that past case law is very explicit about these two questions.

An agreement does not qualify as a ‘by object’ restriction if it has redeeming virtues that compensate for the expected negative effects. Think of Metro II. True, the Court held in that case, selective distribution softens price competition. But it benefits consumers in other ways. As a result, it is not restrictive by object.

The reasons why the Court came to the same conclusion in Maxima Latvija are probably not very different. The abundant references to Delimitis suggest that the contentious clauses were understood to be functionally equivalent to exclusive dealing obligations. And the Court has already held that exclusive dealing has redeeming virtues that are in the interest of both parties to the agreement and thus of consumers.

Interestingly, Itai Ater mentions in his paper, referred to above, that exclusivity obligations agreed upon between ‘anchor tenants’ and shopping malls can yield efficiencies. For instance (and this says much about his academic integrity), the author points out that his empirical analysis may have failed to capture that an exclusivity obligation between an ‘anchor tenant’ and a shopping mall may induce relationship-specific investments or may attract such investments at an early stage.

Having dealt with substantive questions, primarily relating to the notion of restriction by object (see here, here and here), it makes sense to continue this series by switching to a received idea of an institutional nature.

When presenting my research, it is not unusual for me to hear that the Commission ‘is a (very) risk-averse institution’, which, accordingly, would only intervene in clear-cut cases and this after choosing them carefully (perhaps too much).

This is the opposite of what one finds when examining systematically the Commission’s behaviour (which happens to be one of the research areas on which I focus).

And before I move on to explain why, allow me to clarify that I am convinced that, on balance, it is emphatically a good thing that a competition authority takes risks, and takes them often.

I struggle to see how the public interest would be advanced if the Commission only pursued the safest cases and were reluctant to explore new ideas or challenge existing doctrines.

On the other hand, one should be aware of the consequences of a relatively risk-prone attitude of the Commission. The higher the willingness to take risks, the more likely the errors (in law and in fact). And, by the same token, the more important full judicial review becomes.

Examples of the risk-prone attitude of the Commission

If one reviews the Commission’s administrative practice since the 1960s, it seems pretty clear that it has systematically explored the outer boundaries of the case law, sometimes departing from it (by applying a different doctrine, introducing a new one or by explaining why it is not applicable in the context of a particular case).

Off the top of my head, I can think, inter alia, of the following examples (I do not mention areas like State aid, but I have been able to identify a similar pattern there too):

Restrictions by object after Regulation 1/2003

Restrictions by object dominate the enforcement of Article 101 TFEU following the adoption of Regulation 1/2003. This is not only due to the prioritisation of clear-cut infringements by the Commission.

The Commission appears to have chosen to embrace a risky approach that occasionally departs from the logic of the case law. Cartes Bancaires is there for all to see. Even more interesting is ISU, where (as I explained here), it chose not to follow the most recent case law on the issue (Cartes Bancaires itself and Maxima Latvija).

And these are not the only examples. I have devoted many posts to explain why cases like Pay-TV and Lundbeck are controversial from this perspective (Lundbeck is interesting in that there was documentary evidence, back from 2004, showing that the Commission was aware that the practices were in a ‘grey area’, and thus of the risks involved in pursuing the case).

The analysis of effects under Articles 101 and 102 TFEU

Examining the effects of practices (whether under Articles 101 or 102 TFEU) can mean many different things. The analysis may vary greatly depending on what we mean by effect and on the degree of probability that is deemed sufficient to trigger intervention.

And I devoted my Chillin’ talk last year to the question. It is now possible to discern a definition and a methodology from the case law.

The important point here is that the Commission has showed a tendency, over the years, to depart from the notion of effect as defined by the EU courts.

The administrative practice that followed Delimitis is perhaps the best example in this sense. That landmark judgment clarified (once more) that a restriction in a firm’s freedom of action does not amount, in and of itself, to an anticompetitive effect. In addition, it laid down a test that revolved around foreclosure (defined as the ability of a rival to enter the relevant market).

In Langnese-Iglo and Scholler, the Commission did not follow Delimitis (as noted by commentators at the time). It chose instead to take the risk of committing to its traditional approach (under which a restriction in a firm’s freedom of action is sufficient to establish an anticompetitive effect).

One can identify a similar pattern in recent Article 102 TFEU decisions, of which Servier is a great example. More on this soon: it is a fascinating question that keeps me busy.

Refusals to deal

There is much talk these days about interim measures. It is said that, since IMS Health, the Commission has never adopted a decision of this kind. But it is not always explained why the decision was quashed by the (then) Court of First Instance.

IMS Health was just an interim measures decision, but the interim relief sought was far-reaching by any standard (it imposed an obligation that effectively altered a firm’s business model).

One could argue that the context was not ideal for the Commission to take the risk of embracing a heterodox reading of Magill, but the authority chose to give it a try.

The interim measures decision argued, controversially, that the Magill conditions were alternative, and not cumulative, and thus that an obligation to license could be imposed even absent evidence that a refusal would prevent the emergence of a new product.

No surprise that the Court of First Instance took the view that imposing a duty to license on an interim basis should not be based on a peculiar reading of the relevant case law.

The Commission took similar risks again in Microsoft. The Court of First Instance acknowledged in its judgment that the decision had not established an abuse in light of the conditions set out in Magill (in particular, the Commission had not attempted to show that the refusal prevented the emergence of a new product). This second time, however, the risk taken by the authority paid off.

Rebates

I could go on for a while, but I will finish this overview with the case law on rebates. As we all know, the Court declared in Hoffmann-La Roche that rebates conditional upon exclusivity were abusive.

In the years that followed, the Commission progressively expanded the scope of the prohibition rule to encompass target rebates (in Michelin I and British Airways) and even standardised volume rebates (in Michelin II).

This risky approach was successful in the sense that the doctrine encompassed a wider range of potentially harmful conduct. On the other hand, it created a sense of legal uncertainty that prompted the Commission to review its approach to the enforcement of Article 102 TFEU.

The importance of judicial review

The willingness to take risk by stretching, or departing from, existing doctrines is not without consequences.

The interpretation of the law may become less consistent and intervention less predictable as a result – past cases would not provide reliable guidance on the outcome of future investigations and conflicting lines of case law may emerge.

The complex reality of Article 102 TFEU – and the subsequent attempt by the Commission to remedy the consequences of its own approach, mentioned above – speaks for itself. Perhaps the risks were justified to avert potentially harmful conduct by powerful undertakings, but the cumulative effect of individual decisions resulted in the absence of clear and meaningful boundaries between abusive and lawful conduct.

Against this background, judicial review is of paramount importance to preserve the general interest (including legal certainty) and explore why and when it is justified to depart from, or ignore, the prevailing legal doctrines.

I do not believe the law is cast in stone and should never be revisited. And I am sure none of our readers believe the role of judges is to ensure the law never changes.

The point is instead that existing doctrines have a point and a rationale – they are typically the product of careful thought by courts. Accordingly, the case law cannot be dismissed or stretched without an appropriate explanation (which, in turn, may or may not persuade the review courts).

The CMA found the ideal timing for the publication of its decision in the Impulse Ice Cream case: it came out when ice cream consumption peaks (mid-August) and right before the Intel judgment (when, incidentally and among other things, ice cream consumption starts to decline and coffee consumption starts to go up).

Following its investigation of the question, the CMA concluded that there were no grounds for action against Unilever under Article 102 TFEU and its national equivalent. When I write that the CMA shows the way with this decision, it is not because of the outcome (I have already explained that the outcome of individual cases does not matter to me).

What I find interesting is how the authority approached a potential infringement both from a policy and a substantive perspective. There are many lessons to draw from both:

• From a policy-making standpoint, the decision is a valuable reminder that ‘no infringement’ decisions and ‘no grounds for action’ decisions are essential in any competition law system. What an authority does not do is as important as what it does.
• From a substantive standpoint, the CMA shows that an effect-based approach can be conducted effectively and that it is a valuable means to make the best use of (and not to drain, as some like to claim) an authority’s resources.

Allow me to discuss them in more detail.

The importance of ‘no infringement’ and ‘no grounds for action’ decisions in policy-making: we should have many more of them

We live in strange times. Some people like to believe that competition law (like rock ‘n’ roll) can save the world. Some people think that every concern justifies intervention.

Competition law is valuable for society at large because of the lessons learned over decades of enforcement. Experience shows that, while competition law is indispensable in a social market economy, it cannot save the world (this is also true of rock ‘n’ roll, by the way). Experience also shows that sometimes intervention may not achieve anything meaningful and can sometimes be counterproductive (in the sense that it may harm the competitive process and ultimately consumers).

It is very important that competition authorities regularly convey these two messages to the business community and the wider public. Explaining why not everything is a competition law concern, and that intervention is sometimes counterproductive is likely to reduce the appetite for sweeping (and typically unreflective) regulation. My sense is that it will also protect competition authorities against demands for tailor-made action (which sometimes requires a fresh, untested and/or incoherent theory of harm).

The best instruments to show the world that competition law cannot and should not try to get to El Dorado are ‘no infringement’ and/or ‘no grounds for action’ decisions. In this regard, Impulse Ice Cream is a model: it is succinct yet sufficiently detailed to understand why, in light of the case law and administrative practice of the European Commission, Unilever’s practices were unlikely to have exclusionary effects.

The effects-based approach is administrable and can save resources to an authority

Those who oppose the effects-based approach like to claim that it is unmanageable: their hyperbolic stories describe armies of economists engaged in impossibly complex calculations and unable to come to meaningful conclusions. It would no longer be possible, the argument goes, to know whether something is lawful or unlawful, and dominant companies would be able to exploit this uncertainty to avoid intervention. And, if there was any doubt, courts would be overwhelmed with numbers and welfare estimations.

The CMA’s decision shows that these claims do not stand up to serious scrutiny (to tell the truth, I suspect that many commentators do not even take the claims seriously; it is just an irresistible rhetorical point). In about 4 pages, the authority applies the framework laid down in Post Danmark II and concludes that there was no point in keeping the case open and thus devoting some of its limited resources to it.

I will take just one of the practices examined by the authority to show how the effects-based approach works in the real world (spoiler: it does not involve infinite rows of sleepless econometricians). The CMA raised concerns, inter alia, about some ‘large’ package deals offered by Unilever to supermarkets (the decision refers to ‘buy 8 [packages] get 4 free’ or ‘buy 12 get 6 free’).

Unilever is probably dominant on the relevant market, and some of its products are must-haves. In spite of these factors, the CMA concluded that exclusionary effects were unlikely to result from the large package offers. As explained in the decision, the firm gave these offers in February or March, where ice cream consumption is typically low, and were available for just one month. In addition, the purchasing decisions made during winter and early spring were deemed unlikely to affect purchasing decisions during the summer months; instead, such decisions were found to be determined by other considerations.

These conclusions are not only sensible, but also in line with the case law. Needless to say, nothing of what the CMA does is out of the reach of a generalist court. Far from that. A court applying Delimitis or Maxima Latvija to a dispute is more than able to consider the same factors in an Article 102 TFEU case. There seems to be no reason why what is considered acceptable and administrable in the context of Article 101 TFEU is not acceptable and administrable in abuse cases. This is something that is implicit in some of the most recent rulings of the Court.

Any lessons for Intel?

Before I forget: Impulse Ice Cream is relevant for the (never-ending) discussions around Intel in one important respect. The case provides yet another example that the tripartite division between quantity rebates, loyalty rebates and ‘third category’ rebates is, insofar as it exists, neither meaningful nor workable.

Why do I say that? As I started reading the decision, I thought ‘it looks to me like a set of quantity rebates, as the discounts are given with respect to each of the orders; this would make them prima facie lawful’. The officials at the CMA (who obviously have more information) characterised it instead as a ‘third category’ rebate scheme. As sensibly explained in the decision, the characterisation of the practices, in any event, matters less than their impact on the competitive process. This point is buried in a footnote, but it is key.

I will say more: the boundaries of each category are so blurred, and all rebate practices are so similar in their nature and potential effects, that I fail to see what the tripartite division achieves, other than creating confusion and opening the door to arbitrary decision-making. We need legal categories, that is out of the question, but legal categories need to make sense. We will get to know next week whether the Court agrees with me on this one!

Last Tuesday I intervened at the GCLC’s annual conference, focused on the notion of restriction of competition.

What follows is an abridged version of my intervention (that does not include some of the worst jokes). Pablo, who gave me significant input and appears as co-author in the slides, and myself will be writing a joint piece on the subject soon.

***

One remarkable thing that conferences organized by practitioners have with European Commission initiatives is that they all seem to be aimed at working less. We don’t want fines, we don’t want abuse of dominance cases, less focus on key industries, less interventionism, etc. And now we want to clarify what a restriction of competition is?  If competition law is fun (and if we all make a living) it is because it is open ended and evolutive. So in a way these attempts at clarification could be regarded either as a collective suicide or a collective attempt at output limitation [I illustrated the point with a gif of the collective suicide squad from “The Life of Brian” that some considered a bit gore…] Even Commissioner Vestager observed this yesterday.

My role in this collective suicide is to comment on what has been discussed these past couple of days and try to extract, from a legal viewpoint, common threads and lessons.

I had many comments on what has been discussed here, but no common thread. But then last night at the speakers dinner Masssimo Merola gave a brief speech and said that yesterday he had realized about both what he knew and what he did not know. I thought voilà, there’s my structure. So let’s focus on those two questions: what is it that we know, and what is it that we don’t know.

***

If 50 years into EU Competition policy and centuries after the common law doctrine of restraints of trade was developed a bunch of experts are getting together for two days to ask ourselves something as basic as what is a restriction of competition, an outside observer would immediately think that instead of a bunch of experts we’re a bunch of ignorants!

If we are here enquiring about the notion of restriction, that is because the wording of Articles 101(1) and 102 TFEU do not provide an operational test.

Indeed, if anything that limits competition or commercial freedom were to be legally considered as a restriction in the sense of 101(1), then everything would be caught by 101. As also noted by Sir C. Bellamy yesterday, restraint is the essence of every contract…. particularly marriage.

In my view, there is no point in trying to figure out what a restriction is by looking at the literal wording of Article 101(1) and 102 TFEU. It is a hopeless exercise to find an operational test from such vague provisions. They were intended to be open, adaptable, judge made. That is why I like to say that competition law is a distillation of common sense infused with mainstream economics.

It was said yesterday that there is no room in 101(1) for “consumer harm” and that perhaps we need to re-write article 101. But I think that provision was drafted the way it was precisely so that it would encompass everything: consumer harm and the contrary. It is not because there is no explicit reference in the letter of the provision to consumer harm that the case law cannot evolve in that direction, as it arguably has.

But do we really not know what a restriction is? If the conference has taught us anything is that the problem is not one of the concept in itself, but one of methodology, and even of terminology.

What is a restriction of competition in colloquial and in economic terms is perfectly clear (as clearly explained by Jorge Padilla): it’s a reduction of competitive constraints. The problems rather lie in how a legally relevant restriction is established (Do we look at the short or the long term? Process or outcome? How do we look at the positive effects of a restriction, or an apparent restriction? Can we tell a restriction following a quick look (by object) or do we have to examine in detail all circumstances relevant to its effects (by effect)?)

***

Instead of trying to draw lessons with a top-down approach, I propose to analyse what has been said in the conference in a different way: bottom-up.

Let’s focus then on what we know and on what that tells us:

What we know

If we look beyond case-specific often conflicting discussions on the notion of restriction and we pay attention to what the Court has actually clarified over the years, we believe it is fair to say that we actually know a great deal about the sort of conduct that amounts to a restriction of competition.

It is true that the case law is uneven, and that some issues are yet to be clarified, so let’s focus only in those areas where there really is a significant consensus.

• First, and most important of all, the case law is very useful to understand what a restriction of competition is not.

Read the rest of this entry »