Chillin'Competition

There is so much about which I want to write that I should not indulge in commenting on Alfonso’s last post. And I really did not want to. But it is happening. It is not often the case that I choose the topics of my posts. They start writing themselves in my head and before I realise I find myself typing, as I am now. So blame the topic, not me. And blame my co-blogger for starting it all with his initial question 😉 . Rest assured that a couple of posts on information exchanges and geo-blocking will start writing themselves in my head soon (primarily because I have much to say about the two).

The fact that I disagree with Alfonso’s last post does not mean that it is not an entirely reasonable one. He says that he does not regard the case law on exclusive dealing and loyalty rebates as the ‘epitome of absurdity’. I do not think it is. It is actually very far from being so. The argument (at least mine) has never been that it is absurd (because it is not), but that it is difficult to reconcile with other aspects of the case law and as such a permanent challenge for the coherence of the competition law system as a whole. Why do we disagree, then? Let us take a look at some of his arguments.

Experience and economic analysis (the Cartes Bancaires test)

The lessons of experience: Alfonso is right to say that the case law assumes that exclusive dealing and loyalty rebates always have restrictive effects on competition. He refers to the Cartes Bancaires test but he does not discuss the lessons of experience, which are valuable ones. Experience shows that that rebate schemes rewarding loyalty do not necessarily prevent the rivals of a dominant firm from thriving or even increasing their market share. This is the fundamental lesson to draw from British Airways and Michelin II. If, as in these cases, there is cogent evidence showing that competition has not been restricted as a consequence of these practices, then it is safe to claim that negative effects cannot simply be assumed to be the necessary consequence of exclusive dealing and loyalty rebates. Unless, of course, one refers to a particular type of negative effects that cannot be seen or proved but is nevertheless there in some form, like a supernatural entity.

The lessons of economic analysis: In relation to the lessons to draw from economic analysis, Alfonso says that he does not know enough about it to conclude whether it suggests that the assumptions underlying the case law need to be refined. What can we conclude from (both theoretical and empirical) economic research? These lessons are elegantly summarised in the Commission Guidance and the EAGCP Report of 2005. I always recommend Massimo Motta’s careful dissection of Michelin II because it is a lucid and accessible piece addressing the question. Against this background, it seems safe to claim that it would be inappropriate to presume that dominant firms implement exclusive dealing and loyalty rebates with the aim of excluding competition. Economic research also shows that negative effects on competition cannot simply be assumed to result from these practices. This is, therefore, one of these areas where the experience of the case law and economic analysis go hand in hand.

I have written elsewhere that EU courts have consistently displayed remarkable intuition about the motives behind business practices. Delimitis is one of the rulings that is very much in line with formal economic analysis. In this remarkable judgment (discussed by Alexander Italianer in the speech he gave back in December), the ECJ concluded that exclusive dealing cannot be considered to be anticompetitive by its very nature (in other words, anticompetitive motivations cannot be presumed) and that its restrictive effects cannot simply be assumed to be the consequence of the fact that access to some outlets is foreclosed. The cumulative effects of similar arrangements may lead to such effects in a particular context, but this is something to be established on a case by case basis. I fail to see why the finding that exclusive analysis is not restrictive by its very nature would not be relevant in the context of Article 102 TFEU. This insight, which relates to the rationale behind the practice, seems to apply irrespective of the degree of market power enjoyed by a particular company.

The lessons of economic analysis have been embraced by EU courts. To name a few, they played a fundamental role in Cartes Bancaires (two-sided markets), in Airtours (in relation to collective dominance), in Tetra Laval (in relation to conglomerate mergers), or in Woodpulp II (concerning the notion of concerted practice). With these precedents in mind, the question arises of why the insights of professional economists would be ignored in relation to a given set of practices if they have informed the case law in other areas.

The ‘making life more difficult’ standard

Alfonso suggests that it is appropriate to prohibit prima facie under Article 102 TFEU any practices that are assumed to make competitors’ life more difficult. Again, this is a defensible position, and one that is very far from being the ‘epitome of absurdity’. That is not the issue. The issue is that practices that in theory always make competitors’ life more difficult are not always prohibited prima facie in the case law. Below-cost pricing by a dominant firm always makes rivals’ life more difficult, but it is not always prohibited as abusive under Article 102 TFEU (just take a look at AKZO and Post Danmark I). Similarly, the ECJ made it clear in Deutsche Telekom and TeliaSonera that it cannot simply be assumed that a margin squeeze always makes competitors’ life more difficult. It would be necessary to establish the likely effects of the practice on rivals’ ability to trade in the retail market in the context of a particular case. In this regard, the ECJ did not follow the Commission decision in Deutsche Telekom.

As can be seen, the real issue here is that Article 102 TFEU case law does not follow a uniform approach, not that the approach followed in relation to exclusive dealing and loyalty rebates is the ‘epitome of absurdity’. The fundamental claim I made in my paper is that the case law, as it stands, does not treat practices that are comparable in their nature and likely effects in the same way. It is, in other words, a matter of legal coherence. In my view, the most sensible way to address the observed tension in the case law is to embrace the lessons of experience and economic analysis across the board.

Administrability vs ‘hordes of public officials, and countless hours of lawyers and economic consultants’

The case law on exclusive dealing and rebates is defended because it is easier to administer than the alternative, which would require ‘hordes of public officials, and countless hours of lawyers and economic consultants’. The administrability of legal tests is an important factor for any lawyer and for any economist who is aware of the demands of legal practice. If I disagree with this point is because, first, he seems to assume that standards to establish, on a case-by-case basis, the likely impact of a practice are very difficult to administer and impossibly demanding in terms of resources. This is clearly not the case.

My second objection to his argument about administrability relates to the coherence of the competition law system as a whole. If Alfonso’s argument applied in the context of merger control, then all conglomerate mergers involving a dominant firm with the ability to foreclose competition (as in, say, Microsoft/Skype) would be prima facie prohibited irrespective of their likely effects on competition. The fact is that the likely effects of conglomerate (and horizontal, and vertical) mergers are routinely examined in concreto. Nobody seems to object to it as an impracticable model or one imposing an unjustified drain on the resources of the Commission. Why, then, would it be a problem for the Commission to establish the likely effects of potentially abusive practices in accordance with the principles that the authority has committed to follow in the Guidance?

Alternatives to exclusive dealing

Finally, Alfonso takes the view that the prima facie prohibition of exclusive dealing and loyalty rebates is not an issue insofar as the same objectives of these practices can be ‘achieved through other, perfectly legitimate and less risky, means’. Is that really the case? What are these other perfectly legitimate means? Tying? Clearly not, as it is also prohibited prima facie. Are quantity rebates perfectly legitimate? Maybe, or maybe not. Possibly, maybe sometimes. As I argued earlier this month, it is far from clear when and why quantity rebates are acceptable under the case law. The pending ruling in Post Danmark II shows that, indeed, it is difficult to say that this route is really less risky for dominant firms.

On my response

As far as I am concerned, this will be the last post on this most interesting (at least for me) exchange. I know, however, that some people did not like the answer I gave. The only way forward of which I can think is to create a new section in the blog, called ‘Choose your favourite Article 102 TFEU decision’. Readers are all invited to send via email the name of a Commission decision of which they are particularly fond (whatever that means). And I hereby commit to discuss its legal, policy and economic aspects 😉 .

I laughed when Alfonso asked whether there is an Article 102 TFEU decision adopted by the Commission that I like. I laughed in part because I understand why someone might come up with such a question, but also because the image of an academic curmudgeon grumbling from an ivory tower is undeniably comical. After the healthy dose of laughter, I realised the question was more interesting than I initially thought. As is often the case with jokes, it is one that raises pretty fundamental issues. The short answer, and the sad truth, is that I do not see Commission decisions that way. I will elaborate a bit to make sure Alfonso feels compelled to answer the question that I have for him and that you will find at the bottom of the post.

I am not interested in the outcome of individual cases. Really not. I am not even sure that it is possible or meaningful to like or dislike a particular decision examined in isolation. The many posts I have written in relation to the Google investigation would have been written in any other context giving rise to the same issues. When I wrote about the GC ruling in Intel, I never sought to argue that the company had not breached Article 102 TFEU. I am interested in the way in which the law is shaped and evolves over time, and not so much in the ‘who wins’ or ‘who loses’ a specific case. These questions are necessarily relevant for civil servants and practitioners, but matter little to me.

Just take a look at what Alfonso and I wrote about Intel to illustrate the above. He presents the ruling as a ‘victory’ for the Commission and explains how the General Court carefully crafted the judgment in anticipation of the appeal. My entry, on the other hand, was devoted to the key assumption underpinning 35 years of case law on exclusive dealing and loyalty rebates. After reading my paper on these issues, a friend told me, only half-jokingly, that the title was misleading, as it was not really about Intel, but about everything. Well, exactly.

The question is also valuable in that it provides an opportunity to emphasise the difference – that can never be emphasised enough – between law and policy-making. A case may well be sound from a policy standpoint, but this fact alone does not mean that it is necessarily justified from a legal one. It is not unusual to hear people say that a particular case ‘must be right’ because the company in question is superdominant or has done something that looks mischievous. Some people have reacted to my posts on the Post Danmark saga by saying that action by the Danish competition authority was justified because it concerned the behaviour of an incumbent in a recently liberalised industry. Again, this is not the point, and it is not the reason why I wrote about the case. More important, these are dangerous arguments.

Ignoring the law to focus on the outcomes of individual cases is, and has always been, a recipe for disaster. A particular rule may seem unproblematic, from a policy-making standpoint, if it is confined to an exceptional case arising in very particular circumstances. We know well by now that, once a rule is laid down, it is only a matter of time before it is expanded in other contexts that have little to do with the specific factual scenario in which it seemed appropriate and innocuous. Alfonso likes to quote Areeda emphasising the importance of defining sensible limiting principles to enforcement in competition law. As you can see, I like to do so too. It is indeed one of the most important lessons to draw from decades of enforcement.

Just think of Intel again. I never argued that the outcome was wrong. In fact, the case is particularly interesting in that it exemplifies the commitment of the Commission to devote its resources to cases where negative effects on competition are likely. As a result of this commitment, reflected in the Guidance, it may well be the case that the prima facie prohibition of exclusive dealing and loyalty rebates does not make a practical difference when action by the Commission is at stake. This fact does not mean, however, that the perpetuation of a rule that fails to capture the lessons of experience and economic analysis is unproblematic.

The Commission may carefully choose to intervene in clear-cut cases, but what if a national court is asked to intervene in a case where anticompetitive effects are implausible? In the era of private enforcement and decentralisation, it is simply no longer possible to rely alone on the expertise of the Commission and its ability to prioritise cases. Whether or not the authority ‘gets it right’ in a particular case is not the only relevant question anymore. As the Post Danmark saga shows, much has changed since Hoffmann-La Roche, and rules and standards need to be crafted to ensure that they are accurate and administrable. These are the concerns that prompted my paper on Intel and my many posts on Article 102 TFEU. Nothing to do, as you can see, with whether I like or dislike a specific decision.

In light of this extensive answer, I feel entitled to ask Alfonso whether he believes exclusive dealing should be prohibited absent an objective justification or whether, instead, Article 102 TFEU enforcement should follow the principles set out by the Commission in the Guidance and by the Court in Delimitis. Alfonso?

My blogging inactivity over the past few days has to do with spending two weeks in Luxembourg for hearings with a short visit to Madrid for this seminar, which, btw, exceeded the already high expectations. In intense weeks like these ones I tend not to post even stuff I’ve written before because the downside of having your own clients read your blog is that they may get the impression that you’re blogging instead of preparing [Note to them: don’t worry, nothing further from the truth…].

This is all to justify say that today’s post will not be very brainy; here is just a compendium of short reflections spurred by recent experience:

On litigation. I’ve always said that litigation is the most fun part of my job. The problem now is that with the prevailing “amicable” solutions in antitrust (settlements and commitments) the scope for litigation is getting narrower and narrower (as clearly reflected in the joint slides by Judge Van der Woude & Nicolas Petit available here). The result is that most competition-related litigation nowadays is State-aid related. I, for one, am involved now in over 30 ongoing State aid appeals. I guess it’s a consolation that at least one area of competition law is still within the realm of judges.

On state aid. State aid now gives rise not only to most litigation, but also represents a higher percentage of the output of DG COMP than mergers, cartels or abuse of dominance. Nevertheless, it receives much less attention from the people who typically comment on antitrust issues. In previous posts I’ve given my view of the possible causes for this (see here) but today I’ll focus on a consequence: in State aid the Commission can get away with some absurd arguable or inconsistent decisions with no one even noticing. Fortunately, EU Courts are trying to remedy that with some sensible case-law, which also goes similarly unnoticed.

On the Legal Service. Some time ago I also wrote a post on the Legal Service (see here) which earned me some criticism from fellow practitioners who thought I was being too nice to “the enemy”. After some recent hearings in which I’ve been opposing and disagreeing with them I have to stand firm on commending their work, particularly given the acts that they sometimes are obliged to defend 🙂

On the press and competition law. In the past few weeks there were also a few cases on which I have worked that appeared in the general press, and this also triggered a reflection. The coverage that non-specialized press often does of competition cases is shameful subject to improvement, often trying to present everything under a simplistic light attempting to give the impression that everything is a scandal (what they have done with the alleged dairy cartel in Spain is a very good illustration of this). This makes me wonder whether that is also the case with regard to other issues that I don’t know first-hand. Like with lawyers, the differences between good and bad journalists can be enormous, and in Brussels there are competition journalists who know more about competition law than many lawyers, but at the national level…  Considering the increased politicization of competition law (see here), the messages conveyed to the general public are increasingly important, and wrong messages may well lead to wrong decisions.

On Spanish competition law. Some of you have recently inquired about why, despite being two Spaniards, Pablo and I don’t comment on Spanish competition law developments. Well, aside from the fact that the scope of the blog is wider, we’ll comment on those whenever we have something positive to say… The only recent welcome development which deserves a comment is the sensible Supreme Court Judgment on fines on which we still have to comment (I’m waiting to see whether the consequences are the apparently intended ones). This Judgment, by the way, reminded me of something that is applicable to many debates on competition law enforcement and judicial review: much more than institutional arrangements, what matters is having prudent and knowledgeable people doing the work.

On Pablo’s most recent posts: A few of you reacted in private to Pablo’s recent posts on Post Danmark II asking whether he’s ever liked any abuse of dominance case ever done by the European Commission. I thought that was a very good question (and have been teasing him with it for a few days) and one that it’d be interesting to pose him in public. So, Pablo?

I explained last week why it is difficult to infer from the case law an operational test for the assessment of standardised rebates. The purpose of this second post is to start a discussion about the principles that should guide the definition of such a test. I do not believe a single approach is possible or conceivable, and I welcome your views on the question. As in other areas, there is more than one substantive standard that would appropriately capture the principles underlying the case law. Regardless of one’s stance, I see the following ideas as a sensible starting point for a discussion around the definition of a legal test:

• The ECJ presumes since Hoffmann-La Roche that quantity rebates are pro-competitive (or, if one prefers, a valid form of competition on the merits). Any legal test should be devised in a way that is compatible with this principle (both in theory and in practice).
• Quantity rebates do not always have a negative impact on the competitive process. One of the fundamental lessons to draw from Michelin II is that exclusionary effects are not an inevitable consequence of the implementation by a dominant firm of a system of quantity rebates.
• In Post Danmark I, the Court considered that the award of selective price cuts to attract the customers of a competitor is not abusive by its very nature. Absent evidence of predatory pricing within the meaning of AKZO, it is necessary to show that the practice has (or is likely to have) exclusionary effects.

Standardised rebates vs. selective price cuts. I would conclude from the above that a system of standardised rebates like the one examined in Post Danmark II should only be deemed abusive under Article 102 TFEU where there is evidence of its actual or likely exclusionary effects. This is the substantive standard that applies to selective price cuts à la Post Danmark I. I cannot think of a reason why a scheme of standardised quantity rebates should be subject to a more stringent test. In Post Danmark I, the price cuts were specifically devised to attract the customers of a rival. What would justify a stricter treatment of price reductions that apply uniformly across the board?

If anything, a system of standardised rebates is clearly less suspicious and more obviously pro-competitive than the application of selective price cuts. In the case of quantity rebates, price reductions are the consequence of an increase in the amount acquired from the dominant firm. Selective price cuts, instead, reward a decision to switch suppliers. It is therefore safer to presume that the former reflect the cost savings made by the dominant firm. If this is so, I would claim that the need to provide evidence of an exclusionary effect in the case of quantity rebates became inevitable the moment Post Danmark I required evidence in this sense in relation to selective price cuts.

Establishing exclusionary effects in practice. The next question is of course that of how exclusionary effects are established in the context of a particular case. The issue can be approached in two ways. One possible approach is to assume the exclusionary effects of a rebate scheme from the fact that it is ‘loyalty-inducing’. This legal test would be problematic for several reasons, some of which have already been mentioned above and in last week’s post.

• One reason is that we know from the case law (the ‘experience’ mentioned by the Court in paragraph 51 of Groupement des Cartes Bancaires) that the exclusionary effects of rebate schemes cannot simply be assumed from their ‘loyalty-inducing’ nature, even when applied by dominant firms. They may or may not be manifested in the context of a particular case.
• Another reason is that the point of the selective price cuts examined in Post Danmark I was not conceptually different from that of ‘loyalty-inducing’ rebates. The price cuts in question sought to attract the customers of a rival. In other words, they sought to reward disloyalty. I cannot think of a valid reason why ‘loyalty-inducing’ rebates would be prima facie prohibited irrespective of the effects they produce and ‘disloyalty-inducing’ price cuts allowed absent evidence of exclusionary effects.
• Finally, one should bear in mind that all rebates are, at least to some extent, ‘loyalty-inducing’. Rebates, irrespective of their nature and objective, provide an incentive to buy more from a particular supplier. As a result, it seems possible to argue in virtually every instance that a given scheme is ‘loyalty-inducing’ and as such should be prohibited. Thus, not only would this approach fail to provide the basis for a practicable legal test but it would also contradict the principle whereby quantity rebates are presumed to be pro-competitive.

A second possibility is to endorse the logic underlying the Court ruling in Post Danmark I. In essence, this approach would involve showing that the practice it at least likely to harm the ability and the incentive to compete of equally efficient rivals. A legal test revolving around this question would have the advantage of consistency. The same substantive standard would apply to all practices that are not deemed abusive by their very nature. In the current state of the case law, this category includes not only quantity rebates and selective price cuts, but other practices like ‘margin squeezes’, the abusive nature of which is established in accordance with the same principles.

There are several ways in which the impact of a system of quantity rebates on rivals’ ability and incentive to compete can be established. The Court held in Post Danmark I that equally efficient competitors are unlikely to be excluded where the dominant firm charges above-cost prices. Accordingly, it would be necessary to show that the system of quantity rebates is in some way predatory (that is, that it forces rivals to sell below cost). I welcome views on the question, but I struggle to find how it would otherwise be possible to establish the exclusionary effects of schemes like the one examined by the Court in Post Danmark II. Such schemes do not seek to reward loyalty, nor are they conditional upon customers buying a certain share of their needs from the dominant firm. AKZO and Post Danmark I (as opposed to Hoffmann-La Roche and Delimitis) therefore look like the closest precedents.

A legal test based on these principles would not only have the advantage of consistency but would be infinitely easier to administer than the existing unstructured standard. We tend to forget, when thinking about these matters, that the AKZO test was developed by the Court, not the Commission. In its decision, the latter favoured an unstructured standard based on a range of factors. According to the Commission in AKZO, pricing below cost would not be a decisive criterion to establish the abusive nature of a predatory pricing campaign. The Court rejected this approach and crafted a substantive test confining administrative action within a set of boundaries well-defined in advance. Post Danmark II raises, in my view, the exact same issues.

Intel has given rise to much excitement in the competition law community. No doubt it is a high-profile case where much is at stake. If you asked me, however, I would say that the pending preliminary ruling in Post Danmark II is more important for the future of Article 102 TFEU. For some reason, it has attracted relatively little attention (one notable exception is Luc Peeperkorn’s recent article, mentioned by Alfonso a few days ago). Now that the date of the hearing (26 March) is approaching, it makes sense to say a few words about it. From the outset, I have to thank Christian Bergqvist, from the University of Copenhagen, for his availability to discuss the background and some aspects of the case.

Some authors have claimed in the past months that the case law on exclusive dealing and rebates has the advantage of providing legal certainty. It is clear that exclusivity obligations and rebates conditional upon exclusivity are, absent an objective justification, unlawful under Article 102 TFEU. A qualified prohibition rule of the kind is undeniably easier to administer than a standard (a different story is of course that the former is less accurate and more error-prone than the latter).

What these authors do not always emphasise is that the case law would be easier to administer if it prohibited altogether, absent an objective justification, all other rebate schemes. Even easier to administer would be a rule allowing all rebates. The fact is that schemes that are not formally conditional upon exclusivity are, at least in principle, only prohibited where they are shown to be ‘loyalty-inducing’ (or ‘fidelity-building’). As explained by the GC in Intel, this is a matter to be established following an assessment of ‘all the circumstances’ relating to the rebate scheme in question. The lawfulness of these rebate schemes is, in other words, formally assessed in accordance with a standard.

Some standards are good and some are not. A good standard is one that is structured around a set of well-defined proxies and that is predictable for firms and easy to administer for courts and authorities. Bad standards, in turn, are unstructured and make it impossible for firms to anticipate the outcome of a particular case. Post Danmark II shows that the standard under which rebates not formally conditional upon exclusivity are assessed is closer to the second category. The criteria identified in the case law to determine whether a given rebate scheme is ‘loyalty-inducing’ do not provide the basis for an operational legal test that can be readily applied by national courts and authorities.

Post Danmark II is particularly interesting because it is about standardised rebates (this is explicitly mentioned in the preliminary reference). A scheme of that kind is in principle lawful under the case law. Quantity rebates apply in the same way to all customers and increase with the amount supplied and are thus presumed to reflect the cost savings made by the firm. Is that the end of the story? Does it mean that the scheme in question is compatible with Article 102 TFEU? Not really. We know, at least since Michelin II, that quantity rebates can be found to be ‘loyalty-inducing’ and as such prohibited. As in that case, the rebates in Post Danmark II are retroactive. Is the retroactive nature of the rebates alone sufficient to establish an abuse? An additional factor that was considered in Michelin II is the reference period considered for the award of the rebates. Are all rebates granted over a one-year period prohibited?

One can try and read and re-read the relevant rulings, but I am afraid that nothing in the case law clarifies how the abovementioned factors (some hinting at the lawfulness of the scheme, some suggesting a violation of Article 102 TFEU) are weighed against one another to establish whether, on the whole, a particular scheme is ‘loyalty-inducing’. The case law, in other words, does not provide a set of principles for the systematic assessment of the nature of rebate schemes. The different criteria have so far been considered in an unstructured way and in light of the facts specific to individual cases.

In practice, and in the absence of a legal test, it is sufficient for a competition authority (or a claimant) to identify some factors hinting at the ‘loyalty-inducing’ nature of the scheme to establish a violation of Article 102 TFEU. This is at least the lesson I draw from Michelin II and British Airways. As soon as any such factors are identified, an abuse is prima facie established and the burden of proof shifts. It is then for the dominant firm to show that the rebate scheme is economically justified.

EU courts, most recently in Intel, have repeatedly held that rebate schemes that reflect cost savings are compatible with Article 102 TFEU. Similarly, they have clarified that a case-by-case assessment is necessary for rebates other than those conditional upon exclusivity. The case law has not evolved in a manner consistent with these principles. In reality, and for all practical purposes, rebate schemes seem to be prima facie prohibited irrespective of whether they are formally conditional upon exclusivity. I would assume that it is uncontroversial to state that, where the law fails to capture the principles on which it is based, as is true here, it needs to change or at least be refined.

The reasons why the case law has moved away from its underlying principles are not a mystery. It has developed in the context of annulment proceedings, where the primary task of EU courts is to review the legality of administrative action. In such a procedural context the GC and the ECJ are, an analysis of the case law suggests, less likely to develop a substantive test that captures the principles against which the lawfulness of the practices in question is assessed. Unsurprisingly, they are more likely to focus instead on whether the analysis carried out by the Commission is sound and whether it supports the conclusions drawn by the authority.

The pending ruling in Post Danmark II shows that the case law, as it stands, does not provide meaningful guidance to courts and authorities in the era of decentralisation – remember there is a decision by the Danish competition authority at the origin of the case. The good thing about the case is that, as much as Post Danmark I, it is a unique opportunity for the Court to clarify the law in the area and to contribute to legal consistency. These are, after all, the very reasons why preliminary references are so important in the EU system of judicial remedies.