Intel v Commission and the problem with wrong economic assumptions
(by Pablo Ibañez Colomo)
Voices that relativise the problems with Article 102 TFEU case law are not infrequent. It may be true that the case law is not beyond reproach in all respects, the argument goes, but perfection is not of this world. The fact that rulings are often criticised simply means that Article 102 TFEU is an inherently controversial provision and that the stakes in abuse cases are generally very high, not that there is something fundamentally wrong with the preferences expressed by EU courts. And in any event, the alternative, economics-based, approaches have their problems too. The current case law is just the expression of a legitimate choice.
There is of course some truth in this position. At the same time, I find a bit defensive and as such problematic because it can become an obstacle to an honest and constructive exchange of ideas. I can think of at least a fundamental aspect that is uncontroversially (or objectively, if one prefers) wrong with Article 102 TFEU case law. What makes it even more interesting is that it fails to attract the attention that, in my view, it deserves. We all know that exclusive dealing and loyalty rebates are (absent an objective justification) abusive under Article 102 TFEU. The assumption underlying this rule is discussed far less often and is crucial to understand the case law. In paragraph 77 of Intel, the Court repeats the old formula whereby the abovementioned practices, as opposed to quantity rebates, ‘are not based – save in exceptional circumstances – on an economic transaction which justifies this burden or benefit but are designed to remove or restrict the purchaser’s freedom to choose his sources of supply and to deny other producers access to the market’.
This statement, as a matter of economics, is incorrect. Contrary to what the Court holds, there are perfectly valid pro-competitive justifications for exclusive dealing and loyalty rebates. I am inclined to believe that everyone at DG Comp and the Legal Service agrees by now with this idea, which has long been part of the mainstream. Suffice it to check any textbook on industrial organisation or the economics of competition law. To mention the three I had in my office when preparing this post, take Carlton & Perloff; Bishop & Walker; or Niels, Jenkins & Kavanagh (Hans Zenger’s piece on loyalty rebates is great too). Given its peculiar cost structure, some of these justifications are of obvious relevance in the microprocessor industry.
Article 102 TFEU case law will not evolve until the ECJ acknowledges that a rule-based approach to exclusive dealing and loyalty rebates is grounded on a misguided economic assumption. Interestingly, a shift in this direction would not require a major revolution. The ECJ would just have to accept – finally – that what is true under Article 101 TFEU must by definition be true under Article 102 TFEU. In paras 10-12 of Delimitis the Court holds that there are perfectly valid justifications for exclusive dealing and – by extension – for loyalty rebates. As a result, they are not restrictive by object. Article 102 TFEU case law cannot be based on the opposite assumption (i.e. that these practices are anticompetitive by their very nature because they have no economic explanation other than the exclusion of competition). Paragraphs 89-91 of Intel show the difficulties into which EU courts run whenever the tension between these two lines of case law is raised (Van den Bergh Foods being another excellent example).
I am convinced that an effects-based approach would follow logically from the suggested shift. The additional arguments raised in subsequent cases to justify the current approach are not particularly persuasive. The fact that dominant firms have a ‘special responsibility’ that derives from their status does not mean that an effects-based approach to loyalty rebates and exclusivity is not conceivable. There are recent cases, like Post Danmark and TeliaSonera, where the ‘special responsibility’ of dominant firms is seen as compatible with requiring evidence of an anticompetitive effect.
Paragraph 77 of Intel also made me think of the relationship between law and economics in competition law. It is interesting that the General Court reiterates the Hoffmann-La Roche formula to make it clear that there is a long line of case law supporting its position. ‘Exclusive dealing and loyalty rebates have no pro-competitive justifications because we have always said they do not’, the judges appear to claim. What is an economic argument is dealt with, in other words, as a legal one. From an economic perspective, to be sure, the fact that EU courts have consistently relied on the same assumption does not make the latter any less incorrect.
The Intel judgment also made me think of something I often say. Economic analysis is sometimes presented as an exogenous force that has interfered with EU competition law since the 1990s. What wrong assumptions such as the one discussed in this post show is that this view is not accurate. Economics is hard-wired into competition law – it is an integral part of it. The only debate should be whether to rely on one’s more or less accurate intuitions (à la market definition in United Brands, for instance) or to trust instead the analytical tools developed over several decades by competent individuals devoting their professional lives to a systematic understanding of the economic side of the discipline.
Chillin'Competition 
Pablo, excellent post as always. A couple of comments:
1) Your main point -if I undertand correctly- is that “[c]ontrary to what the Court holds, there are perfectly valid pro-competitive justifications for exclusive dealing and loyalty rebates”.
I don’t think the Court is denying that this may be the case. Actually, it explicitly acknowledges the point; take a look at recital 94 of the Judgment:
“Lastly, it should be noted that it is open to the dominant undertaking to justify the use of an exclusivity rebate system, in particular by showing that its conduct is objectively necessary or that the potential foreclosure effect that it brings about may be counterbalanced, outweighed even, by advantages in terms of efficiency that also benefit consumers (see, to that effect, Hoffmann-La Roche, paragraph 71 above, paragraph 90; Case C‑95/04 P British Airways, paragraph 74 above, paragraphs 85 and 86; and Case C‑209/10 Post Danmark [2012] ECR (‘Post Danmark’), paragraphs 40 and 41 and the case-law cited). However, in the case in point, the applicant has put forward no argument in that regard”
The Court does therefore acknowledge that there may be perfectly valid and pro-competitive justifications for royalty rebates; the problem seems to be rather that those arguments weren’t presented to it (admittedly, I take that reproach to the applicants with a pinch of salt; I’ve no elements to know that this was the case here, but I do know of other Judgments where similar pretexts were used to avoid certain discussions).
Also, don’t forget that the Court did anyhow undertake a detailed and thorough review and verified that the facts of the case were consistent with the presumption the Court endorses.
—
2) At the end of the day, the Court establishes a rebuttable (see above) presumption that these rebates further restrict competition in markets where competition is significantly reduced as a result of the presence of a dominant company.
The presence of a dominant company is the reason why the Court doesn’t accept your contention that “what is true under Article 101 must by definition be true under Article 102”. In this regard see recital 89 of the Judgment:
“Although exclusivity conditions may, in principle, have beneficial effects for competition, so that in a normal situation on a competitive market, it is necessary to assess their effects on the market in their specific context (see, to that effect, Case C‑234/89 Delimitis [1991] ECR I‑935, paragraphs 14 to 27), those considerations cannot be accepted in the case of a market where, precisely because of the dominant position of one of the economic operators, competition is already restricted (see, to that effect, Case C‑310/93 P BPB Industries and British Gypsum v Commission [1995] ECR I‑865 (‘Case C‑310/93 P BPB Industries and British Gypsum’), paragraph 11, and the Opinion of Advocate General Léger in that case, points 42 to 45)”
If you ask me, this reasoning carries some weight. The underlying reasoning is exactly the same as the one justifying the, I would say, uncontroverted idea that conduct that is be legal when undertaken by a non-dominant company may in some settings be illegal if carried out by a dominant one. In other words, I would disagree with the sweeping statement what’s true under 101 must (not) by definition be true under Article 102.
Alfonso Lamadrid
16 June 2014 at 12:08 pm
Excellent comment as always 😉
On your first point:
– In para 77, the Court acknowledges that, in ‘exceptional circumstances’, there may be an economic justification. The fact that (in theory) an objective justification is available does not mean that the statement is a valid one. The premise remains wrong and not in line with reality. Assuming that in ‘normal circumstances’ loyalty rebates have no valid pro-competitive justification remains incorrect as a matter of economics.
On your second point:
– As I mention in the post, the GC had to acknowledge that the ECJ has relied on contradictory assumptions in relation to exclusive dealing and similar practices. It is difficult to reconcile the two strands of the case law. I also suggest in the post that the argument you discuss is a non sequitur. The fact that competition is already restricted does not imply that anticompetitive effects should not be established. This would mean that exclusionary effects would never have to be shown in the context of Article 102 TFEU. But we both know of cases where the application of this provision depends on evidence of anticompetitive effects.
– In any event, the fact remains that the underlying assumptions contradict each other. It is obvious that the probability of anticompetitive effects is greater when a firm is dominant. Nobody would dispute that. However, the fundamental point is that under Article 101 TFEU exclusive dealing is assumed to be a normal competitive practice and under Article 102 TFEU it is assumed to be, absent ‘exceptional circumstances’, a device aimed at excluding competitors (i.e. it is assumed to have an anticompetitive object, as Eric was pointing out in a comment to your post). And the fact that it is not necessary to show effects is the very consequence of this (incorrect) assumption. As in Article 101 TFEU case law, the practice is assumed to be anticompetitive by its very nature, which is in turn why the GC held in Michelin II that ‘object and effect’ are sometimes one and the same thing in the context of Article 102 TFEU. If the underlying assumption were sound from an economic perspective, I would not have any problem with this passage in Michelin II. The problem is that it is not.
Pablo Ibanez
16 June 2014 at 12:53 pm
I concur with Alfonso’s comment. Crucially, the post forgets what the judgment reminds us at §§ 94 and 173: that Intel did not argue that its behaviour was a source of efficiency gains or had any procompetitive justification. The same had happened in Tomra.
In other words, Pablo’s intuition that Intel’s loyalty rebates were efficient and procompetitive may well be accurate -or not. We will never know because it was not the position put forward in the case by Intel. I don’t doubt that Intel hired many “competent individuals devoting their professional lives to a systematic understanding of the economic side of the discipline”, but somehow they chose not to argue that the behaviour was, on balance, pro-competitive or efficiency enhancing (even if this was ‘obvious’). Like it or not, pro-competitive justifications and efficiencies come only as a second step in Article 102 analysis, and need to be explicitly raised and argued convincingly.
Unlike Alfonso, I would not take §94 with a pinch of salt. For the simple reason that, were the GC to have distorted the pleas of a party, this is reviewable on appeal. If it turns out that the GC ignored or misrepresented one party’s pleas on a material point, its judgment would never survive review.
My take on some recent literature (Elhauge, Economides, Maier-Rigaud & Schwalbe, to name just a few) makes me doubt that the club of “competent people” (now, that’s a desirable club membership) is as unanimous as the post makes them to be on the likely effects of loyalty rebates and the adequacy of a rule-based approach.
I understand Pablo’s disappointment and sanguine opinions. I just hope they will not deter other contributions discussing further the Intel judgment and finding some virtue in it. The more the merrier ! Personally, I think it is a well drafted and internally consistent judgment. It will not please everybody, but it’s worth a careful and unbiased reading before forming an opinion.
Eric
16 June 2014 at 4:14 pm
Hi Eric,
Thanks for your comment.
I thought it was clear that I was not claiming that Intel’s practices were ‘efficient and pro-competitive’. The point of my post is a general one that is not related to whether the lawyers and economists working for a dominant firm in a particular case claimed (or had persuasive evidence suggesting) that the practice in question was pro-competitive. I did not say anything about the ‘likely effects of loyalty rebates’, which the economists you mention discuss.
My point instead is that there is no good reason to assume that (or to set a rebuttable presumption whereby) loyalty rebates and exclusive dealing have no valid economic justification; in the same way (and here we agree) that there is no reason to assume the opposite. What really matters is that the latter is at least a plausible explanation for the behaviour. Where the impact of a practice on competition is ambiguous, claiming that its ‘object’ is anticompetitive seems clearly incorrect, in my view. That was my point. Exclusive dealing is not deemed anticompetitive ‘by its very nature’ under Article 101 TFEU, and I see no convincing reason why this should be the case under Article 102 TFEU. That negative effects are more likely to arise when the practice is implemented by a dominant firm? Of course. But this is something to be established on a case-by-case basis. This is after all the logic behind the Guidance as reflected in the Intel decision.
On economists: consensus is not synonymous with unanimity.
And definitely: the more the merrier!
Pablo Ibanez
16 June 2014 at 5:10 pm
As an (institutional) economist as well as a lawyer, I’ve always argued that I’d rather have a (slightly) wrong per se rule than an effects-based test that you don’t know you’ve failed until you wind up on the wrong end of a Commission decision. If the law says that you can’t do loyalty rebates, that’s endlessly better than having to spend (and from a social welfare point of view waste) a fortune on lawyers only to be told that it’s probably better not to do loyalty rebates, just in case.
So I don’t tend to mind very much if the Court overlooks – or seems to overlook – a possible subset of cases where the behaviour in question is social welfare enhancing, as long as it sets out clear legal rules. In this case, though, it doesn’t seem to have done that either.
Martin Holterman
16 June 2014 at 8:40 pm
Pablo’s reaction is understandable in his own intellectual eco-system. Given Pablo’s concept of what a restriction “by object” is (one that facially lacks any plausible economic justification), then obviously he can’t be satisfied with Intel’s reference to “object”. But I don’t think I am overstating things if I say that not everybody shares that conception. Indeed, it can’t be the court’s approach, since it explicitly structures a rule making room for economic justifications and efficiency arguments.
It remains a fair question to ask Pablo if he does not see a tension between criticising the case law for -supposedly- resting on an ‘obviously’ wrong economic intuition…while Tomra and Intel, with an army of competent individuals, did not feel capable to even try to counter that intuition.
The Intel approach seems rather classic: rather than starting from an assumption that a practice of loyalty rebates by a dominant undertaking is usually desirable (call it efficient, welfare-enhacing, pro-competitive, depending on your priors about the interests protected by competition law), the Court identifies loyalty rebates as conduct which is in principle capable of distorting the competitive process, and requires the dominant undertaking to substantiate how this conduct benefits consumers.
The judgment seems to me to come quite close to Martin’s preferred approach (“you can’t do loyalty rebates”), except that it’s not a per se prohibition, you are allowed a defence.
I don’t think we will see a change in EU case law on loyalty rebates until critics of the doctrine and other “competent individuals” are ready to put their money where their mouth is (explaining and substantiating economic justifications in real cases). It should not be so difficult if these are “obvious”. Once the Commission and the Court have accummulated experience with cases where efficiency gains and economic justifications are substantiated, they may develop more refined structured rules (for example, presuming the justification/efficiencies under certain circumstances). But this will not happen if criticism remains based on the alleged pervasiveness of pro-competitive justifications….which nobody is willing to argue in a real case.
Eric
17 June 2014 at 12:32 pm
Being in favour of clear and coherent rules (recognising their welfare enhancing effects as Martin H does) I have to concur with Eric…If we bring the post’s argument ad extremum, any exclusivity arrangement would have to be argued on an effects basis, even a 50 year long exclusivity agreement concluded by a dominant firm (perhaps holding an essential facility) would have to be assessed on a case-by-case basis…this would surely not be in the interest of European competition law and consumers!
Asimo
17 June 2014 at 1:40 pm
That is a very fair question, Eric, and one that goes to the heart of the issue, I would say (like the rest of the comment).
This discussion, to me, is essentially about the two ways in which competition law can be shaped. A possible approach is the one you mention: intervention in a given case depends on whether there is evidence of pro-competitive effects. If the parties are unable to put forward evidence in this sense, the practice will be prohibited. An alternative approach is to intervene only where there is evidence of anticompetitive effects. The practice is allowed unless the authority brings evidence of its negative impact on competition.
The law on exclusive dealing and loyalty rebates is based on the first approach. Merger control, on the other hand, is based on the second. A merger control system based on the former is not inconceivable, however. Mergers would not be allowed unless the parties bring cogent and convincing evidence that the transaction will lead to efficiency gains. Some people might even argue that this would be a better system. It may even be the system that you would personally favour. Only the really good mergers would go through. Changes in the market structure would not be allowed unless there is a good reason. And there is research suggesting that it is not infrequent that the efficiency gains expected from mergers fail to materialise (which would mean that society and consumers gain nothing from them).
The question we should be asking is why merger control is based on a system that depends on a case-by-case assessment of the likely effects of operations. To me, the answer is quite clear. While mergers (in particular horizontal mergers) can impact negatively on competition, there is no reason to believe that there is something inherently perverse about them. Does it mean that mergers always generate efficiencies? No. Does it mean that companies never merge to avoid competing or to exclude their rivals? No. It simply means that it would be incorrect to assume that this is the motivation behind mergers. To the extent that efficiency is a plausible motivation for concentrations, the focus of the analysis revolves around their likely negative effects. Only after this analysis is the burden of proof reversed. Mergers are routinely allowed without an analysis of the efficiency gains to which they lead. And, in line with the above, I suspect that many of them would not have been allowed to go through had the parties been required to show that they would be pro-competitive.
This is exactly the point of my post. It is wrong to assume that there is something inherently anticompetitive about exclusive dealing and loyalty rebates. Thanks to economists, we understand this. It is wrong to assume that they are implemented to exclude competitors, in the same way that it is wrong to assume that companies merge to eliminate competition. Therefore, it makes sense to make intervention depend on an analysis of the likely effects of the practice on competition (you will agree that ‘capability’ is not ‘likelihood’, and Article 102 TFEU is full of examples showing it), instead of shifting the burden of proof and make the assessment revolve around its pro-competitive aspects. What the Guidance did is acknowledge that practices that are in principle ‘capable’ of having negative effects when implemented by dominant firms, like tying, exclusive dealing or rebates, are also ‘capable’ of producing positive effects. As a result, it makes sense to bring enforcement in line with the logic of merger control. To address the question; acknowledging the above and shaping enforcement accordingly does not mean in any way that exclusive dealing and loyalty rebates will never have negative effects on competition (I understand the research you mentioned in your first comment as a reaction to the idea a quasi-per se legality rule for loyalty rebates) or that there are valid pro-competitive justifications in every single case. So no, I do not see a tension.
To Asimo: mergers, even mergers to monopoly, are assessed on a case-by-case basis, and nobody seems to be scandalised by that.
Pablo Ibanez
17 June 2014 at 3:11 pm
The difficulty I have with your argumentation is that it assumes that EU law on exclusive dealing and loyalty rebates is based on an approach where the authority does not bring any evidence of “negative impact on competition”, and therefore everything is prohibited unless shown to be good. This is a caricature.The approach is a structured rule which considers certain behaviour problematic because it is conducive to an impact on market structure without being obvious competition on the merits (i.e. bringing about a direct benefit for consumers; it would be too long to discuss here whether there is an effective ‘discount’ when a monopolist gives a loyalty discount, but there is good literature suggesting that there is not).
Impact on the competitive structure of the market may not be the kind of impact that you want examined, but it’s not the same as showing nothing.
This is where your analogy to mergers becomes interesting. EU merger control is largely structuralist. It was widely applauded even during its first fifteen years where the logic of merger control was exclusively concerned with market structure (Regulation 4064/89 was ONLY concerned with the creation or reinforcement of a dominant position). The “case by case approach” was largely devoted to counting market shares, and the “effects” were largely the addition of market shares, concentration ratios and HHIs. Even today, empirical studies show that Reg. 139 has not significantly changed the outcomes that could be expected under Regulation 4064. The structuralist approach to “effects on competition” which you judge insufficient under Article 102 has been at the heart of merger control analysis for much of its history.
Similar ‘structural’ rules apply in the U.S. to exclusive dealing, which is prohibited upon showing market power and a sufficient foreclosure share (loyalty discounts have been treated under this rule occasionally: Masimo Corp v. Tyco Health Care, LePage). Idem for tying, which is prohibited upon showing (a) separate products; (b) tying market power; (c) tying; and (d) a substantial foreclosure share…. unless there are countervailing efficiencies. This type of rule can be refined, and this is the kind of evolution I referred to at the end of my previous post. It is in the best tradition of antitrust law, where economics informs the creation of structured rules. ‘Case-by-case’ law is an oxymoron.
I will not prolong this ping-pong game. Please don’t take offence in my disagreement. The bottom line is that, much as I enjoy your writings, I am usually not interested in criticism suggesting that judges are stupid or obviously wrong or just “don’t get it”. Things are seldom that simple.
Eric
17 June 2014 at 4:39 pm
Your comment raises many exciting issues that I hope to discuss with you at some point.
I will only address the last comment for the time being. Saying that at the heart of the case law there is an assumption that is incorrect as a matter of economics is emphatically not the same as saying or suggesting that ‘judges are stupid or obviously wrong or just “don’t get it”’. Very very far from that. I know full well that things are not simple, in particular when it comes to devising and refining rules and standards in competition law. This is what makes it a very exciting field.The law evolves at the pace it does and it usually takes time before judges find the way around a particular issue, even when they know or feel that the law needs refinement (and I am not implying that this is the case in this particular area, I write in general terms). The Commission Guidance departs in several important respects from the case law. I do not believe the document was prepared as a reaction to the perceived inability of EU courts to grasp several crucial issues. As myself, I am sure these officials have the utmost respect for the Court, and are very much aware of how incredibly difficult their task is.
The post was precisely about promoting an honest and constructive exchange of ideas. This can only be done, first, if one acknowledges that things are not simple. Of course they are not. The case law in this field is the product of several decades and reflects evolving ideas and assumptions. And it is often the case that the judgments do not even capture faithfully the spirit of the proceedings. I am convinced that trying to untangle some of these assumptions and discussing the extent to which they are in line with mainstream economic consensus contributes to such an honest and constructive exchange. At least it was that in that spirit
Pablo Ibanez
17 June 2014 at 5:25 pm