The author identifies two main takeaways from the Intel decision: (i) as a matter of principle, Article 102 TFEU is only concerned with the exclusion of rivals that are as efficient as the dominant firm. The departure from the market of rivals that are less attractive in terms of, inter alia, price, quality or innovation is deemed to be a natural outcome of the competitive process and as such unproblematic; (ii) practices are only caught by Article 102 TFEU insofar as they are capable of having anticompetitive effects. By the same token, it should always be possible for a dominant firm to provide evidence showing that, in the context in which it is implemented, the practice is incapable of having such effects.
The article – which can be found here – seeks to explore these elements in more detail. It is structured as follows:
- Section II reviews the case and its background. I am not going to do that here (again), but it is worthwhile to highlight a few elements of the author’s analysis. In particular, he finds that the Commission’s: ‘Intel decision is a faithful expression of the administrative practice that followed the review process that started with the Discussion Paper of 2005. On the one hand, the Commission reiterated the rule laid down in Hoffmann-La Roche, under which exclusive dealing and loyalty rebates are prima facie prohibited without it being necessary to examine their concrete impact on competition. On the other hand, the decision examined at length, in line with the principles set out in the Guidance Paper (issued before the adoption of the decision but after the start of the investigation) the anticompetitive effects of the said practices. The analysis of the effects was conducted for prioritisation purposes, not as a legal requirement.’
He also provides a fairly accurate assessment of the academic debate that followed the Commission and General Court’s decision: ‘The discussions were typically normative in nature. In other words, commentators focused on whether the ‘by object’ qualification for exclusive dealing and loyalty rebates is justified.’
He then interprets the judgment of the CJEU as follows: ‘It restates the rule pursuant to which exclusive dealing and loyalty rebates are prima facie prohibited irrespective of their effects on competition. On the other hand, the Court points out that, where a dominant firm provides evidence showing that the rebate scheme is not capable of having an anticompetitive effect, the Commission is required to consider the nature, purpose and operation of the practice, including its coverage, the extent of the dominant position, the conditions imposed on customers as well as any evidence of an anticompetitive strategy. Moreover, the judgment reiterates that, if a prima facie infringement is established, it is possible for a dominant firm to come up with an objective justification for its behaviour and/or claim the efficiencies deriving from it.’
- Section III then reviews the judgment, and considers that there are four main lessons that can be derived from it: (i) as a matter of principle, Article 102 TFEU is concerned with the exclusion of rivals that are as efficient as the dominant firm; (ii) for exclusive dealing and rebates to be caught by Article 102 TFEU, they must at least be capable of restricting competition. If, in the relevant economic and legal context, these practices fail to meet the threshold of capability, they are not prohibited under this provision; (iii) capability is presumed in the context of ‘by object’ practices. In other words, to effectively discharge its burden of proof that a practice restricts competition for the purposes of Art. 102 TFEU, an authority or party only has to establish that the ‘by object’ practice has been implemented; (iv) ‘a dominant firm can adduce evidence showing that the ‘by object’ practice is not capable of restricting competition in the specific economic and legal context in which it is implemented. In other words, the dominant firm may rebut the presumption of capability underlying the legal test applying to ‘by object’ conduct. Where a dominant firm produces evidence in this sense, an authority is required to engage with it and establish that the practice is capable of having anticompetitive effects on the basis of an analysis of the relevant market and the nature and operation of the contentious practice.’
He considers that these principles had already been outlined as regards Art. 101 TFEU in T-Mobile. In fact, the author goes as far as to say that: ‘There should be no doubt after Intel that the same principles apply in the context of Article 102 TFEU. There is, in other words, a consistent approach to ‘by object’ practices across the board.’ In short, ‘Intel merely articulates a principle that was already explicit in the context of Article 101 TFEU.’ There is said to be a reason for this: ‘The case law suggests that the ‘by object’ category – under both Articles 101 and 102 TFEU – seems appropriate for behaviour with a purpose that is objectively anticompetitive. The ‘by object’ approach is considered to be warranted, in other words, where a practice is deemed to have no rationale other than the restriction of competition. (…) Where a practice falls within the ‘by object’ category, it is only sensible to presume that it is at least capable of having restrictive effects on competition. (…) By the same token (and this is the essence of the clarification introduced in Intel), if it appears that the practice is incapable of having anticompetitive effects, it is no longer safe to presume that the rationale for the practice is anticompetitive.’
In a sense, the clarification introduced by the Court in Intel is an acknowledgement that not all restrictions by object are created equal. As the case law stands, the ‘by object’ label applies to a set of practices that is not homogeneous: it applies to manifestly anticompetitive conduct, but it also applies to conduct that may have pro-competitive effects. As such, a mechanism allowing the parties to show that anticompetitive effects are implausible, such as the one introduced in Intel, is indispensable.
- Section IV then focuses on clarifying the circumstances under which a practice is ‘capable’ of restricting competition. It is noted that the Intel ruling fails to define the notion of capability and does not provide details about what dominant firms need to show if they wish to rebut the presumption in practice. However, the decision does provide some guidance on both topics.
The case law suggests that the notion of capability can be distinguished from that of likelihood. ‘Capability’ is an expression which has been used in the past, even if it has never been given a precise definition. The threshold of capability as deployed in cases other than Intel (including in Art. 101 TFEU cases) is a relatively low one, however, and can be equated with plausibility. In other words, the standard of capability is met where a restriction of competition is a conceivable – but not necessarily likely – outcome, and seems to apply to ‘by object’ restrictions. As regards practices which are not by object, the applicable threshold seems to be ‘likelihood’ of anticompetitive effects – which the author equates with a balance of probabilities test. Whether such ‘likely’ effects arise is something that needs to be demonstrated on a case-by-case analysis.
- Section V reviews the type of circumstances where evidence that the conditions of competition would have been the same with and without the contentious behaviour can be adduced. This would seem to include situations where the practice is not capable of altering the conditions of competition as a result of the underlying regulatory framework or the application of IP rights; situations falling under the ancillary restraints doctrine, or whatever equivalent doctrine may apply to unilateral conduct; and, as per Intel, situations where evidence is submitted that the causal link between the practice and its (actual or alleged) effects is implausible.
According to Intel, a firm can attempt to rebut the presumption of capability by claiming that the practice would not force equally efficient rivals to sell below cost. Evidence in this sense does not exclude altogether a finding of infringement but, as suggested in Intel, requires an authority or claimant to identify other factors showing that, in spite of this fact, anticompetitive effects remain plausible. These anticompetitive effects could be shown via some additional factors such as the proxies identified by the Court in Intel, which comprise the coverage of the practice, the extent of the dominant position and the length of the obligations. However, the judgment sheds little light on how claims relating to these proxies would operate in the context of a particular dispute.
- An additional question is what level of evidence dominant firms would have to meet if they want to discharge their burden of proof and overturn the presumption that the exclusive discounting practice is anticompetitive: ‘In this regard, there are two approaches to the issue that come across as reasonable. Under one approach, dominant firms would have to satisfy the same level of evidence that the Commission has to meet to discharge its legal burden of proof. (…)Under [a] second approach, it would be sufficient for the dominant firm to raise doubts about the capability of the practice to restrict competition. Arguably, adducing evidence that has an ‘air of reality’ should be sufficient to trigger an obligation on the authority or claimant to establish that the practice is capable of restricting competition. The main argument in favour of this second approach is that it would be consistent with the presumption of innocence, and with the fact that the burden of establishing an infringement lies with the authority or claimant’.
In the context of Intel, it would seem that if evidence that an equally efficient rival would not have been forced to sell below cost is adduced, the Commission cannot avoid engaging with the effects of the practice. Accordingly, the authority (or claimant) would have to show why the analysis of the dominant firm is incorrect or why other features of the practice and/or the relevant market lead to the conclusion that the practice is capable of having restrictive effects, in spite of the conclusions flowing from the application of the ‘as efficient competitor’ test. It is less straightforward to see how a dominant firm can challenge the presumption of capability in light of factors such as the coverage of the practice, the extent of the dominant position and the duration of the obligations. However, one can think of arguments around the duration of the practice, or the fact that the company is not an unavoidable trading partner.
Comment: Overall, I think this is a really good discussion of the Intel judgment and its implications. I also think that its discussion of how a dominant company may rebut the presumption that its discounting practices are anticompetitive s stronger than that in the paper above. Unsurprisingly, I also really enjoyed the discussion on standards of proof for various practices and for rebutting presumptions that may arise in the context of the analysis of such practices.
Nonetheless, I disagree with some aspects of the author’s analysis of the case. This discussion will be a bit long and technical, and boils down to different interpretations of substantive EU case law on anticompetitive practices more generally – so feel free to skip it unless this is a topic you are interested in.
First, I think that the author may be reading a bit too much into an individual judgment. This is a decision that – we should remember – some authors considered could have merely procedural implications regarding the standard of judicial review that the General Court should adopt as regards infringement decisions.
Second, I am unable to follow most of Section III, and particularly the argument that Intel aligns the case law on Articles 101 and 102 TFEU. While I appreciate the intellectual sophistication of the proposed interpretation of the case law (and how it tries to develop a common approach to Art. 101 and 102 TFEU), I think the analysis is more normative than descriptive. In particular, I don’t see any support for the first three lessons (or for any ‘by object’ qualification of Art. 102 TFEU practices) in the Intel decision itself. This may be a possible interpretation of the case in the context of the case law on Art. 102 TFEU taken as a whole, but the ‘lessons’ that the author identifies do not seem to me to emerge directly from the judgment.
Perhaps more importantly, I don’t think this is the sole possible, or even the best possible interpretation of the case in the context of the case law.
To begin, I do not think that the conclusion that Art. 102 TFEU is concerned only with the exclusion of ‘equally effective competitors’ is supported by the case law. This requires one to overlook all cases of restrictions on parallel imports; the MEO case discussed a few weeks ago regarding the impact of second-line price discrimination where the dominant undertaking is not present in the downstream market; excessive pricing abuses, as well as other exploitative practices; and a number of exclusionary practices, such as abuses of the IP/judicial/regulatory system like the ones in Astra Zeneca or Motorola.
I also disagree that ‘there can be no doubt’ that the same principles govern both Art. 101 and Art. 102 TFEU. One may argue that the same principles govern, or should govern, these two provisions. But there can be, and are, significant doubts about this – or, at the very least, about the level at which those principles operate.
- First, it is rather controversial to argue that there are ‘by object’ restrictions under Art. 102 TFEU at all.
- Second, even if one were to accept that such ‘by object’ restrictions exist for both Art. 101 and Art. 102 TFEU, by object restrictions of collusive practices can only be overturned as a result of the fulfilment of Art. 101(3) TFEU. The requirements set out in this provision are not the same as those required to show that a dominant company’s conduct was objectively justified or counterbalanced by efficiencies that benefitted consumers under Art. 102 TFEU.
- Third, and to my mind more importantly, I do not think that under Art. 101 TFEU one can avoid classifying a conduct as restrictive ‘by object’ merely by showing that a collusive practice is not capable to restrict competition under the same terms as in Intel. This confuses the assessment of whether an agreement or practice is susceptible of restricting competition by object (i.e. the categorisation of a conduct) with an effects-based assessment which is framed by a presumption of illegality.
The categorisation of a conduct as a restriction by object under Art. 101 TFEU and the assessment of the effects of that same conduct are different things. This is apparent, for example, in the Toshiba – Power Transformers judgment. Here, the parties agreed that the parties were not potential competitors because Japanese companies could not enter into the market. The court held that market sharing arrangements had consistently been found to be a restriction by object and that, in such a case, ‘the analysis of the economic and legal context of which the practice forms part may thus be limited to what is strictly necessary in order to establish the existence of a restriction of competition by object’. In other words, the categorisation of a practice as being restrictive of competition ‘by object’ is not about the assessment of the effects of a conduct; and even evidence of the absence of those effects does not suffice to challenge the sanctioning of an object conduct. After all, one simple way of showing that a ‘by object’ agreement cannot have had anticompetitive effects is by showing that it has not been implemented. And, yet, the fact a cartel has not been implemented has been explicitly held not to change the way in which an object restriction is analysed (see, for example, Case C-407/04 P Dalmine v Commission ECLI:EU:C:2007:53, paras. 83-84).
It is true that the case law on Article 101 TFEU does seem to have adopted a (slightly) restrictive tone as to the classification of (new) collusive practices as restrictions by object following T-Mobile and Cartes Bancaires. It is nonetheless a stretch to read these developments as allowing parties to reverse a burden of proof that a practice is restrictive by object by showing that the practice is not capable of having anticompetitive effects, as was allowed in Intel.
Further, the examples provided by the author to support his arguments can also be used to defeat those very same arguments. It may be that the same decision can be used to directly contradict the argument being made (for example, the T-Mobile judgment which the author uses as an example of applying the Intel framework in the context of Art. 101 TFEU explicitly holds that, as regards restrictions by object, “anti-competitive effects result can only be of relevance for determining the amount of any fine and assessing any claim for damages”). In other cases, the judgments are – I would submit – better read not as being concerned with reviewing evidence of absence of anticompetitive effects, but with establishing whether the cartelists were not either actual or potential competitors to begin with as a result of the regulatory / IP framework (which was the ultimate subject of appeal in the Toshiba – Power Cases mentioned above). Establishing that parties to an agreement are at least potential competitors is relevant to understand whether a conduct is capable, by its very nature, to restrict competition – it is not concerned with the identification with the effects of this conduct.
To repeat, I think this approach mistakenly equates the categorisation of restrictions by object (which must be done by reference to the legal and factual context in which it occur – e.g. whether the supposed cartelists are at least potential competitors – and is incumbent on the claimant/competition agency) with the possibility created in Intel of the defendant disproving the prima facie anticompetitive object of a discounting practice. This reflects the fact that I don’t perceive ‘object restrictions’ to be rebuttable presumptions, which is what Intel could be said to have created. As such, I am unable to read Intel as unifying Art. 101 and Art. 102 TFEU case law.
The analysis of the implications of the Intel decision strikes me as being on firmer ground from a descriptive standpoint concerning where Art. 102 TFEU law stands, at least as regards exclusionary (pricing) abuses – even if, for the reasons outlined above, I have my doubts that such an approach is apt to cover all exclusionary abuses, let alone all abusive practices. But even here the argument seems to be more normative than descriptive. Let us think of predatory pricing under AKZO – and, particularly, let us think about the presumption that pricing below average variable cost is abusive unless there are objective justifications. The author notes that: ‘the Court explained in AKZO that the exclusion of competition is, at least prima facie, the only plausible explanation for pricing below average variable costs.’ As he notes, a similar justification could be extracted from Hoffmann-La Roche, according to which the Court adopted a per se rule of illegality because it took the view that exclusive dealing and loyalty rebates have no economic justification other than rival foreclosure. In other words, the Intel decision can be used to support an argument that a presumption of abuse when pricing below average variable costs should be susceptible to being overturned by evidence submitted by a party other than objective justifications. Personally, I think such as argument makes sense, even if I can think of counter-arguments based on legal certainty and the like. This would be an interesting argument to test before the courts, which goes to my point – the author’s is not so much describing the law as making a normative argument about how the case law should be construed, namely in such a way as to achieve the normative goal of coherence between Art. 101 and 102 TFEU.
From a normative standpoint, I would personally take a different tack to articulating Art. 101 and 102 TFEU. Formally, there are no object restrictions under Art. 102 TFEU. As such, I would not argue that we should have a distinction between object and effect restrictions concerning both Art. 101 and 102 TFEU. Nor would I argue that object restrictions are ultimately just strong presumptions that can be rebutted by evidence of absence of effect provided by the parties. Instead, I would argue that we should do away with such a strict framework altogether.
There are object restrictions under Art. 101, which can only be overturned under Art. 101(3) TFEU. But otherwise, what we have are different standards of proof and presumptions of variable strength which may be met or defeated by evidence submitted by the competition agency/plaintiff/defendant as the case may be, in a structured process of discovery of the effects of individual business conducts. The challenge – and the goal – should be to identify the presumptions and evidentiary rules that should apply to certain sets of practices within the different legal frameworks of Art. 101 and 102 TFEU. In other words, we should strive to identify legal tests that are optimally tailored to deal with certain business practices, and that are more precise than the generic formulas that must be applied to collusive and unilateral practices more generally. In a way, this is what we already do – and what Intel does for some types of discounting practices.
Due to the challenges of categorisation, and the ingenuity of businesses, this is likely to be an iterative and never-ending effort. But it also strikes me as a more realistic, and better normatively grounded endeavour, than our current business of submitting business practices to contortions worthy of top-level Chinese circus performers in order to make them fit into arbitrary ‘object’ and ‘effect’ boxes.
As you may have gathered, my disagreements with the paper do not reflect failings by the paper or its author; they reflect a legitimate disagreement about what EU competition law actually is on these matters, and as regards how we would like to see it evolve. Which is why I can say that this is an undoubtedly useful, interesting and thought provoking paper, which I’m sure will be nominated to various antitrust awards by year end.