The draft Article 102 guidelines - initial impressions

143rd GCLC Lunch Talk [1]

September 10 2024

Giulio Federico [2]

Speaking Points [edited for clarity]

?

Premise

Let me start by the premise that Article 102 is one of the hardest areas in competition law. The jurisprudence is complex and evolving (and not always consistent), the economics is complicated (since unilateral conduct is of course much harder than horizontal conduct/agreements/mergers), and there is often only partial consensus on what should be the right approach.

It is appropriate that the European Commission takes the initiative (again, 16 years after the publication of the Guidance Paper) to bring it all together, and ideally find a practical and coherent way forward – a workable common ground in light of recent jurisprudence and modern economics.

For reasons that I will explain, I do not think that the current draft guidelines do a particularly good job at finding this common ground and synthesis. But it is also important to recognize upfront that this is a challenging task, and it is natural that there will be quite a bit of back and forth. In this context, I look forward to an open and constructive consultation process on the guidelines.

Discussion topic 1: what is your take on the most important features of the draft 102 Guidelines?

Let me highlight three features.

(1) Policy implications: reliance on an effects-based approach and on the use of economics

This draft signals a significant policy shift, away from approach taken in the 2009 Guidance Paper.

This is not that surprising, since there has not been much ownership of the Guidance Paper at the Commission for past 10+ years, but nonetheless the policy shift is quite prominent. ?

Some of the key elements of the Guidance Paper - notably the definition of anti-competitive foreclosure (i.e. foreclosure of competitors that ultimately leads to consumer harm in form of higher prices / lower quantity / lower innovation) as a guiding principle, and the focus on likelihood of anti-competitive foreclosure, are not retained in the draft Guidelines. Nor is the careful exposition of economic theories of harm (e.g. for exclusive dealing and for predation) which is contained in the Guidance Paper.

The concept of anti-competitive foreclosure is now replaced with some high-level references to consumer harm (at paragraphs 5 and 51), which are then not repeated or weaved into any of the substantive tests or discussion of each conduct. My concern is that these two references alone, whilst helpful, will not be sufficient to support the articulation/development of explicit theories of harm for each conduct.?

The draft guidelines (as they stand) would de-facto row back on the attempt to modernize the enforcement of Article 102 and to frame it within a more economic approach, which was one of the fundamental objective of the 2009 Guidance Paper (in an effort to align Article 102 to the modernization of other instruments, e.g. merger control and Article 101).

This is unfortunate (and hopefully can still be corrected in the final guidelines), because, if properly applied, the use of sound economics in Article 102 enforcement can actually lead to more effective, focused, and coherent investigations (and hence also more efficient investigations, which is one of the objectives of the Commission’s drive towards a "workable" effects-based approach).?

(2) Is the two-limb test proposed in the draft guidelines too open-ended, and does it provide sufficient guidance to firms?

One of the prominent feature of the draft guidelines is the reliance on a two-limb test:

o?? Limb 1: does the conduct represent competition that is not on the merits?; and

o?? Limb 2: is the conduct capable to lead to exclusionary effects?

This test is based on recent case law (notably Servizio Elettrico Nazionale (SEN)) - ?even though the draft guidelines do not fully apply the principles set out in SEN (I will come back to this in my 3rd introductory point). ?

My concern is that, as presented in the draft, this approach it will lead to a unduly open-ended substantive test, with insufficient clear limiting or guiding principles, and de facto limited guidance for firms (undermining one of the key objectives of the draft guidelines).

This is in part because of the departure from the “As-Efficient Competitor“ (AEC) standard (to which I return below).

The lack of a de minimis threshold on effects (see paragraph 75 of the draft guidelines) is a particular source of concern, as it is not in line with most economic theories of anti-competitive foreclosure, and actually arguably not in line with more recent jurisprudence.

There is also an apparent inconsistency between the standard for objective justifications and efficiencies, and the standard for effects under Limb 2, which would inevitably tilt the balance of Type 1 and Type 2 errors.

I will expand on these concerns in our discussion later on, in relation to the general framework and evidentiary standards. ??

(3) The role of AEC standard and of the AEC test

Of course, in light of the recent jurisprudence, the draft has to deal with the use of the AEC test, and the more general concept of the AEC standard (namely, that foreclosure is anti-competitive only if it is capable of excluding as efficient competitors).

The AEC standard

A remarkable feature of the draft is the relatively limited role played by the AEC standard, given prominence of this concept in the case law, especially in the wake of the Court’s judgement in Intel in 2017 (e.g. see SEN, Super League, and as of today, also Google Shopping).

The AEC standard is a difficult concept to apply and one can argue that it has actually become too prominent in recent case law. However, ?given that the draft guidelines profess to codify the case law, its absence as a general principle (for both Limb 1 and Limb 2 of the proposed test) is surprising.

In terms of Limb 1, the AEC standard is the guiding principle behind the definition of “competition on the merits” set out in SEN (and in the Advocate General’s opinion in that case), since this concept is primarily based on replicability by an AEC (i.e. could an as-efficient competitor profitably match the conduct of the dominant firm?). This is the way in which SEN makes sense of previous case law. This also a reasonable way as a matter of economics to define anti-competitive conduct (under Limb 1), before looking at exclusionary effects - to ensure that we are looking at the “right” effects (i.e. not just any exclusionary effect, but only anti-competitive ones).

In relation to Limb 2, not only is AEC standard not adopted in the draft guidelines, but it is actually rejected as a relevant factor (see paragraph 73)!

The use of the AEC standard would not only be more consistent with the jurisprudence, but it would also be a reasonable way to define more precisely both Limb 1 and Limb 2 - possibly with some well-defined exceptions (e.g. naked restrictions), building also on some of the recent case law (e.g. Superleague).

The AEC Test

With respect to the AEC test (i.e. a quantitative price-cost test) the draft clearly indicates its use only for price-based abuses (predation; rebates; and margin squeeze) but not for exclusivity rebates.?

I think that this is right as a matter of economics and policy, and a helpful clarification compared to Guidance Paper (and the use of the AEC test in the Intel case), which had been subject to some ambiguity.

However, the departure from the results of the AEC test even for price-based conduct is not helpful, in particular as a general principle under Limb 1 (see paragraph 57, where the draft guidelines state that pricing above ATC may depart from competition on the merits), and in the specific example of rebates (see paragraph 144(b) and footnote 325).

For exclusivity rebates, even if the AEC test not adopted by the Commission, it will still likely feature in any discussion on replicability on conduct by an AEC, so some guidance on how the Commission would approach this issue would still be required. ??

Any departures from AEC test for price-based conduct needs to be ring-fenced much more carefully in the guidelines, in order not to chill pro-competitive conduct (e.g. at least the intent to exclude should be part of the evidence required in any case involving above cost pricing).

Discussion Topic 2: The general framework distinguishing “conduct departing from competition on the merits” and “capability to produce exclusionary effects”

As I mentioned earlier, the combined effect of the two limbs as defined in the draft guidelines may lead to quite a vague standard for intervention – providing limited guidance and legal clarity, and not enabling a proper self-assessment by firms.?

In relation to Limb 1 (competition on the merits):

• besides the general reference to consumer benefits (paragraph 51 – which is helpful, as mentioned earlier), the draft guidelines offer very limited general principles, and primarily rely on a (non-exhaustive) list of factors to take into account, and on legal tests for specific conducts.
• The draft guidelines also depart from SEN, since replicability by an AEC is not the general guiding principle behind the definition of competition on the merits, but it is just one possible factor relevant for assessment.
• My concern is that the assessment of Limb 1 will either be quite mechanical (relying on proposed legal tests for specific conducts) or unduly open-ended.
• What would help here instead is either a very clear statement on the fact that only conduct that is likely to lead to consumer harm (i.e. anti-competitive foreclosure, as defined in the Guidance Paper) would be found to depart from competition on the merits; or the more explicit adoption of the entire framework put forward in SEN, notably including the adoption of the AEC standard and the related notion of replicability.

In relation to Limb 2 (capability to produce exclusionary effects):

• The framework presented in draft guidelines is more standard (e.g. the list of relevant factors at paragraph 70 is based on paragraph 20 of the Guidance Paper), but, in large part due to the loss of the notion of anti-competitive foreclosure (i.e. paragraph 19 of the Guidance Paper), it is not clear which effects we are to worry about under Limb 2? Just effects on as-efficient competitors, or any competitor? Would any shift in market share in favour of the dominant firm qualify under Limb 2? And what is the evidentiary bar for findings effects?
• Under one interpretation of the draft guidelines, any (largely) theoretical effect on any competitor compared to a plausible counterfactual amongst several possible outcomes, with no de minimis threshold, could satisfy Limb 2.
• The other concern here is that the generic reference to fact that Limb 1 is satisfied may not be sufficient to ensure that we are only dealing with anti-competitive effects under Limb 2.

Overall therefore the risk is that the combined effect of Limb 1 and Limb 2 (as defined in the current draft) could support a unduly low and open-ended standard for intervention, de facto based on the form of the conduct, and not its (likely) effect on consumers.

One way to address this concern would be to tighten the definition of Limb 1, and introduce the AEC standard (in line with the recent case law), with well-defined exceptions where the AEC standard can be departed from (but not as general rule).?

Discussion Topic 3: The evidentiary requirements/burden to demonstrate a conduct’s capability to produce exclusionary effects, including causality, presumptions, naked restraints, “object abuses”

Let me focus first on presumptions, and then say also a few words on causality and standard of proof.

Presumptions

In principle, presumptions are difficult to apply to unilateral conduct, but as a matter of economics can work – but of course the standard and process of rebuttal would be key here.

We will need to see how the proposed approach to presumptions and rebuttals will work out in practice, but some of the language in the draft guidelines under Limb 2 (e.g. paragraph 60, which sets out the need for “substantially different circumstances from background assumptions”) seem to go beyond approach set out in case law, e.g. Intel.

Besides this introductory point, my main concerns are on how to trigger the presumption and on its practical implications.

On the trigger for the presumption, I think that the issue of appreciability of effects should be part of what the authority would need to prove to trigger the presumption. I say this because it does not make sense to apply a presumption of anti-competitive effects if a conduct (say, exclusivity, tying or predation) only cover a very small share of the market.

Such a conduct is unlikely to impair the ability of a competitor to compete for the rest of the market, and hence its ability to constrain the dominant firm.

In this vein, paragraph 75 of the draft guidelines (on the lack of de minimis, and on the statement that any exclusionary effect constitutes a further weakening of competition) is troubling and generally incorrect as a matter of economics.

It appears to fundamentally mis-interpret the logic of anti-competitive foreclosure. That is, under many standard economic models of foreclosure, the customer dealing with the dominant firm and that is subject to the conduct (say, an exclusive deal) actually benefits from the conduct (i.e. often the customer needs to be compensated for the restriction / exclusivity), and other customers suffer because of the resulting weakening of competitors (e.g. this is known as a "divide & conquer" strategy).

But this mechanism only works if the coverage of the conduct is sufficiently extensive to hinder a competitor. If coverage is low, it is hard to see how conduct can (eventually) lead to consumer harm and hence be anti-competitive.???

My other concern is related and of a practical nature. If the authority needs to do very little to trigger the presumption (e.g. as in the proposed approach for exclusive dealing), the risk is that it will not investigate the market sufficiently, and that it will be unable to build, articulate and validate a plausible theory of harm.

This can in turn lead to weaker cases and quite possibly issues in Court. A case at hand in my view is the Qualcomm exclusivity case, where the Commission de facto applied the approach to exclusive dealing put forward in the draft guidelines, and it lost the case at the General Court arguably because of the lack of a theory of harm, and of evidence for it (besides the procedural issues noted by the General Court).

Evidentiary standards

On the standard of proof, the issues of balance of probabilities and of the appropriate counterfactual are as usual key.

The draft guidelines are explicit in not taking a “balance of probabilities” standard (i.e. the draft guidelines don’t say for example that foreclosure effects need to be more likely than not, compared to the counterfactual absent the conduct).

This not an issue in principle – recall for example the Court of Appeals in Microsoft (2001), indicating that “it would be inimical to […] allow monopolist free reign to squash nascent, albeit unproven, competitors at will”. ?This is also clear from US and EC jurisprudence on pay for delay.

But it is not clear what the standard proposed in the guidelines actually is (see section 3.3.2). The logical alternative to a balance of probabilities standard would be a “balance of harm” approach (i.e. can the authority show that overall consumer welfare is likely to fall because of the conduct)? However, the draft guidelines do not adopt this standard either, and do not refer at all to consumer welfare in the discussion of the standard for effects.

Under the standard adopted instead it appears that any increase in the probability of exclusionary effects will do (i.e. the draft indicates that? “it is sufficient to establish that conduct contributes to likelihood of exclusionary effects” (paragraph 65) and that “it is sufficient to establish a plausible outcome against various possible outcomes” (paragraph 67)), with no connection with likely impact on consumers, and size of possible harm.

This is problematic as it may again support in practice a formalistic approach to the assessment of effects.

It is also an issue that the same standard of proof does not appear to be used for efficiencies and objective justifications, as the standard of evidence between pro- and anti-competitive effects needs to be symmetric (as a matter of logic), to properly balance Type 1 and Type 2 errors.?

Discussion Topic 4: The conducts subject to specific legal tests, in particular the selection of the five practices in question and the representation of the tests and their consequences

The draft guidelines present legal tests for five practices: exclusive dealing; tying; refusal to deal; predation; and margin squeeze.

If the legal test is met, both limbs are fulfilled (so that the legal test de facto replaces the two-limb test).

Two of the practices are subject to a full presumption of anti-competitive effects (exclusive dealing and predation), if the legal test is met.

Two more of the practices are subject to a partial presumption of effects (certain forms of tying; certain types of margin squeeze), which de facto means that part of the legal test (the part on effects) is simply presumed.

As a first general comment I would note again that there is very limited discussion in the draft guidelines of the underlying theory of harm and the overall anti-competitive rationale of each practice (including the incentives by the dominant firm) - this is a notable departure from the Guidance Paper.

In terms of specific comments on the treatment of each conduct subject to a legal test, here are some initial (non-exhaustive) thoughts:

• Exclusive dealing. The proposed legal test here seems very light (simply a formal confirmation that the practice exists?) and, as mentioned, at least the appreciability of effects (e.g. coverage) should be part of the legal test in my view. Also, even if the AEC test is not carried out, in practice, the agency will need to explain why competitors cannot match the exclusive deal offered by the dominant firm (e.g. why is competition in exclusives not possible?; is there some form of leverage?). As in Guidance Paper, it would also be desirable if the underlying theory of harm to consumers were set out in the draft guidelines.
• Predation. The discussion in the draft guidelines is relatively standard, but also in this case the coverage/appreciability of the practice and the underlying theory of harm should be part of assessment (e.g. what is the mechanism by which prices will eventually be raised to harm consumers? are there barriers to re-entry for competitors?). This does not imply the need to prove recoupment, but it would require an articulation of the anti-competitive mechanism at work. It is positive that the principle of legal certainty is recognized in the draft guidelines (see paragraph 117), but it is unclear why the same is not done for other similar pricing practices (e.g. rebates).
• Tying. The application of partial presumption here is quite obscure,? and it is not clear why this is necessary. My main policy concern though is that, especially in the digital economy, product innovation / enhancements may take the form of an apparent tie (i.e. a product is improved with features of another product), so the risk is that application of a presumption here (or more generally a restrictive approach) may deter pro-competitive product innovation. The draft would also benefit from a more explicit articulation of the theories of harm at work, also in light of recent cases (e.g. Google Android, and the related recent case in the US).
• Refusal to deal and margin Squeeze. The main economic issue here (as noted by several commentators in the past) is that constructive refusals to deal (i.e. margin squeeze; self-preferencing; “access restrictions”) are treated differently than outright refusals to deal, with a looser legal test. This may lead to endless debates on which box to place a specific conduct in, rather than on the underlying theory of harm and mechanisms. I recognize that the case law on this topic is quite prescriptive and complex (as confirmed by today’s judgement on Google Shopping), but there is probably some scope here for the Commission to take some leadership and present a more unified approach. It is also a pity that the useful discussion that was found in the Guidance Paper on the role played by regulation (both in terms of the assessment of indispensability and of the incentives to foreclose) is not retained in the draft guidelines, since many of cases involving margin squeeze related to regulated sectors.

[1] Panel discussion with Inge Bernaerts (DG Competition) and Giorgio Monti (Tilburg Law), chaired by Damien Gerard (Belgian Competition Authority).

[2] Partner, Oxera. All views in these speaking points are personal, and do not necessarily represent those of colleagues at Oxera. I currently do not represent any parties involved in Article 102 proceedings.? These speaking points build on the useful discussion with Lluis Saurí, Massimo Motta and Helen Jenkins at the Oxera webinar on the draft Article 102 guidelines of September 4 2024.