From Guidance to Guidelines: my presentation on exclusionary abuses
We had a great discussion on exclusionary abuses last Tuesday at LSE. We were delighted to see how many of you joined us online to what was (at long last) a Chillin’ event (rest assured there will be many more to come).
During the event, I shared a presentation outlining my main ideas, which you can find here. I would very much welcome your thoughts on the main points I made, which I can summarise as follows:
Enforcement under the Guidance is in a much better shape: the days of Michelin II and British Airways are long gone, which is why we tend to forget how much in need of reform Article 102 TFEU policy was at the time.
In the years that have followed the Guidance, enforcement has remained robust (both in qualitative and quantitative terms), and has become wiser (in the sense that the intellectual foundations are much more solid) and more predictable (again, we tend to forget how unpredictable enforcement was in the old days).
The analysis of effects is not a luxury or an indulgence from which we can dispense. It is the key tool to distinguish between conduct that is on the whole pro-competitive and behaviour that can be expected to harm the competitive process. One should not forget that the majority of practices potentially caught by Article 102 TFEU are, most of the time, normal expressions of competition on the merits.
The above said, it is worth reflecting on some of the criticisms that have been directed at enforcement under the Guidance. Is the analysis of effects overly cumbersome? Is intervention difficult to tell in advance?
Structured legal tests could be part of the answer to these concerns while preserving the essence of the Guidance. From this perspective, the forthcoming Gudelines provide a good opportunity.
Contrary to what is sometimes said, the evaluation of the actual or potential anticompetitive effects of a practice need not be open-ended or all-encompassing. It can be structured around a number of solid proxies and finite lists of criteria. This, in fact, is what the Court of Justice has been doing over the past decade (just think of Intel).
The case law could be codified in line with this principles. For instance, there is little doubt that the coverage of a practice is central to the assessment of the actual or potential impact of a potentially abusive practice. Why not rely on a bright line (say, 30% coverage) to define the instances in which effects are more likely?
If effects can be established always and everywhere, then the test of effects is a bad one. Any attempt to recalibrate policy should not come at the expense of predictability. If the test is reconfigured so that the analysis of effects becomes little more than a formality, it will not be possible to anticipate enforcement (the Commission would always find a way to show an impact on competition).
In the same vein, I explained that the hard questions cannot be avoided. These hard questions include:
- The meaning of effects.
- The threshold of effects. We all agree that Article 102 TFEU is concerned with actual or potential effects. But the notion of potential effects does not answer the question of what threshold applies. AG Kokott suggested that the case law demands a threshold of likelihood (probability of >50%) when looking at potential effects. I agree with her understanding of the relevant rulings.
- Causality: The difference between competition law (understood as a system aimed at protecting the competitive process) and economic regulation aimed at restructuring markets so that they conform to the particular vision of an agency is the issue of causality. As emphasised by the EU courts in recent years, Article 102 TFEU is only triggered when the actual or potential effects are attributable to the behaviour of the dominant undertaking.
I very much look forward to your comments. As you know, I have nothing to disclose.
Chillin'Competition 
What do you mean by « coverage »?
Clément
27 April 2023 at 7:21 pm
I mean: we are talking about 102 cases, wouldn’t « coverage » equal to market shares in most instances?
Clement
27 April 2023 at 7:29 pm
Thanks, Clement!
Great and very useful question. The coverage of the practice is not the same as the market share of the dominant company.
You can have a dominant company with a market share of, say, 80%, and a practice with a coverage of 10%.
Suppose, for instance, that the dominant company I describe above implements a system of loyalty rebates, but that this system only applies to customers that represent 10% of customer demand on the relevant market.
Pablo Ibanez Colomo
27 April 2023 at 9:06 pm
Hi Pablo,
Great post & talk as always! Just a few points:
1. I am not at all sure that enforcement has remained “robust” or that the Commission simply reacts to some defeats as Jorge mentioned in the discussion. The very existence of the DMA is an explicit recognition that antitrust enforcement came too late (perhaps 10 years after the opening of some seminal cases) and did too little. Name a case where the remedy fully satisfies you. In most cases, I would expect you to be dissatisfied with the remedial response. So I am not sure on what basis you ground your statement that enforcement has remained robust and I am wondering what are your benchmarks.
2. I agree with you that the Court has done a lot of good work to clarify the law (hence “wiser”). But I am not sure about predictability. I see many commentators disagree about the role of the AEC test (whether and when it is necessary). But on top of this, the specificities of how the test is to be applied remain unclear (e.g. how much of the market has to be contestable to raise concerns?). In addition, as you mention the exact meaning of anticompetitive effects and the standard of proof remain a tad unclear. Equally unclear is what exactly counts as procompetitive and when such effects can counter or justify a prima facie abuse (see, for instance, Google Android where the Court engaged in a rather limited discussion of the objective justification ground).
3. The exact role of the AEC test as part of a theory of harm is also not clear. Even if we find a negative result of the AEC test does this mean that the conduct is an abuse of dominance? Monti discusses this question here: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4056280. For instance, if rebates foreclose one as-efficient rival by foreclosing access to one key customer would that be enough to establish an abuse? The rival can try and sell to others and consumers most certainly will not be harmed. On this basis, Giorgio says “the AEC test does not operate as a stand-alone smoking gun to show illegal conduct: the rebates still need to be assessed in their wider economic context”. But then the Court has not explained how the Commission is supposed to do that. So again, it seems to me that the law lacks clarity and the Intel test is less structured than you would hope for.
4. Another problem of the AEC test is that it is underinclusive. Even if in theory a discount of 5% could be offered by an as-efficient rival as Domco, there may be no such efficient rival out there (see PostDanmark I). But there could be promising firms, new entrants, or start-ups that could evolve to become efficient if they were not priced out. In this setting, the AEC test produces undesirable results from an economic perspective: it allows for practices that harm consumer welfare. So more work needs to be done to find out under what conditions a Domco can hurt less efficient rivals and by doing so harm competition and consumers. In this regard see a recent decision by the Greek NCA here https://www.epant.gr/en/enimerosi/press-releases/item/2415-press-release-imposition-of-a-10-342-773-fine-on-coca-cola-3.html.
5. Many practices in digital markets may have no price effects (for a nice illustration here: https://blogs.law.ox.ac.uk/business-law-blog/blog/2022/07/how-big-tech-barons-smash-innovation). In this setting, the AEC test is of rather limited value there (even though some studies raise some optimism https://academic.oup.com/jeclap/article/13/2/125/6542017). So I am curious if you have any tests in mind to rectify this problem, especially in light of the forthcoming 102 Guidelines.
6. Can you tell me a bit more about how you understand causality? What are the available options and how can we ensure that we do not pose on the Commission a standard impossible to meet? I have a similar concern with your 30% coverage threshold. Are there any economic studies that support such a threshold? In theory, it seems reasonable but how practical is it? For instance, if all customer demand in a market is 10,000 outlets and a widget seller offers suspicious rebates, the Commission will have to collect evidence from 3,000 stores to build its case. How realistic is that?
Apologies for writing too much – Your post & presentation triggered lots of thoughts! I am looking forward to reading your views!
Stavros Makris
28 April 2023 at 1:07 pm
thanks so much, Stavros! All great points
The DMA was not in any way a recognition that competition law enforcement came too late. If a market is structurally unsuited for competition, no amount of Article 102 TFEU enforcement is ever going to change that.
The DMA is a recognition that competition law is not, and has never been, a toolkit for the systematic regulation of markets.
If you look at what Article 102 TFEU has achieved across the board, it is quite impressive, in terms of breadth and depth.
On the AEC test: as I mentioned during the talk. The AEC test is just a proxy; it is not central to the operation of the regime. And I can see it is becoming a distraction.
On causality: I will write a full post on it. Let me mention on the coverage that I would not see why you need to knock on the door of 3,000 separate stores. If they have concluded the agreement with the dominant undertaking, I suspect all those contracts would be with the dominant undertaking.
More fundamentally, on coverage: does it even make sense to argue that a practice has potential effects on competition when its coverage is minimal? It would make no sense, which is why it has become a central consideration in the case law.
Pablo Ibanez Colomo
29 April 2023 at 8:16 am