Chillin'Competition

‘Self-preferencing’ is the buzzword of the day. This concept was discussed at length in the Special Advisers’ Report and is now being explored in earnest by national competition authorities around the EU. We have heard news about the Dutch Authority exploring Apple’s alleged favouring of its own app. Right before the break, the Italian authority announced the opening of an investigation against Amazon and its practices.

This flurry of enforcement activity, yet another sign of the turning tide, cannot surprise an attentive observer. Self-preferencing is a powerful rhetorical device: it provides authorities with the key to potentially countless investigations and allows them to monitor closely the activity of large firms in the digital arena. More importantly, it is through this device that they can adopt wide-ranging remedies changing, potentially in fundamental ways, some business practices.

The genius of self-preferencing as an epithet is that it turns the most pervasive and the most banal of practices into prima facie violations of Article 102 TFEU. Vertical integration is, indeed, pervasive in the digital (and the analogue) world. And instances in which vertically-integrated firms treat their affiliates more favourably than their rivals are equally pervasive. Just stroll around your local supermarket after reading this post.

Self-preferencing may become the tool to achieve ‘common carrier antitrust’ (that is, a state of affairs in which vertically-integrated firms are expected to comply with perfect neutrality obligations vis-a-vis customers and suppliers). This is an idea I explored previously on the blog.

As I was reflecting on the above, and on the dangers of enforcement-by-slogan, I was reminded of Areeda’s classic piece on essential facilities. In that piece, Areeda warned against an epithet-based approach, emphasising that it tends to lead to the expansion of a doctrine ‘to the limits of its language’, ultimately resulting in outcomes that are intellectually indefensible (Areeda’s choice of words was more colourful and perhaps more effective, but I guess you get the point).

If arguments revolving around self-preferencing are going to play a fundamental role in the coming years, it makes sense to ask some hard questions by reference to first principles. And it is also important to face the consequences of certain choices.

Self-Preferencing has never been seen as a concern, but as an expression of competition on the merits

It may be hard to believe, but there was a time when vertical integration and, by the same token, firms favouring their affiliates were not deemed problematic in themselves. They were not considered suspicious conduct deserving close scrutiny but the natural manifestation of the competitive process.

One of my favourite examples of the traditional approach dates back from the days when people smoked. In Tabacalera v Filtrona, the latter firm challenged a self-preferencing decision by the former (at the time, the tobacco monopoly holder in Spain): Tabacalera had decided to favour its own affiliate for the production of cigarette filters.

The Commission explained in its press release that this form of self-preferencing was not in itself an ‘abnormal act of competition’. In fact, few things are more normal than a firm relying on its affiliate to meet its own needs for a particular product (or favouring it when in doubt).

A second example (another one of my favourites) along the same lines is provided by the Guidelines on non-horizontal mergers. Again, self-preferencing is not presented in the document as something justifying remedial action, in and of itself, after a merger. It is instead presented – and was examined in individual cases – as an expected outcome of  a vertical merger, which may be problematic only where certain circumstances are met.

Key takeaway: I guess the main message here is that self-preferencing cannot be seen, in and of itself, as a departure from competition on the merits. Questioning this practice, in essence, is tantamount to questioning vertical integration and the rationale behind vertical integration – and I do not think anybody is ready to suggest (or at least not yet) that vertical integration is inherently bad, or presumptively bad.

Thus, if self-preferencing is going to dominate enforcement, we need to think carefully about the legal conditions under which it is deemed unlawful (i.e. the legal test) and the rationale behind such conditions. Which takes me to my next point.

Indispensability: the hard question that cannot (and should not) be avoided

I have explained many times on the blog that I am not interested in the outcome of individual cases. What matters, in my view, is the analysis, and how the analysis relates to the applicable case law. And I noted, when I read the Google Shopping decision, that none of the hard questions about the legal status of self-preferencing were addressed.

The Commission never spelled out in the decision the legal test against which the lawfulness of self-preferencing was assessed. It merely explained that Article 102 TFEU can apply to leveraging practices. This is not only correct but wholly uncontroversial. The problem is that (i) it is not a legal test and (ii) fails to address any of the issues.

If you have not done so, take a look at paragraph 334, which, according to the decision, provides the case law underpinning intervention. The Commission refers to several cases in footnotes 350 and 351. A cursory look at these judgments shows that there are virtually as many judgments as there are legal tests:

• CBEM-Telemarketing: in this case (the first to be mentioned by the Commission), the legal test requires evidence of indispensability (the reference to indispensability is explicit in the judgment, a point expressly confirmed in Bronner).
• Tetra Pak and Microsoft (Media Player): as the law stands, tying is prima facie unlawful under Article 102 TFEU – the practice is prohibited in and of itself (bearing in mind that it is always possible for a firm to rely on Murphy/Intel to rebut the underlying presumption that the practice is capable of restricting competition).
• TeliaSonera: a ‘margin squeeze’ is only prohibited where it has, or is likely to have, anticompetitive effects.

As you can see, the outcome in Google Shopping is perhaps correct. The difficulty is that we do not know which of the above three legal tests was the applicable one in the case, as the question was never tackled as such in the decision.

The uncertainty in this sense will have to be addressed, sooner or later, by the EU courts. It cannot be dismissed simply by arguing that leveraging has in the past been found to be abusive. For the same reasons, the question of whether indispensability is a legal condition in some, or all, self-preferencing cases will have to be addressed.

If you ask me: I always say that Google Shopping is just a re-run of Commercial Solvents: a firm, following a decision to vertically integrate, engages in a raising rivals’ costs strategy. But bear in mind Commercial Solvents, like CBEM-Telemarketing, required indispensability (and for good reasons).

Key takeaway: Perhaps it makes sense to depart from Commercial Solvents. Perhaps the reasons behind the introduction of the indispensability requirement in that case (and in CBEM-Telemarketing) are not compelling anymore. Perhaps one should not require indispensability in the specific circumstances of the case (for instance, it makes sense not to not to require indispensability in TeliaSonera/Slovak Telekom scenarios, as I explained here). 

But these questions have to be addressed explicitly. Avoiding the inevitable for a while can only lead to uncertainty and inconsistencies in the case law.

The need for a robust analysis of anticompetitive effects

One could argue that, so long as we can show that self-preferencing is potentially a source of anticompetitive effects, the above discussion is largely irrelevant. This is a very reasonable position to take. There is only one problem: it all depends on how anticompetitive effects are defined, and how they are assessed in practice.

This is a question that keeps me busy – and to which I devoted my talk at the last Chillin’ event. As I explained then, if the threshold of effects is low, pretty much any behaviour would be in principle prohibited, and would be treated, in practice, just like ‘by object’ conduct. The requirement to show anticompetitive effects would be a mere formality.

If we understand anticompetitive effects to mean ‘any competitive disadvantage’, self-preferencing will be, always and in any circumstance, prima facie abusive. Complaints about self-preferencing are, in fact, complaints about the affiliate receiving a competitive advantage.

We would face the same situation if an authority were able to discharge its burden of proof by showing that anticompetitive effects (however defined) are plausible. Such a low threshold would be met in the vast majority of cases. After all, the reason some practices deserve scrutiny is because they are deemed a plausible source of anticompetitive effects.

I believe it is possible to infer from the case law what is meant by effect. However, there seems to be a disparity between the case law and the practice of competition authorities. My work over the years has taught me that authorities have a tendency to equate any competitive disadvantage with anticompetitive effects, and tend to prefer to set the threshold of effects at the level of plausibility.

This disparity will inevitably arise in self-preferencing cases. And, like the indispensability issue, the question of what we mean by an anticompetitive effect will have to be tackled directly and in a clear and meaningful way.

Key takeaway: Saying that self-preferencing is abusive where it is a potential source of anticompetitive effects does not solve any of the problems discussed above. Unless the notion of effects is defined, considerable uncertainty, and considerable scope for opportunism, will remain.