NEW PAPER | Self-Preferencing: Yet Another Epithet in Need of Limiting Principles
I have just uploaded on ssrn a paper on Self-Preferencing (see here). Those with a good memory may remember that the paper started its life as a blog post. Given the topicality of the issue and its growing importance, I felt it made sense to expand some of the arguments and revisit other arguments explored in other recent papers of mine.
I advance three main ideas that I believe were worth emphasising in the current environment.
The first is that self-preferencing is, in itself, a manifestation of competition on the merits. Not only is there nothing inherently suspicious in a firm favouring its affiliate, but often it is inextricably linked to the pro-competitive benefits expected from horizontal or vertical integration.
The second idea is that self-preferencing is not a sound legal category. Legal categories are necessary. However, not all of them are useful or meaningful. The fundamental problem with self-preferencing is that it is a catch-all concept that brings together behaviours that are fundamentally different.
In addition, it overlaps totally or partially with well-established categories. In this sense, it is not clear what its added value may be.
Suffice it to think about it for a second. Tying and bundling are forms of self-preferencing (in essence, tying amounts to giving a competitive advantage to an affiliate). A refusal to deal is also a variation on the same theme: the integrated firm refuses to deal with rivals because it intends to favour an affiliate.
These practices are also very different if one takes into account the nature of the intervention to bring the infringement to an end. We know from experience that intervention in some cases requires a court or an authority to change the design of a product and/or change a firm’s business model. Other cases can be addressed by means of a one-off negative obligation.
As you can see, it is difficult to see how the competition law system will improve with the introduction of a legal category that overlaps with existing ones and that obscures the differences between them.
With takes me to my third point: if self-preferencing is going to be embraced, the system would benefit if the implications of doing so were carefully pondered.
I have discussed Van den Bergh Foods in previous papers (see in particular here). Its implications are clear: where intervention in a case would involve forcing a company to transfer an asset or to enter into agreements with third parties, indispensability is an element of the legal test.
This case law makes sense. Indispensability is a sensible filter to limit the instances in which the system is required to the tricky and error-prone tasks of redesigning products and/or changing business models. The institutional implications of abandoning this filter are substantial, and they would need to be addressed.
Finally, there is much talk these days about making self-preferencing prima facie unlawful. This is another issue that deserves careful thought. From a positive perspective (in particular following Generics and Budapest Bank), self-preferencing is not the sort of practice that can be treated as a by object infringement.
From a normative perspective, a robust effects analysis, at the very least, seems necessary. Turning self-preferencing into a by object (or quasi-by object) does not come across as optimal.
I very much look forward to your comments. If you were wondering, I am happy to clarify, in accordance with the ASCOLA declaration of ethics, that I have nothing to disclose.
Chillin'Competition 
Many thanks for your thought-provoking analysis, Pablo. I was at first surprised by the definition of the conduct in your paper (at p2): as a situation where ‘an instance in which an integrated firm favours its affiliates at the expense of rivals’, hinting at the need of a separate legal entity which rivals may benchmark their terms against. I thought at first that might not be necessary, but perhaps it is, or should be. Let me explain this a bit more.
I am perhaps naively assuming that self-preferencing is not proposed as a general rule for all markets, but one that would apply in a subset of cases. That category has not been defined other than by flagging it in the digital debate, and I could not agree more that it would be important to define ‘meaningful criteria to identify the conduct which would be categorised and scrutinised as such.’ (p24).
I wonder if that category should be restricted to markets where, as you put it, ‘as a result of the underlying economic context, a dominant firm finds itself in a position that is analogous to that of an undertaking that benefits from exclusive rights or other regulatory privileges granted by the State.’ (p9), a logic that may apply independently of whether ‘the edge that a firm enjoys is the result of government action by means of, for instance, the award of exclusive rights’ or results from other elements, including successful competition on the merits.
As in the telecommunications field, the solution might involve mandatory legal separation (hence my suggestion above) and non-discriminatory access. In that sense, the ban on self-preferencing would belong to the margin squeeze family. Whether, or more precisely, in what proportion should this be done via regulation or classic competition tools is an relevant question, but that is a whole other discussion. But again, as you note, all this should be defined in a much clearer way than what we have seen to date.
Marcos Araujo Boyd
18 July 2020 at 8:44 pm
Thanks, Marcos, for the insightful comment!
I have the impression we see eye to eye on the matter. As you say, self-preferencing is the sort of concern one observes in telecommunications markets. Insofar as one can identify separate levels of activity, one can think of separate entities (at least in theory).
It also seems to me that the idea is to confine the scope of the doctrine to a subset of cases: the question that follows is how to define such a subset. This is where the discussion on indispensability becomes relevant. And perhaps why the remedy may make a big difference in practice.
Pablo Ibanez Colomo
20 July 2020 at 4:57 pm