Chillin'Competition

I hear very often that the outcome of Google depends on whether you call it a refusal to deal or a tying case. If one sees it as a refusal to deal case, then it would be necessary to establish, at the very least, that non-discriminatory access to the search engine is indispensable. If not, the threshold would be much lower (although the Commission committed to establish anticompetitive foreclosure in tying cases in the Guidance). This approach cannot be right. The legal test that applies to a practice should not depend on what one ‘sees’, or on picking the words that one prefers (and which, no surprise, tend to coincide with the legal test that one also prefers).

What should matter when thinking about the applicable legal test to a given practice is the underlying issue, and in particular whether it is closer in nature to those underlying refusal to deal cases or tying cases instead. How could one draw the line between one and the other? This is a topic that I discuss very often in class with my students. The easiest way to go about the question is to think backwards about the case. In other words, it makes sense to think first about the remedy and then about the legal test.

In technology markets, it is not always easy to distinguish between tie-ins and refusals to deal, in the same way that it is difficult to distinguish between vertical and conglomerate mergers. What is clear, on the other hand, is that remedies in tying and refusal to deal cases are very different in nature. This explains, in turn, why the legal test under Article 102 TFEU is also very different.

The typical remedy in a tying case is an example of the good old, unsophisticated competition law. The likely or potential anticompetitive effects of tying can be easily addressed by breaking the tie-in, that is, by preventing the dominant firm from conditioning the acquisition of one product to the acquisition of another. Once tying is prohibited, concerns about the ability of the dominant firm to extend (or strengthen) its position on the market for the tied product disappear.

The remedy in refusal to deal cases takes competition law out of its comfort zone. Remedial action is no longer about a one-off, proscriptive form of intervention but about positive obligations that require monitoring. If a refusal to deal is found to be abusive, it is inevitable that the remedy will regulate the conditions of access to an input or a platform owned by an integrated firm. Remedies will, as a result, greatly interfere with the way in which a company does business. These are some of the reasons why competition law has traditionally limited to exceptional circumstances the instances in which regulated access obligations are imposed on a dominant firm.

The fact that concerns are addressed by means of access obligations on regulated terms and conditions tells you something else about the underlying issues in a case. If the dispute relates to the conditions of access to a platform or input, it also means that it does not really involve two separate products, as is in principle required in tying cases (just take a look at the Guidance). The dispute, in other words, relates to a complex product that integrates different components and not to a tie-in of distinct products. What about the Internet Explorer case? Was it not about access and tying at the same time? Indeed, but bear in mind it also was a commitments decision, which did not address substantive issues. In fact, I fully agree with the view, taken by some authors, that it was a refusal to deal case in disguise.

Where does the above leave us in relation to Google? I have written several times in the blog that I suspect that the underlying issues in the case are closer to those at stake in refusal to deal cases (hence why I mentioned last week that I am not convinced that anticompetitive foreclosure would or should suffice). It seems now clear to me that the Commission does not challenge the integration of Google’s services into the search engine, which suggests that talking about distinct products would be entirely artificial. This issue became very clear when the different rounds of commitments were discussed. The Commission has now explicitly mentioned that the remedy should be non-discriminatory access to the platform.

One could wonder whether it has become meaningless to distinguish between refusal to deal and tying cases. If drawing the line between the two is so difficult in practice, would it not make more sense to require anticompetitive foreclosure across the board, and get rid of the indispensability/new product conditions? I think it would be a terrible idea. The reasons why the legal test in refusal to deal cases is so strict are in fact as compelling as ever given the shift of the discipline towards IP-intensive and technology-intensive markets.

A separate question is of course whether the indispensability and new product conditions will be watered down in EU competition law to the point that they are no longer reliable indicators of administrative action in the field. In one way or the other, and irrespective of what courts and authorities formally say, we may end up in a situation in which, for all practical purposes, the test applying to refusals to deal and to tying cases is the same. I am more inclined to agree with this question, which is not one of principle. This is a phenomenon in which I am interested and which I follow closely. Who knows, maybe I will teach my students in a few years’ time that the Microsoft saga was the first nail in the coffin of the refusal to deal doctrine and that Google was the definitive one.