Chillin'Competition

A comment on the suspension of the Bundeskartellamt’s Facebook decision seems like a perfect fit for the start of the new academic year at Chillin’Competition. The Order is truly a must read for anyone interested in competition law, particularly in digital markets. The fact that it’s written in German complicates that a bit. Let’s hope this helps. This post is lengthy, but that’s the price to pay for not reading the full original.  If short on time, go to the highlighted bits.

We already commented on this case at the time it was opened (see here) and decided (see here). Last week the Higher Regional Court of Düsseldorf suspended the decision pending a final decision on the case, expressing “serious doubts” as to its legality and using some pretty strong language. This development seems to have surprised many. Not so much us. If anything, we are pleased that it is very much in line with our understanding of the law, as consistenly expressed in this blog. 

After reading an excellent Twitter summary (here), I couldn’t help spending some of my last hours of holidays reading a pretty good Chrome translation (available here:) of the German original version and writing this post. The quotes used in this post are based on that automated translation (so please check against the original) and on the input of my colleague Konstantin Jörgens. To help find references, my comments also refer to the numbering of paragraphs in the translation (not present in the original). Note also that the Order discusses German law, but that it relies on principles common to EU Law (and makes an ironic(?) reference in passing to “the desired alignment of national competition law with that of the Union” (para. 29)).

The Court suspended the decision arguing that even a summary examination of the factual and legal situation leads to the conclusion that it will be set aside (para.25). Its (annotated) reasoning follows. [In Part II we will build on these elements to discuss why the Order is a perfect illustration of sensible and necessary judicial review, and by no means an obstacle to proper enforcement in digital markets].

Is there an exploitative abuse? Our first comment on this case said that “admittedly, and theoretically, the Bundeskartellamt could build an exploitative case alleging that Facebook sets infra-competitive privacy terms and conditions. However, this does not seem to be the reasoning underlying the investigation. Perhaps this has to do with the difficulties in determining which is the “competitive” level of privacy (…) possibly in the light of these difficulties the authority is prepared to take a shortcut, automatically equating an alleged “violation of data protection provisions” by a dominant company with an abuse of dominance.

The Court shares the same view, and that’s essentially why it suspended the decision. It underlines that an exploitative abuse may take place when a dominant company imposes business conditions that differ from those which would likely result from effective competition. In line with our first post, the Court understands that there is in theory no reason why one could not run an exploitative case in relation to privacy policies. Crucially, however, it rules that “the [Bundeskartelamt] did not carry out sufficient investigations into an “as if competition” and consequently did not provide any meaningful findings on the issue of which conditions of use would have formed in the competition” (para. 27; the Order comes back to the counterfactual also later at para. 47). Facebook did not have to show what the competitive level of privacy would have been; it was for the Bundeskartellamt to look into it, but it didn’t.

Failure to assess the counterfactual, again. As repeatedly held by EU Courts, and as you will have read us write a thousand times, a proper counterfactual analysis is the best sanity check for any given theory of harm. Contortions to avoid the sanity check suggest that the authority itself is aware of the pitfalls of its case. A competition authority may have a margin of appreciation in conducting complex analysis, but for that very same reason it cannot entirely do away with them. This logic, by the way, is very much in line with the tendency we see in EU Courts (see e.g. here). Most of the discussion that follows, regarding causality, is also in essence about the counterfactual. In addition, the Court also faults the Bundeskartellamt’s assumption that users “prefer a fee-based network to a free but ad-supported one” because the authority “made no reliable and meaningful findings” (para. 78).

Assessing the consumer harm: data processing as a voluntary consideration for free and non-indispensable services. The Court observes that the data gathered by Facebook is duplicable and can also be made available to third parties  (para. 31). It also observes that the decision failed to address why “all” of the data collected by Facebook was excessive (para. 32), and that there is “no loss of control” on the part of users because the data processing took place in compliance with Facebook’s terms of use and with users’ consent. According to the Court, the fact that use of the network is conditional on consent to the processing of the data a issue “requires to  balance the benefits of using an ad-supported (and thus free) social network with the consequences of the use of the data” (para. 35 and later also at paras. 71 and 85, and more at length at 76) and users remain free to use or not use free-ad-supported Facebook depending on their values and preferences. The Court repeatedly stresses that in Germany there are more Facebook non-users than users, which shows that non-usage is evidently an option. As we said in our comment on the opening of the case, “Facebook is not an unavoidable trading partner and consumers are not locked in to it; if consumers don’t think it’s worth giving data in exchange for the service, they won’t join. So, again, we agree.

What’s essential to a business model? Our post commenting on the decision underlined that the Bundeskartellant did at least not target the processing of data generated by Facebook’s own website because “[t]his is an essential component of a social network and its data-based business model”. Setting the business model as a red line seemed sensible. But the Court here takes a wider –and arguably even more sensible- view of what is essential to Facebook’s business model (described at the very outset in paras. 8 and 9, as well as later in 35 and 71, as offering free services financed via tailored online advertising in exchange for users agreeing to the terms of service). Later at 93 the Court points to a “lack of reliable explanation on if and to which extent  the use of the added data boosts advertising revenues to finance the social network”.

What matters (in a sanctioning regime) is the company’s behavior, not users’ psychology. The Bundeskartellamt had argued that users do not read terms of service, but the Court dismisses this argument observing  that based on a realistic interpretation , this most probably is due to “indifference or convenience of the Facebook user” and that no one had claimed there was any informational deficit on the part of Facebook (para. 37; at 71 the Court adds that “there is no evidence that Facebook obtains the consent of users through coercion, pressure, exploitation of lack of willpower or otherwise unfair means”). At 84-85 the Court explains that “whether the users act out of indifference or because they do not want to spend the necessary time and effort (…) does not matter” as their decision is ultimately “free, uninfluenced and autonomous”.

Not every legal violation is sufficient to give rise to an abuse. The Court does not agree with the Bundeskartellamt’s interpretation of the German case law. It discusses the Supreme Court’s rulings and explains that only unlawful behavior that has an effect on the protected goods of competition law (freedom of competition and openness of market access) can be equated to anticompetitive conduct. Our first post on the case presented it as part of the tendency “of extending the “special responsibility” of the dominant firm in order to comply with the law, and not just with competition law, with literally any legal provision”. Well, in a quote that deserves a proper translation, it argues that the “special responsibility” only regards competition, and does not extend to legal compliance by way of avoiding any possible violation of the law (paras. 44 and 46).

On Causality. In our comment on the opening of the case we noted that “the Press Release does actually say –or suggest- something which is arguably sensible (albeit contrary to Continental Can and Astra Zeneca) when explaining that it needs to check whether there is “a connection between such an infringement and market dominance”. Well, the Order deals at quite some length with this issue (I spare you the discussions on German case law), noting that a link of causality between dominance and the disapproved behavior (“or at least the anticompetitive effects of its behavior”) is required both under EU and German law (paras. 53-56). The Court observes that the suspected exploitative abuse does not result in a structural weakening of competition (para. 58) and that its effects on consumers are unrelated to dominance (para. 59).

Perfect understanding of “anticompetitive effects”. If you have read Pablo’s posts (e.g. here) or head me speak recently (e.g Lesson 7 here)  you will have heard that one of our recent obsessions has to do with the watering down of the notion of effects. Our contention is that according to the case law mere disparity of treatment is not enough, and that there can only be anticipative effects when rivals’ ability and incentive to compete are hindered. Here, the Court explicitly says precisely that, that“not every economic disadvantage inflicted on another company constitutes a hindrance in the antitrust sense. What is needed is an impairment of the competitive and entrepreneurial options for action and decision-making” (90). Amen.

Barriers and effects need to be convincingly shown, not simply assumed. In our comment on the decision we remarked the simplistic approach of assuming, without the necessary analysis, that “practices are problematic because they enable companies to improve their products and offer ads that are more relevant to users (…).But, unfortunately, there seems to be little appetite to deal with complexity and ambiguity these days, particularly when it comes to certain “online platforms”.

The Court in para. 93 argues that the idea that additional data increases barrier to entry because data is relevant to generate advertising revenue is “incomprehensible” and that this is a question that “requires closer examination and a detailed explanation” /”a review and conclusive presentation by the antitrust authorities. That’s what’s missing” Why? Because, the Court explains, direct network effects mean that the value of the Facebook network increases as the number of users increases and the real barrier to entry lies in rival’s need to offer an equally attractive offer capable of gaining a sufficient number of users. The Court takes the view that the decision has not “substantiated and demonstrated” how the processing of the data at issue could affect market entry. It also observes a “lack of reliable explanation on the extent to which and in what scale the use of the data boosts advertising revenues to finance the social network”. This analysis was “indispensable” because the key to entry does not lie in obtaining the highest possible advertising revenues but a sufficient number of users. At the end of 93 the Court explains what type of analysis was required. The same is true about allegations of leveraging in other (not properly defined) markets, where the decision shows a “serious lack of reasoning” (para. 94) “lacking robust and comprehensible explanations” (para. 95).

A competition law problem? The conclusion to our first post on this case was that “there may be a market failure, but one that has to do with asymmetries of information, not market power. In other words, whether consumers know or not what terms and conditions they are accepting may be a public policy issue, but one that, in my humble view, is not for competition law to address”. The Court appears to share this belief. It explains that only with the help of the causality requirement “it is possible to avoid antitrust enforcement beyond the regulatory purpose of abuse control and to prevent the antitrust authority from prosecuting non-competition related infringements”(…) “unfairly disadvantageous terms (…) can also be based on informational market failure and the resulting systematic asymmetry of information to the detriment of customers) (para.61). The Court understands that “this possible alternative causation link justifies both the unlawfulness of the decision” (because the Bundeskartellamt bore “the burden of determination”/proof) (para. 76) and the legitimacy of consumer protection rules. Like. Earlier on the Court had made a point in this regard that we have also made before: the interests of those affected by the same behavior on the part of non-dominant firms are no less worthy of protection (para. 47), which is another reason not to leave these matters to competition law.

***

Stay tuned for Part II, with our comments on the reactions from other commentators and on what this development may/should mean for competition enforcement.