Chillin'Competition

Yesterday the ECJ rendered its awaited Judgment in the Eturas case on the boundaries of the notion of agreement and, specifically, on concerted practices. (An English version was not yet available last night, so what follows uses my own relaxed translation from the French version).

The case originates in a preliminary reference from the Lithuanian Supreme Administrative Court in the context of annulment proceedings against a national decision finding that travel agents had coordinated the discounts applicable to clients.

What makes the case particularly interesting, is that the alleged coordination would have taken place via an online travel booking system (E-TURAS, owned by Eturas) used by more than 30 travel agents in Lithuania. The question that attracted interest to the case was, in essence, that of whether a common electronic platform facilitates collusion, an issue that has elicited interesting comments in recent times  (see here for an interesting paper by Ezrachi and Stucke, and here for a New Yorker piece (When bots collude) on a US case involving coordination through an algorithm). Online poker players discovered these problems a long time ago, so my brothers say…

The facts

According to the Lithuanian authority, prior to the alleged infringement the director of the Eturas had sent an email to other travel agents asking on a vote on whether discounts should be reduced (there is only record of one agent having received it). Following those emails

• a system notice was sent via the internal E-TURAS messaging system announcing that on the basis of the declarations, suggestions and wishes of agents, discounts were in principle capped at 3% (the message was available and could only be consulted in a section of the system called “information messages”; there is only evidence that two agents accessed it; no one replied and no one took public distance from the message either) and, subsequently
• a technical restriction was set in the E-TURAS system (integrated in the websites of the agents) limiting to a maximum 3% the discounts available for online bookings (the technical restriction did not preclude larger individual discounts, but those required additional technical actions)

The national competition authority found an infringement as it observed that agents had not publicly distanced themselves from the initiative, could have reasonably assumed that others had received the same message and were likely to abide by it, and it inferred that agents had previously discussed these actions. An annulment action was partially upheld and fines were reduced, but both the authority and the companies appealed before the Administrative Supreme Court, who referred two questions to the ECJ:

The Judgment

—First question: is the fact that a message was sent enough to establish that its addressees became or should have become aware of its content, and that by not opposing it they acquiesced in a way that would make them liable under the competition rules?

The ECJ starts off recalling the principles that companies are to determine their behavior autonomously (27) and that “passive modes of participation” (i.e. complicity/facilitation) are also caught by Art. 101 (and refers here to the recent Judgment in Treuhand (28)) (our comment on that one is available here).

The Court nevertheless explains that, whereas there is a presumption that companies take into account the info exchanged with others for determining their behavior (33, citing the infamous T-Mobile), the question posed by the national Court is not really about that, but rather about an evidential issue (34). As recalled in paras. 29 to 32, issues pertaining to the standard of proof are not harmonized by Regulation 1/2003 and therefore remain a matter of the national law at issue provided that the principles of effectiveness and uniformity are respected.

The principle of effectiveness requires that national rules on evidence don’t render the application of EU law impossible or excessively difficult; accordingly, it requires that not only direct evidence, but also indicia, be valid to prove an infringement (35-37). However, if, as in the case at issue, the national Court has doubts as to whether the travel agents became or must have become aware of the content of the system notification, then the presumption of innocence applies, in the sense that the mere sending of the notification cannot be in itself enough to infer awareness of its content (38-39). It could nonetheless be used in combination with other objective and consistent indicia to establish a rebuttable presumption of awareness (40).

According to the Judgment, the national Court should not make it excessively difficult for the companies to rebut this presumption; for instance, they should be able to show that they had not received the message, consulted that part or read it until some time had elapsed.

—Second question: in the event of a negative reply to the first questions, what elements should be taken into account in determining whether undertakings participating in a common system like the one at issue take part in a concerted practice?

The ECJ recalls that a “concerted practice” implies not only an element of concertation, but also an ensuing behavior on the market and a causal link between the two (42)

After repeating the facts of the case (43), the ECJ holds that (a) those agents that were aware of the content of the system notification could be presumed to have tacitly acquiesced provided that the other elements (behavior + causal link) are met and thus be liable as from the moment of reception (44); and (b) those agents whose awareness cannot be established cannot be presumed to have participated in a concerted practice by virtue only of the implementation of the technical restriction (45).

A presumption of the participation of a given undertaking in a concerted practice can nonetheless be rebutted by publicly distancing itself from that practice or by reporting it to administrative authorities. In addition, the Court says that, according to the case law, in the absence of a collusive meeting there may be other ways of rebutting the presumption (46).

With regard to “public distancing”, the Court observes that in a case like this, in which it is impossible to identify all addresses of the message, it cannot be required that a company distances itself in the eyes of all other addressees (47), and takes the view that a “clear and explicit objection” addressed to the system administrator would be enough to rebut the presumption (48). AG Szpunar had suggested something similar in his Opinion, adding also that ignoring the communication or instructing employees not to abide by it would not have been sufficient.

With regards to the other means of rebutting the presumption of having participated in a concerted practice, the Court states that, in circumstances such as the ones at issue, the causal link could be rebutted by showing that the company systematically applied a discount exceeding the cap (49).

My comments:

On the first question -whether prove that a message was sent is enough to presume the awareness of the recipient- is rightly treated by the Court as a matter of evidence subject to national law; the Court’s reasoning in the regard is sound and consistent.

Usual readers of the blog have my issues with the principle of procedural autonomy (it is an undeniable problem that the outcome of cases dealt with the same provisions – Art. 101 and 102- can vary so much depending solely on the authority that handles it because in real life cases procedure matters, in practice often more than substance.

We all know that the principle of procedural autonomy is often eroded by the principles of effectiveness and uniformity (there are several examples in competition law; see here). When those principles are applied, however, the result is typically to enlarge the scope of rules, powers of prohibitions (in practice effectiveness tends to weigh more than uniformity…). But in Eturas the Court says something interesting, even if obvious: above all, above other presumptions, and also above effectiveness, is the presumption of innocence, and in the face of doubt, the solution is clear. So, there are two outer limits to procedura autonomy in this regard: proving something cannot be too easy (presumption of innocence) nor too difficult (effectiveness).

In its response to the second question, the ECJ’s approach consists in applying the same principles (presumption that info will be taken into account + requirement of public deviation) but with some flexibility to account for the peculiarities of the technology at issue (i.e. admitting other means of distancing, not requiring that distancing be done with regard to all others, and not suggesting, like the AG’s Opinion did, that it could be necessary to also inform customers).

The Court also admits that distancing can be shown by conduct on the market, but provided that it is “systematic” (AG Spuznar’s Opinion had proposed not to accept market conduct as a valid element of rebuttal as it could hardly be distinguished from “cheating”). This is sensible, and in fact there are several precedents in which EU Courts accepted distancing that did not involve express declarations (e.g.  Cases T-208/08 and T-209/08, Gosselin Group v Commission,  para 161 or Cases T-122/07 to T-124/07, Siemens AG Österreich, paras 60-61).

Furthermore, the ECJ introduces a twist –not so visible in AG Szpunar’s Opinion- which is to make the establishment the concerted practice conditional on the finding that a given company was aware of the content of the message. Note in this regard that the second question posed by the national Court was consciously and expressly framed in a way that assumed that the companies were not aware of that content. Nevertheless, the ECJ rejects the possibility that a concerted practice is proved in the absence of evidence on awareness of the message (45). This is sensible and in line with previous case law.

The implications of the Judgment are interesting:

Companies sharing important IT functions with competitors will now have to carefully craft protocols  to avoid possible liabilities. I have my ideas on how to do that, but I (or rather my firm) bills for those 😉

Questions remain open, particularly concerning system administrators. For instance, in para. 48 the ECJ accepts that distancing may be done only vis á vis the “system administrator”. But in that case, what obligations fall upon the system administrator? Would the immediate forwarding of that message or the withdrawal of the initial communication exempt it from liability? If yes, would that  run counter the effectiveness of the prohibition given that a signal was already sent? If not, what incentives would it have to correct its behavior?

Also, would a “suggestion” or “recommendation” on pricing –as opposed to a technical restriction- be enough of a signal to give rise to a concerted practice?  Is the legality of the system administrator’s conduct conditional on the actions adopted by the users of the system? (after all, in the absence of a concerted practice, its conduct would be unilateral and hardly objectionable, unless perhaps framed as an individual abuse of collective dominance –in what would be a mixture of the EU’s Irish Sugar and the US’s Ethyl case on facilitating practices-). And, in in this sense, could the use of a common IT system be considered as a structural link relevant to establishing collective dominance? I guess not, but take a look at Italian Flat Glass…

My bet is that it won’t take long before these questions come up again.