When Competition Policy Meets Science Fiction (Part 2)

(Part 1 here)

Further discussions in London focused on a second constellation, characterized as “Hub and Spoke”, in which competitors do not predetermine the pricing rules themselves but outsource this function to a third party. The main differences between this and the “Messenger Scenario” (as described above with reference to the UK poster case) are that the parties here a) use the same automated repricing tool and b) the computer programme calculates prices based on its own blueprint and not directly executing the rules set by the human sellers. As noted by some of the conference speakers and thoroughly discussed in the book Virtual Competition, the use by competing sellers of automated repricing capabilities offered by a single provider can lead to some measure of de facto price alignment. This result is potentially worrisome and competition enforcers should be prepared to address it in a suitable way. This shouldn’t be too difficult where the dampening of price competition produced by the automated repricing tool is intentional on the side of the competitors, meaning that it is the original reason why they have chosen the same algorithmic tool, or that they are at least aware that the technology as concretely employed has or could have this effect. In this case, the automated repricing software jointly employed by the sellers could be seen as the hub (or "brain") that facilitates collusion by controlling the wheel's spokes.

Also amidst the gales of technological change, therefore, the notion of awareness is likely to remain central to the assessment of collusive behaviour, as recently stressed by the CJEU in Eturas, where the Court applied Art. 101(1) TFEU to an online travel booking system used by 30 travel agencies in Lithuania. The administrator of the booking system posted a notice in the system mailboxes informing the agencies of a technical (and automatic) restriction on the discount rates they could offer their own clients. The first preliminary question addressed to the CJEU by the referral national court asked whether the simple proof of the system notice allows the presumption that the “economic operators were aware, or ought to have been aware, of the system notice introduced into the computerised information system”. According to the CJEU, it is up to national law to decide if proof that a message has been sent to the booking system’s mailboxes is sufficient to prove that the addressees were aware, or ought to have been aware, of its content. The presumption of innocence however precludes the referring court from deducing the undertaking's awareness of the message content from the mere dispatch of the message in the booking system. Instead, a presumption of awareness may be based on ‘other objective and consistent indicia’ that the undertaking tacitly assented to an anticompetitive action (for instance, in this case there had been prior communication between the system administrator and the travel agencies regarding a possible capping of discount rates). If awareness of the content of the message can be demonstrated, the acquiescence in that initiative may be inferred unless the undertaking opposes to it (e.g., by reporting the initiative to the authorities). In a nutshell, the “unusual method of communication” between the undertakings concerned, namely the system notice, is a sufficient basis for the finding of a concerted practice aiming to a discount restriction provided that the travel agencies were aware of the content of the communication. This also means that, as argued by the Advocate General Szpunar, “the mode of communication in itself is not relevant, especially since the participants in collusion may be expected to avail themselves of the possibilities offered by the advance of technology”.

On a slightly different note, it should also to be highlighted that future cases are likely to be substantially more challenging than the one considered by the CJEU in Eturas, which had some rather rudimentary technology at its core. Thus, the provider of a jointly employed, big data-fueled repricing tool like for instance Feedvisor could work out profit maximization strategies for the benefit of its high-paying clients that are much more sophisticated (and opaque) than a bare price alignment, based on rich and complex sources of market data, the ranking criteria (algorithms) employed by the marketplaces, the flow of information coming from the sellers, in-depth consumer data, etc. Among the many tactics creatively employed by the automated repricing tool, only a few – difficult to spot, and for intermittent periods of time – could possibly be considered price dampening by way of horizontal collusion.

Also discussed as at least tangentially part of the “Hub and Spoke Scenario” was the so called Uber Dilemma. By joining the car service platform, the driver agrees to charge her riding services according to the fares worked out by Uber’s algorithm. This is a simple, middling vertical agreement between the platform and the driver. Once the platform acquires market power, other drivers could become aware that by joining the platform they would feast on supracompetitive prices (higher fares and, subsequently, higher commissions earned by the platform). At this point, that is likely to materialize after the platform has already tipped into dominance, competition enforcers could detect the familiar scent of horizontal collusion in the market, possibly by way of hub-and-spoke conspiracy. But as soberly intimated by one distinguished competition enforcer and keynote speaker at the conference, “intervening after tipping may be futile”.

(Part 3 follows after a well-deserved vacation :-)).

No comments:

Post a Comment