The potential anticompetitive impact of the automation of decision-making has become a topic of great interest in recent times. Self-learning pricing algorithms are focusing the agenda from the European competition authorities due to two potential harmful effects on business conduct: price discrimination and collusive practices. It is precisely the second effect that is generating greater uncertainty and debate. One of the key questions is whether it is possible to design a fully-autonomous software capable of colluding without any explicit human intervention and contact between competitors.
Algorithms are instances of logic (or rules) written in software to produce a certain output. Software incorporating business decision-making algorithms is capable of accumulating an enormous amount of data that can be used for tracking and adapting to competitors prices practically in real time. The increasing sophistication of these mechanisms allows to monitor agreed prices, to detect deviations from agreements or to signal the intention to engage in concerted practices much more effectively. Consequently, tacit collusion becomes easier to create and less expensive to maintain.
The European Commission has recently published its Final Report on the E-commerce Sector Inquiry, where it states that the “availability of real-time pricing information may trigger automatized price coordination”. The Commission’s concerns are further emphasized in its contribution to the Background paper of the Organization for Economic Cooperation and Development (OECD) Roundtable on “Algorithms and collusion” of June 2017, where it suggests that they may be examined under traditional EU competition law.
In this scenario, European competition authorities face the challenge of determining what behaviors should fall under the scope of Article 101 of the Treaty on the Functioning of the European Union (TFEU). In fact, a grey area exists between the right of companies to intelligently adapt to the behavior of competitors and anticompetitive tacit collusion.
Bearing in mind that there is less and less human intervention in this area, another major challenge is the attribution of responsibility for such acts, which requires demonstrating that there has been an intention to coordinate prices when designing the algorithms. Commissioner Margrethe Vestager, in a speech at the 18th Conference on Competition (16th March 2017), pointed out that “companies can’t escape responsibility for collusion by hiding behind a computer program” and that “pricing algorithms need to be built in a way that doesn’t allow them to collude”. Therefore, it seems that responsibility for anticompetitive conduct derived from algorithms may be attributed to the firm.
Some authors believe it is plausible that even in the absence of explicit instructions, and thanks to its self-learning ability, software that is programmed to maximize benefits issues collusive responses without having been programmed for this purpose. The European Commission’s Chief Economist has stated that he is not yet convinced that computers can tacitly collude.
Although this is a fairly recent issue, some competition authorities have already had the opportunity to rule on the matter.
In U.S. v. David Topkins (2015), a former e-commerce executive was found guilty of conspiring to fix the prices of posters sold on Amazon Marketplace. In this case, the U.S. Department of Justice found evidence that the defendant had participated in the design and implementation of a pricing algorithm to monitor the agreement.
In August 2016, the Competition and Markets Authority (“CMA”) of the United Kingdom issued a decision finding that Trod Limited and GB eye Limited infringed competition law by agreeing that they would not undercut each other’s prices for posters and frames sold on Amazon’s UK website. The agreement was implemented using automated repricing software configured by each of the parties. Amazon itself was not involved in the cartel and has not been investigated by the CMA.
The European Commission has not yet dealt with a “pure” pricing algorithm case. However, the ruling of the Court of Justice of the European Union (“CJEU”) in preliminary ruling Case C-74/14 Eturas may give some guidance on how to assess cases of tacit collusion in an online environment. In June 2012, the Lithuanian Competition Council fined 30 travel agencies and Eturas (as a facilitator) for participating in a concerted practice in relation to discounts applicable to bookings via the E-TURAS online travel booking system. The Competition Council found that Eturas, the administrator of E-TURAS system, had set a technical restriction that limited to 3% the discounts available for online bookings. The administrator had previously sent a notice to all travel agencies using the system informing about this restriction.
The National Court dealing with the appeal filed to the CJEU questions for a preliminary ruling, asking whether the communication and introduction of technical limitations on discounts applicable to the sale of products through a computerized booking system constituted a concerted practice for the purposes of Article 101 of the of the Treaty on the Functioning of the European Union (“TFEU”). The CJEU ruled that if the travel agents using the platform had knowledge of the content of the message of the administrator which potentially gave rise to anticompetitive collusion, they may be presumed to have participated in that agreement unless they took steps to distance themselves. However, travel agents must have the opportunity to rebut the presumption that they were aware of the content of the message by proving, for instance, that they did not receive the message or that they did not read it until some time had passed since its dispatch. Hence, the CJEU confirmed that actual knowledge was required for an infringement to exist and that the sending of the message alone was not sufficient to give rise to a presumption of knowledge, but knowledge could be inferred from “objective and consistent” indicia.
The debate on the implications of algorithms from a competition law point of view will remain one of the trending topics in the upcoming years. The Commission has already warned companies that pricing software will be one of its enforcement priorities.