EU ANTITRUST: HOT TOPICS & NEXT STEPS

EU ANTITRUST: HOT TOPICS & NEXT STEPS 2022

Prague, Czechia

of a merger, shifting the burden of proof of credible negative evidence to the parties could effectively block mergers. This proposal might really be a kind of “firing squad”. Similarly remarkable is the suggestion that the exclusionary conduct is anticompetitive regardless of market power. The presumption could be rebutted by its addressee, which is a firm or group of firms that have over 50% market share or otherwise have (only) “significant market power” (not inevitably dominant position – ibid, p. 10). We need to monitor the discussion of these proposals closely, as they may be important and maybe fatal for the development of AT worldwide. 3.2 Pricing Algorithms These are one of themost visiblemanifestations of the digitalisation of the economy and also one of the most dangerous, because they blur the distinction between the forbidden conscious collusion (meeting of minds) and the silently suffered tacit collusion (parallelism), which is not and cannot be prohibited because it would undermine the essence of market interaction. Algorithms undermine the category of intent or mutual understanding underlying the classic cartel doctrine. They can contribute to open collusion because they act as information intermediaries between conspirators and speed up the exchange of relevant information necessary to maintain a cartel for mere moments. Algorithms can also facilitate collusion made tacitly, due to the negligible cost of information and of responding to it immediately. Even if tacit (unintentional) collusion is not punishable by law, this does not mean that it is desirable (Ezrachi, Stucke, p. 66). Algorithms in general can also be pro-competitive (OECD, p. 14). On the supply side, they increase transparency, which leads to product improvement and new product development, so they increase competitive pressure. On the demand side, they support consumer demand and provide information on non price parameters (quality, consumer preferences). Given that it is practically impossible to distinguish the outcome of a spontaneous price following (parallel behaviour) from conscious understanding, regulation in the form of a rebuttable presumption of knowledge of such consequences of pricing algorithms (Salaschek, Serafimova, p. 14) that would force competitors to insert coordination-preventing mechanisms into algorithms has also been considered. The burden of proof of intent or purpose to collude otherwise will be on the competition authority. There is also a view that certain types of pricing algorithms that support supracompetitive prices (such as self-learning price-setting algorithms) should be banned automatically ( per se ), and that perhaps the time is ripe to ban any conduct with a potentially anticompetitive tendency that does not have a corresponding pro-competitive offset, even if such conduct does not meet the characteristics of an agreement in the traditional sense (Gal, p. 116 ff.).

354