2.0 "What About Bob Test": Pricing algorithms and a proposal for new additional parameters that can inform antitrust investigations around the world

Abstract

This paper examines the role of pricing algorithms in antitrust investigations, focusing on two main questions: Which categories of pricing algorithms are most frequently targeted in recent antitrust cases? Which parameters can competition authorities use to distinguish legitimate, efficiency-enhancing algorithms from practices that threaten competition? The article is situated within the scope of coordinated conduct facilitated by pricing algorithms. The paper is structured as follows: Section 1 introduces the topic and its main concepts and distinctions. Section 2 presents the "What About Bob" analytical test and its three parameters: (1) Are the pricing algorithm's recommendations mandatory or optional? (output) (2) The pricing algorithm relies on public or confidential data? (input) (3) the algorithm can manipulate market conditions through adaptive learning? Section 3 conducts a case law analysis of ten antitrust investigations from the US, Brazil, Canada, and Poland, classifying them by algorithm category and three parameters test. The analysis finds that dynamic pricing algorithms and hub-and-spoke coordination are most common in recent investigations, and that the distinction between mandatory and optional recommendations, as well as the use of public versus confidential data, are central to antitrust assessment. Also, 70% of the antitrust cases are still ongoing. Section 4 summarizes the findings; and the conclusion proposes five additional parameters to complement the original test, leading to a 2.0 "What About Bob Test" with eight parameters. The five new parameters are the following: (4) Is there evidence of prior explicit collusion? (5) Are the pricing algorithms recommended by a trade association or union of the market? (6) Are the companies aware of a collective adoption by competitor of the pricing algorithm (7) Is the pricing algorithm used by companies in the market that jointly have dominant position? (8) Does the algorithm deliver significant market efficiencies given the market's characteristics? Considering that pricing algorithms remain a relatively recent phenomenon and most investigations are ongoing, we propose that-except in cases of explicit collusive agreements facilitated by the algorithm-all other categories should be assessed under the rule of reason as the guiding principle of antitrust analysis. This approach would help ensure that the use of algorithms is not unduly discouraged, given their potential to generate significant benefits for market development worldwide

Keywords: pricing algorithms, artificial intelligence, antitrust, competition