Collusion by pricing algorithms in competition law and economics

Abstract

Software programs based on algorithms have become common in pricing because they outperform humans at automatising tasks in terms of speed, complexity, and accuracy of analysis. In many online markets, repricing algorithms have replaced the human decision-maker. As with any other technology employed in the market, repricing algorithms empower human activity toward both positive and negative consequences. Their properties enable market transparency and efficiencies but also entail collusion risks beyond traditional oligopolies. This paper analyses why repricing algorithms can facilitate anti-competitive coordination and what is the scope for Art. 101(1) TFEU to tackle it. Acknowledging the limitations of EU competition law against collusion by autonomous algorithms, we qualify the antitrust concern through the economics and computer science understanding of pricing algorithms. Algorithmic pricing does not always lead to higher prices, although even simple algorithms can learn complex reward-punishment schemes that resemble collusive pricing strategies.