Tackling Algorithmic Collusion Without Reinventing the Wheel

Abstract

Self-learning pricing algorithms increasingly allow firms to sustain supracompetitive outcomes without explicit communication, exposing weaknesses in competition regimes that rely on finding “agreements” to establish conspiracy. This paper argues that abuse of dominance provisions, rather than traditional cartel or conspiracy rules, offer the most promising path for addressing algorithmically facilitated coordination across U.S., EU, and Canadian law.

The recent RealPage prosecution in the U.S. illustrates these challenges. Rather than solely prosecuting RealPage for collusion, the Department of Justice has alleged that RealPage’s revenue-management system uses large pools of sensitive competitor data and machine-learning tools to shape common pricing behaviour across rental markets. The case shows how a single algorithmic intermediary can influence market-wide outcomes when competitors neither communicate nor intend to coordinate, underscoring the difficulty of fitting such conduct into agreement-based notions of cartels.

The analysis concludes that abuse of dominance-based tools provide a more coherent doctrinal framework for algorithmically facilitated coordination. However, the RealPage dispute also demonstrates that current legal standards were not designed for AI-mediated market structures. Ensuring effective oversight will require modernized regulatory frameworks capable of responding to coordination that arises from shared algorithms rather than traditional collusive intent.