Delegate Pricing Decisions to an Algorithm? Experimental Evidence

2025-10-31

AI Summary

This study examines how delegating pricing authority to self-learning algorithms with collusive tendencies affects market prices and competition in repeated Bertrand games. Using experimental economics, we implement three institutional treatments: (1) human-only pricing, (2) full algorithmic delegation, and (3) human-vetoable algorithmic recommendations (“vetoable recommendations”). Results show: (1) delegation rates are significantly higher under vetoable recommendations; (2) both algorithmic treatments yield systematically lower equilibrium prices—and thus higher competitive intensity—than human-only pricing; and (3) human–algorithm collaboration does not facilitate collusion but instead suppresses prices, challenging the prevailing “algorithms inevitably foster collusion” hypothesis. This is the first controlled experimental study to demonstrate the dual regulatory role of human veto rights—shaping both delegation behavior and market performance—thereby providing novel empirical evidence for algorithm governance and antitrust policy design.

Technology Category

• Game Theory and Economic Paradigms: Mechanism Design
• Multiagent Systems: Mechanism Design
• Humans and AI: Learning Human Values and Preferences

Application Category

• Economics, Online Markets and Human Computation: The sharing economy
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

We analyze the delegation of pricing by participants, representing firms, to a collusive, self-learning algorithm in a repeated Bertrand experiment. In the baseline treatment, participants set prices themselves. In the other treatments, participants can either delegate pricing to the algorithm at the beginning of each supergame or receive algorithmic recommendations that they can override. Participants delegate more when they can override the algorithm's decisions. In both algorithmic treatments, prices are lower than in the baseline. Our results indicate that while self-learning pricing algorithms can be collusive, they can foster competition rather than collusion with humans-in-the-loop.

Problem

• Examining delegation of pricing decisions to collusive self-learning algorithms
• Comparing human-set prices versus algorithm-influenced pricing outcomes
• Investigating whether algorithms foster competition or collusion with human oversight

Innovation

• Self-learning algorithm sets prices automatically
• Human override capability for algorithmic decisions
• Algorithmic pricing fosters competition over collusion