Abstract:There has been substantial recent concern that pricing algorithms might learn to ``collude.'' Supra-competitive prices can emerge as a Nash equilibrium of repeated pricing games, in which sellers play strategies which threaten to punish their competitors who refuse to support high prices, and these strategies can be automatically learned. In fact, a standard economic intuition is that supra-competitive prices emerge from either the use of threats, or a failure of one party to optimize their payoff. Is this intuition correct? Would preventing threats in algorithmic decision-making prevent supra-competitive prices when sellers are optimizing for their own revenue? No. We show that supra-competitive prices can emerge even when both players are using algorithms which do not encode threats, and which optimize for their own revenue. We study sequential pricing games in which a first mover deploys an algorithm and then a second mover optimizes within the resulting environment. We show that if the first mover deploys any algorithm with a no-regret guarantee, and then the second mover even approximately optimizes within this now static environment, monopoly-like prices arise. The result holds for any no-regret learning algorithm deployed by the first mover and for any pricing policy of the second mover that obtains them profit at least as high as a random pricing would -- and hence the result applies even when the second mover is optimizing only within a space of non-responsive pricing distributions which are incapable of encoding threats. In fact, there exists a set of strategies, neither of which explicitly encode threats that form a Nash equilibrium of the simultaneous pricing game in algorithm space, and lead to near monopoly prices. This suggests that the definition of ``algorithmic collusion'' may need to be expanded, to include strategies without explicitly encoded threats.

The impact of collusion on price behavior: Empirical results from two recent cases

Asymmetric Pricing Caused by Collusion

What Do We Know About Algorithmic Collusion Now? New Insights from the Latest Academic Research

Unprofitable Cartels: Evidence from a Natural Experiment in the UK

Collusion Between Supply Chains under Asymmetric Information

Algorithmic Collusion by Large Language Models

Collusion risk in corporate networks

Collusion by Algorithm: Does Better Demand Prediction Facilitate Coordination Between Sellers?

Algorithmic collusion: A critical review

Algorithmic Collusion and Price Discrimination: The Over-Usage of Data

Learning Collusion in Episodic, Inventory-Constrained Markets

Deciphering Algorithmic Collusion: Insights from Bandit Algorithms and Implications for Antitrust Enforcement

Algorithmic Collusion: Supra-Competitive Prices via Independent Algorithms

Is Collusion-Proof Procurement Expensive?

Algorithmic Collusion in Assortment Games

Collusive Price Leadership Among Firms with Different Discount Factors

Algorithmic Collusion Without Threats

Hypothesis Testing Approach to Detecting Collusion in Competitive Environments

Regulation of Algorithmic Collusion

Algorithmic Collusion: Genuine or Spurious?

AI Pricing: Adoption of Artificial Intelligences and Collusive Price