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.

Collusion, inattentive consumers and shrouded prices

Collusion Between Supply Chains under Asymmetric Information

Collusive Price Leadership Among Firms with Different Discount Factors

Asymmetric Pricing Caused by Collusion

Algorithmic Collusion in Dynamic Pricing with Deep Reinforcement Learning

Collusive Outcomes Without Collusion

Algorithmic Collusion Without Threats

Learning Collusion in Episodic, Inventory-Constrained Markets

Algorithmic Collusion: Supra-Competitive Prices via Independent Algorithms

Competing to Commit: Markets with Rational Inattention

By Fair Means or Foul: Quantifying Collusion in a Market Simulation with Deep Reinforcement Learning

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

Data‐driven collusion and competition in a pricing duopoly with multinomial logit demand

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

Robust Algorithmic Collusion

Algorithmic Collusion in Assortment Games

Bertrand Vs. Cournot Competition in Distribution Channels with Upstream Collusion

On Mechanism Underlying Algorithmic Collusion

Frontiers: Algorithmic Collusion: Supra-competitive Prices via Independent Algorithms

Dynamic pricing with multiple consumers and alternating offers under retailer competition: theory and experiment

Algorithmic Collusion: Insights from Deep Learning