Behavioral Heuristics and Market Patterns in a Bertrand-Edgeworth Game

This paper studies price-setting behavior in a Bertrand-Edgeworth game, a traditional setup that provides a stylized representation of real-world markets (such retail markets). It explores methodological complementarities between analytical, experimental and agent-based simulation approaches. The game under analysis only admits hard-to-characterize mixed-strategy Nash equilibria. This paper uses families of heuristic rules for individual behavior to analyze the resulting market prices. A set of laboratory experiments finds that price-setting choices of agents are described reasonably well through a sales-based simple pricing rule. Average market prices tend to converge from above to quasi-steady states with small individual dispersion, occasionally disturbed by agents who search for profits by raising prices. Salient features of the experimental outcomes can be represented through simulations with artificial agents who apply the sales-based heuristics with parameters calibrated to the experiment.