Adaptation and convergence of behavior in repeated experimental Cournot games

This research examines results from laboratory experiments in which five human subjects participate as sellers in a Cournot oligopoly environment. The central issue is whether repeated play among a group of privately informed subjects will lead to convergence to a unique, static, noncooperative Nash equilibrium. The experiments were designed so that the implications of different hypotheses about adaptation and convergence, such as the best response dynamic and fictitious play, could be distinguished. The results provide, at best, only partial support for the hypothesis that behavior of privately informed subjects will converge to the static Nash equilibrium when play is repeated. Total output averaged over time periods and across experiments is greater than, but still close to, predicted equilibrium total output. However, observed intertemporal variation in total output and heterogeneity in individual choices are inconsistent with convergence to the static Nash equilibrium.