Overcoming Free-Riding in Bandit Games

Abstract This article considers a class of experimentation games with Lévy bandits encompassing those Bolton and Harris (1999, Econometrica, 67, 349–374) Keller, Rady, Cripps (2005, 73, 39–68). Its main result is that efficient (perfect Bayesian) equilibria exist whenever players’ payoffs have diffusion component. Hence, the trade-offs emphasized in literature do not rely on intrinsic nature bandit models but commonly adopted solution concept (Markov perfect equilibrium). an artefact continuous time: we prove arise as limits discrete-time game. Furthermore, it suffices to relax strongly symmetric equilibrium.