This paper considers a sequential social learning game with a general utility function, state and action space. We establish that the value of private information converges to zero almost surely in every Perfect Bayesian equilibrium of any sequential social learning game.We use this result to show that totally unbounded signals are necessary and sufficient for asymptotic learning to hold in every sequential social learning game. Finally, we assume that the utility function of each agent is a private random draw and establish robustness of our results. (Joint with M. Mueller-Frank).

About Abstract