Communication in a Complicated World

Steve Callander (Stanford University)
Nicolas Lambert (Stanford University)

Abstract: The internal organization of political and economic institutions is often predicated on enhancing the flow and use of expert information. However, underlying existing models is a canonical representation of expertise that takes a very simple yet unrealistic form. Beginning with the famous model of Crawford and Sobel (1982), an expert is modeled as knowing a single one-dimensional piece of information that the non-expert doesn't. In this paper we construct a richer model of expertise, one in which the advantage of expertise is much greater. This representation applies the Brownian motion to model the mapping from actions to outcomes. In this setting an expert knows a continuum of correlated pieces of information that the non-expert doesn't. We characterize equilibrium policy advice in this setting and show that it takes varied and richer forms that resonate with practice and that avoid several undesirable features that flow from the traditional approach. We explore what this implies for how expertise relationships should be structured for optimal policy formation.