Neighborhood-Based Information Costs

Neighborhood-Based Information Costs

By Benjamin Hébert, Michael Woodford
January 31,2020Working Paper No. 3751

We derive a new cost of information in rational inattention problems, the neighborhood-based cost functions, starting from the observation that many applications involve exogenous states with a topological structure. These cost functions summarize the results of a sequential information sampling problem (because they are uniformly posterior-separable) and capture notions of perceptual distance. This second property ensures that neighborhood-based costs, unlike mutual information, make accurate predictions about behavior in perceptual experiments. We compare the implications of our neighborhood- based cost functions with those of the mutual information in a series of applications: security design, global games, modeling perceptual judgments, and linear-quadratic-Gaussian settings.