Albert Ma, Boston University

In a general credence-good model, a consumer’s potential loss is a continuous random variable. Observing the loss value, an expert can provide a repair. If the expert’s price offer is accepted, the loss will be avoided. Not knowing the true loss, the consumer never learns if the repair would have been worthwhile. We characterize perfect-Bayesian equilibria, all of which are inefficient. In closed form, we derive separating and pooling equilibria; in the former, the consumer can infer losses from prices, in the latter, the consumer can infer that losses reside in an interval. We also endogenize the expert’s information acquisition. The first best is achieved if the expert can commit to only knowing if the loss is below or above a certain threshold. The general framework can be extended to a two-dimensional model in which cost and loss are random and correlated, and to consider a market with multiple experts.

DATE: Friday, March 4, 2022

TIME: 3:30-5:00 p.m.

LOCATION: Fronczak 444