Website CV Paper
Fields
Theory, Industrial Organization
Job market paper
Misspecification-Aware Sequential Social Learning
I study a model of sequential social learning à la Smith and Sørensen (2000), in which players are concerned that their model of the signal distribution may be misspecified and choose actions according to a criterion that is robust to misspecification. I show that this concern creates a trade-off in information transmission: greater concern makes extreme actions more informative but less frequent. When misspecification is sufficiently small, introducing a small concern for misspecification accelerates learning and improves the discounted sum of expected payoffs. This welfare gain reflects a second-best logic: although concern lowers a player’s expected payoff at a given belief, this loss is outweighed by the gains to future players from more accurate beliefs.