Mon, 11/15/2021 - 13:00 / 14:00
405, Viale Romania
Speaker: Emanuele Tarantino , Luiss
We propose a theoretical analysis of the conditions under which estimation of search costs is biased when a Bayesian agent searches in the presence of social learning. We extend the canonical empirical sequential search model by allowing agents to observe the choice of one of their social connections. We find that social learning changes agents’ sequential search decisions. We compute the estimator of search cost distributions under various standard datasets available to empirical researchers. We find that neglecting social information biases the estimate of search cost distributions. The direction of the bias depends on the dataset’s content. We also suggest offline estimation techniques or experiments that are useful to identify and correct the direction of the bias.