Estimating search with learning

In this paper we estimate a dynamic model where consumers are searching for the best good and learning about the price-quality relationship in the Bayesian fashion. Thereby we relax one of the critical assumptions made in the existing studies on empirics of search: that consumers know the distribution o¤ers and therefore do not take into account the information collected during the search process. Besides being more realistic, the assumption of learning allows to make an inference about consumer’s prior beliefs from his search decisions: the variation of posterior beliefs among consumers can explain these decisions in a way that is complementary to variation in search costs. We estimate models with and without learning on a unique dataset of search histories by consumers booking a hotel online. A statistical test between the two models shows that the data favors the learning hypothesis. We also …nd an evidence that consumers underestimate the price of quality, relative to the price regression.