Estimating demand in online search markets, with application to hotel bookings,” unpublished manuscript

When consumers search before making a purchase, their choice sets have two properties: they are limited and endogenous to consumer preferences. Both factors introduce bias in the estimates of discrete choice models of demand. We show that conditional search decisions, together with purchases, provide enough information to identify heterogeneous consumer pref-erences together with search costs. An attractive feature of this approach is that it does not require data on search cost shifters, which is rarely available in online markets. We estimate a structural model of search, using a unique dataset of searches for hotels online. The results show that both the limited nature and the endogeneity of choice sets due to search can lead to signi cant over-estimation of price elasticity. We also provide estimates of the joint distribu-tion of search costs and the price coe ¢ cient, indicating a correlation between these consumer level parameters. This suggests that the search behavior and the price sensitivity may have common determinants. PhD student at Columbia University. I am indebted to my advisors, Michael Riordan, Bernard Salanié and Katherine Ho. I am grateful to participants of Applied Micro Theory and Industrial Organization seminars at Columbia, the search workshop at Gronningen, to Jose Moraga-Gonzalez and Paulo Albuquerque for their helpfu