Econometrics & Applied Economics Seminar - Andy Wu (The University of Melbourne)

Title: Estimating Consumer Search with Learning in the Retail Gasoline Market

Abstract: This paper estimates a model of consumer search for gasoline in the presence of two types of imperfect price information: ex-ante uncertainty about each station's price, and uncertainty about the distribution of all stations' prices. We develop a model in which consumers formulate their prior belief of the current price distribution using the prices observed on past driving trips, and then Bayesian update their beliefs with each new price observed, before deciding whether to purchase gasoline or continue searching for a cheaper price. We estimate this model by utilizing a unique data set of station-level daily gasoline sales and prices, combined with data on the empirical distribution of various traffic flows. Our empirical results suggest that consumers place a relatively high weight on new price observations when formulating their beliefs of the overall price distribution. We find that price distribution uncertainty is the primary component of imperfect price information, and if it were eliminated, consumers could achieve 70 percent of the total savings that could be realized by having perfect price information.