Estimating Dynamic Demand in Differentiated Durable Goods Markets Using Price Alert Thresholds

Retail prices are volatile and decrease over time in many markets for differentiated durable goods. Consumers in these markets have an incentive to delay their purchase and wait for lower prices. This paper uses a novel data set from a price alert service for TVs sold on Amazon.com to estimate substitution patterns across products and over time. Users of a price alert service submit a price threshold for a product they want to purchase and receive an alert when their threshold is reached. I estimate consumer preferences using a discrete/continuous-choice model in which the price threshold is the solution to an optimal stopping problem. The estimates imply that many consumers respond to a price increase by delaying their purchase rather than by choosing a different TV. This suggests that TVs mostly compete with their own future selves rather than with other TVs. To measure how much consumers benefit from delaying their purchase I consider a counterfactual in which consumers buy a TV immediately upon arrival at the market. ∗I am deeply indebted to my advisors Amit Gandhi, Ken Hendricks and Alan Sorensen for their guidance and support. I also thank Rasmus Lentz, Jack Porter, Dan Quint and Yu Zhu for their helpful comments. I am very grateful to Dan Green and the team of camelcamelcamel.com for providing the data which made this project possible. All remaining errors are my own. Contact: Department of Economics, University of Wisconsin-Madison, 7310 Social Science, 1180 Observatory Dr., Madison, WI 53706, [email protected].