Estimating the Effects of Adverse Selection in Used Car Markets

In this paper, I address the long-standing question of whether adverse selection prevents used cars from reaching owners who value them most highly. In doing so, I confront the challenge of identifying the effects of adverse selection separately from the effects of efficient sorting of vehicles based on their conditions. This latter process would usually occur simultaneously to adverse selection and also affects the distribution of vehicles that trade. Using the prediction in Hendel and Lizzeri (1999), that adverse selection and efficient sorting both increase the rate of price depreciation, I propose to use their joint effect as an upper bound on the effect of adverse selection. My estimate of this joint effect, based on proprietary data on one million dealer used car sales and trade-ins, is close to zero, a result that indicates that adverse selection is unimportant. Using Consumer Expenditure Survey data, I provide additional support for this conclusion by showing that vehicles that were recently purchased from a dealership received approximately the same number of repairs as comparable continuously-held vehicles. I conclude with a discussion of the role that sellers’ concerns for their reputations may play in limiting information-based inefficiencies. ∗I am grateful to Steven Berry and Fiona Scott Morton for discussions and guidance. I also thank Romana Primus at Whaling City Ford and Paul Trembley at East Rock Auto Repair for background information on the used car and car repair markets. Philip Schmidt-Dengler and various seminar participants provided useful comments.