Structural Estimates of Depreciation from Wholesale Auctions of Used Ford Windstar Vans

This paper estimates the depreciation of over twenty varieties of Ford Windstar passenger vans, using several hundred thousand observations from a deep collection of wholesale auction transactions. The data permit the estimation of individual vehicle ages apart from differences between resale- and model-years, so that age-, date-, and vintage- effects are all identifiable. Depreciation is modeled in two parts, estimated separately: obsolescence, wherein new model-years depress the resale values of older cohorts uniformly; and ordinary wear-and-tear, which reduces the resale values of individual vehicles in a nonuniform manner summarized by different individual-level service-lives. Obsolescence (and inflation) are estimated first, in a regression setting of third-order functions of age and miles to hold individual effects constant, allowing varietal and quarterly time dummies to trace each variety's counterfactual "as-if-new" price through time in a way that upholds the useful fiction of a Hicksian aggregate. The estimated as-if-new prices then anchor individual-level resale-price profiles, which are embedded in implicit service-life densities that vary over time and age. Quarterly moments of prices and miles are then matched by adjusting parameters for wear-and-tear.