Rental car service tenure and downstream vertical integration in the automobile industry

U.S. and foreign automobile manufacturers acquired minority or controlling interest in several of the largest daily rental car firms between 1987 and 1991. The purpose of this dissertation is to develop and estimate a model that explains observed changes in the length of time vehicles are retained in service fleets across the merger regimes. I construct a stochastic dynamic programming model similar to Rust (1987) in which a rental car firm decides to keep or replace a vehicle each period as new information about costs and demand becomes known. The solution to the firm's problem is an optimal stopping rule that defines the distribution of vehicle tenure. The premise of my thesis is that changes in tenure are due to the internalization of two technological externalities at the downstream level. The two externalities are an input-substitution distortion described in Vernon and Graham (1971) and arbitrage costs.;Data on rental car service tenures were collected from archive files at the Virginia Department of Motor Vehicles. One of the contributions of this research is the collection of firm specific data.;I estimate changes in vehicle tenure using a semiparametric hazard rate model. The baseline hazard is specified as a piecewise linear spline function with spikes. The hazard rate is assumed to be higher during certain periods due to restrictions manufacturers place on vehicle purchase agreements. The unobserved heterogeneity component is represented by the gamma distribution.;My estimation results show that the acquisitions had significant effects on the length of time the rental car firms keep vehicles in their fleets. In particular, my results support the hypothesis that the automobile manufacturers internalized two technological externalities when they acquired the rental car firms. Though the results vary across firms, when controlling for manufacturers' vehicle repurchase offers, unobserved heterogeneity and cohort-effects, the input-substitution distortion dominates in two of the five acquisition cases and the arbitrage externality dominates in the remaining three cases. An overall merger dummy variable indicates that the input-substitution distortion dominates.