An analysis of consumer holdings of durable goods, with and without transaction costs

Theoretical work on competitive models of consumer durables has been scarce because even simple models of durables are necessarily dynamic and yield closed-form solutions only under very restrictive assumptions. Empirical work on consumer durables has suffered both from insufficient theoretical underpinnings and from the paucity of data which afflicts microeconometrics generally. This investigation addresses each concern, developing both a theoretical model of consumer holdings of durable goods and an empirical application to consumer holdings of automobiles.;Chapters One and Two address the need for improved theoretical analysis of consumer holdings of durables. Chapter One presents a model of individual consumer holdings of durable goods, in which the holdings are motivated by a fixed cost of transacting in the secondhand market. Even a small transaction cost implies that durables which are nearly new are not traded in equilibrium. Furthermore, a sufficiently large transaction cost closes the secondhand market altogether. Chapter Two explores the problem of choosing among durables which may differ in both a depreciable component (such as aggregate mileage or collisions) and a nondepreciable component (such as wheelbase or maximum horsepower). A characterization theorem for equilibria of the model is provided, as well as a brief computational example.;Chapter Three comprises the empirical application to the automobile market. While many investigations of consumer demand for automobiles have employed a static discrete choice framework, we choose a dynamic discrete choice approach which better addresses the fundamentally dynamic nature of the automobile purchase and replacement decision. Estimation is based on Rust's (1988) method for estimating the parameters of discrete control processes and Berkovec's (1985) analysis of equilibrium automobile holdings with an active secondhand market. Data for the investigation come from the Sydney Transportation Panel, which provides detailed information on a large number of individual household vehicle fleets over a four year time span. The resulting model allows for the estimation of consumer choice probabilities which depend only on fundamentals.