Consumer Durables and Inertial Behavior: Estimation and Aggregation of (S,s) Rules

This paper presents an (S,s) model for automobile consumption and estimates it using a data set of US households. The model allows for unobserved heterogeneity in both the target level and the band width, takes into account the possibility of a zero desired level, constrains the band to be non negative and allows asymmetric bands. The model is estimated on a novel data set which contains information on both stock values and automobile expenditure for a large number of households observed over a period of a year. The (S,s) rule is specified in terms of the ratio of car stock to non durables. The shortcuts usually employed in the empirical literature on (S,s) rules can be avoided thanks to the richness of the data set and the rigorous specification of the stochastic model. Having estimated the model and considered `goodness of fit' measures, aggregation issues are considered. First, the paper presents a number of negative results. Then, several simulations aimed at evaluating the effects induced by inertial behavior on aggregate dynamics are considered.