Investigating Household Purchase Timing Decisions: A Conditional Hazard Function Approach

The purchase timing decision is an important component of the dynamics of a household's purchase behavior. This decision is influenced by marketing and other variables, and the modeling of this dependence has recently received attention in the literature. In this paper, we build on previous studies and develop a comprehensive stochastic model that incorporates the major factors influencing interpurchase times. Specifically, we use a generalized version of Cox's proportional hazard model to test among competing probability distributions for the interpurchase times while incorporating effects due to marketing variables, observed household characteristics, and unobserved heterogeneity across households. The empirical finding from analyzing the IRI coffee data, suggests that the interpurchase times cannot be adequately described by probability distributions such as exponential, Erlang-2 or Weibull. The effects of unobserved heterogeneity are significant, and they impact the estimates of the effects of the covariates. We also find that a nonparametric procedure for estimating the effects of unobserved heterogeneity provides the best overall fit to the data and yields covariate estimates that are more consistent with prior expectations. Our model is validated by replicating the substantive empirical findings on an additional product category.