Investigating purchase timing behavior in two related product categories

The authors' objective is to investigate the purchase timing behavior of households in two product categories for which the decision of a household to make a purchase in one category at a given point in time influences purchase in the other category. Interdependence in purchase timing between the two categories is modeled using a bivariate hazard function approach that accounts for unobserved heterogeneity. The authors provide empirical results from five durable and nondurable goods product category pairs-pasta and pasta sauce, liquid and powder detergents, clothes washers and dryers, soup and yogurt, and room air conditioners and dishwashers. The latter two categories are included in the analysis to ensure that the model can distinguish between true and spurious dependence caused by categories being purchased together. The authors find that the estimated cross-category correlation parameter obtained from the bivariate hazard model is consistent with prior expectations in each of the product categories. As in the literature on univariate hazard models, the authors find the bivariate hazard functions to be nonmonotonic in nature. Using the proposed model, the impact of price reductions in one category on the purchase probabilities in both categories is examined using the pasta/pasta sauce product pair. Given the positive relationship in the purchasing in these two categories, the authors find that a price cut in one category increases the purchase probabilities of both pasta and pasta sauce.