Household heterogeneity and state dependence in a model of purchase strings: Empirical results and managerial implications

Abstract

The problem of separating the effects of household heterogeneity from state dependence in brand choice models is important from a theoretical as well as a managerial perspective. To address this problem we propose a choice model that treats a household's string of purchases as the unit of analysis, in contrast with conventional models of individual purchases. Our formulation accounts for unobserved household heterogeneity in preferences and in responsiveness to marketing variables, as well as Markovian state dependence. We use three scanner panel data sets to test models that differ in the nature and extent of heterogeneity and state dependence accomodated. We find empirical evidence for state dependence as well as heterogeneity in each product category. We demonstrate via simulation of a price promotion how the separation of the effects of state dependence from unobserved heterogeneity enables us to (1) correctly evaluate the profitability of a promotion by explicitly taking into account the multiperiod effects of promotions resulting from state dependence, and (2) prioritize market segments for the targeting of promotional activities.