The Decision of When to Buy a Frequently Purchased Good: A Multi-Period Probit Model

Abstract

Increased availability of scanner-based panel data has enabled researchers to better understand nondurable commodity purchase dynamics. In this study, we focus on one component of the purchase process--when to buy. The relationship between the discrete purchase decision and a set of household and purchase characteristics is quantified using a simulated maximum-likelihood procedure. Given the longitudinal nature of our data, unobserved heterogeneity is addressed by adopting an auto-correlated error structure. Our empirical application is household purchases of cheese. We find evidence of significant persistent unobservable household heterogeneity, which is not eliminated by the inclusion of lagged exogenous variables.

Keywords

• autocorrelation
• discrete decision
• panel data
• simulated maximum likelihood