Dynamic food demand and habit forming behaviors: Bayesian approach to a Dynamic Tobit panel data model with unobserved heterogeneity

Incorporating dynamics such as habit formation in analysis of demand can make estimation more reliable and help to explain the “stickiness” in consumer demand behavior when consumers receive new information about products, such as a food safety event or recall. Scanner data allow many repeated observations of the same household so are ideal for analyzing the impact of habit on demand. In addition to that, scanner data allow us to easily observe the presence of zero purchases. The presence of zero purchases is an important econometric issue in empirical modeling on food demand in the sense that ignoring the censoring issue could lead to biased estimation results. This paper investigates the impact of state dependence on food demand using Nielsen 2009 and 2010 HomeScan data. In this paper, we take into account the censored nature of food expenditure data and employ a Bayesian procedure to estimate the dynamic demand models on dairy products. By controlling the individual heterogeneity in the model the source of endogeneity for the lagged dependent variable is removed