Modeling whether, what and how much to buy the right way
Whether, what and how much to buy are central decisions in consumer goods markets. Marketing research commonly uses a sequential approach where quantity decision is conditional on purchase incidence and brand choice (e.g., Ailawadi et al. 2007). This approach assumes separability between decisions and suffers from selectivity bias. The bias can be overcome by explicitly controlling for it (e.g., Zhang et al. 2005) or by using one unifying utility function, a method considered as ”state of the art in analyzing purchase behavior in a single product category” (e.g., Song and Chintagunta 2007). However, this latter method puts restrictive assumptions on the influence of prices on choices, which may affect managerial implications derived from the model results. Our study investigates the effect of selectivity bias by comparing these three approaches. We focus on the extent to which managerial implications arising from these frameworks differ. We use household panel data for three categories which exhibit sufficient variation in variables of interest (e.g., price level, storability). We show that the superiority of one framework versus the other depends on the specific category and its characteristics. The managerial implications of using the ”wrong” framework are demonstrated by conducting two simulation studies; these show that price elasticities substantially deviate across frameworks.
