Modeling Preferences for Common Attributes in Multi-Category Brand Choice

Abstract

Characteristics based discrete choice models of demand have been used extensively in both economics and marketing. The basic endeavor in these models is to view products as bundles of characteristics, with consumer preferences defined over this characteristics space. In the context of brand choice in packaged goods categories, Fader and Hardie (1996) show the advantages of this approach in terms of parsimony, as well as model fit. More importantly, this modeling approach lends itself to important applications, such as predicting demand for new products. In this paper, we propose a multi-category brand choice model that is based upon the conceptualization that the intrinsic utility for a brand is a function of underlying attributes or characteristics, some of which are common across categories. Our general premise is that preferences for attributes that are common across categories are likely to be correlated. Further, we project the unobserved component of preferences for attributes and sensitivities to marketing mix variables to a lower dimensional space of unobserved factors. The factors are interpretable as unobservable household that explain similarity in choice behaviors across categories. Since the traits transcend categories, we can use household specific factor estimates derived from purchasing in existing categories to predict preferences for attributes in new categories. The proposed model is applied to household panel data for three closely related snack categories, and for two less related food categories. We find strong correlations in preferences for product attributes, such as brand names and low-fat or fat-free. In two cross-category targeting applications, we demonstrate that these high correlations in product attribute preferences across categories imply that (1) one can use the model estimates to improve forecasts of preferences for an attribute in a new category, and (2) that one can score potential targets for a new product in an existing category based on prospects' probability of choice.