Understanding store-brand purchase behavior across categories

This paper investigates whether the tendency to buy store brand is category specific, or an enduring consumer trait. We develop a multicategory brand-choice model with a factor-analytic structure on the covariance matrix of the coefficients. The methodology allows us to elicit the basic latent tendency for a household to buy store brands, while controlling for other causes such as price sensitivity. The model is applied to a set of ten food and nonfood product categories. We find strong evidence of correlations in household preferences for store brands across categories. Using a two-dimensional factor structure, we find that one of the factors explains a substantial amount of variation in store-brand preference, while the other factor explains price sensitivity-consistently across categories. The presence of these factors in all categories indicates that there are unobservable household-level traits that are non-category specific, i.e., stable across product categories. Using data from five holdout categories, we find that household estimates of these latent factors are very useful in predicting demand for store brands in new categories. Other potential applications for store managers are discussed.