Analysis of household shopping behavior across retail formats

Today, grocery and drug retailers must develop policies in a competitive environment that includes other retail formats. Mass merchandisers and warehouse club stores, in particular, offer pricing, promotion and assortment policies that are different from those of traditional retailers. This research examines how these policies affect revenues of retailers in different formats, incorporating competition from other stores of the same format as well as other formats. We specify a Multivariate Tobit mixture model that captures two key shopper decisions: which store chains to shop, and how much to spend in each store chain. We model household packaged goods expenditures in each store chain as a function of the retailer's marketing policies, other unobserved retailer-specific factors, travel costs and household demographics. We apply the model to the packaged goods expenditures of 175 households over a two year period at the ten largest store chains from five retail formats in a major geographic market. Our results suggest that, overall, store assortments have a greater effect on shopping behavior than pricing and promotions. Retailers' feature advertising activity affects shopping behavior, while unadvertised discounts do not. Also, shoppers' travel times negatively affect shopping, an effect that is consistent across formats. Interestingly, all of these effects are much greater for grocery stores than for non-grocery stores. The differential effects lead to substantially different recommendations for traditional retailers trying to maximize returns in light of competition from multiple formats. The research contributions of this paper are threefold. One, we conduct the first panel data study which captures packaged goods spending both in grocery stores and other formats. Second, we go beyond the common choice modeling framework to include household spending. This is similar in spirit to adding a purchase quantity decision to a brand choice model. Because the variation in household spending directly affects retailer sales and profits, expenditures are an important consideration for retail decision makers. Third, our specification of a Multivariate Tobit model is the first such application in the marketing literature.