A Dynamic Model of Cross-Category Competition: Theory, Tests and Applications

Recent marketing studies use scanner data to diagnose the influence of a change in a brand's marketing mix on other brands in the same category. A few studies also use scanner data to model inter-category effects between substitutes (e.g., tea and coffee) or complements (e.g., tea and sugar). No study models the dynamic effects of cross-category competition though.We present a dynamic model that shows how different products in a typical market basket influence demand of one another. Empirical tests use single-source, market-level data for ice cream and some substitute and complement categories. On the basis of in-sample fit, out-of-sample forecasts, and formal causality tests, this study demonstrates that the marketing mix decisions for ice cream influence the sales and market shares of ice cream toppings and frozen yogurt. As a check on the predictive validity of the model, model parameters, using 16-month data, predict behavior in the subsequent 3 months.