Using a Variety-Seeking Model to Identify Substitute and Complementary Relationships among Competing Products

Abstract

Current techniques using brand-switching data to infer competitive relationships among products are based on the assumption that those brands most frequently switched between are the closest substitutes. For product classes in which variety seeking is an important determinant of consumer behavior, this assumption is probably violated. A technique is proposed which accounts for the fact that consumers may switch between brands not because the brands are substitutes, but because the brands complement one another to meet the consumers’ desire for variety. The approach utilizes constrained nonlinear optimization to determine model coefficients that describe substitution and complementarity among brands. The proposed technique is illustrated with individual consumption histories for a frequently purchased packaged good and is shown to extend the range of insights provided by current brand-switching techniques.