Identification and Estimation of Demand for Bundles

We study the identification and estimation of a mixed logit model of demand for bundles. We generalize the original model proposed by Gentzkow (2007) in three ways. First, we allow the demand synergy parameters (capturing complementarity and substitutability) to be bundle-individual specific and treated as random coefficients. Second, we allow the joint distribution of the random coefficients to belong to any parametric family. Third, our arguments are not specific to the three-bundle case but are directly developed for choice sets of any size. We propose sufficient conditions for identification and for lack of it. Our sufficient conditions for identification also guarantee consistency and asymptotic normality of standard MLE and GMM estimators, which are robust to both price endogeneity and sampling error in the observed market shares of bundles. Finally, we use our methods to investigate the welfare implications of bundle-level pricing strategies in the ready-to-eat (RTE) cereal industry in the USA. Preliminary results highlight the existence of strong demand synergies among different RTE cereal brands and an interesting interaction between mixed-bundling pricing and market structure: mixed-bundling pricing increases firms’ profits only in the absence of competition (i.e., perfect collusion or monopoly), while it hurts both consumers and firms as soon as some competition is present.

Please sign up for meetings below: