Optimal Bundling: Characterization, Interpretation, and Implications for Empirical Work

This paper studies pure bundling. Specifically, I show that, under some conditions, a firm optimally chooses to sell only the full bundle of a given set of products if and only if the optimal sales volume of the full bundle is weakly larger than the optimal sales volume for any smaller bundle. I argue that this characterization can be interpreted as follows: pure bundling is sub-optimal when there is considerable variation across consumers in how complementary they find disjoint sub-bundles, and/or when this variation correlates negatively with their price sensitivity. I then demonstrate{using simulated data and a random-coefficient discrete choice demand model{that capturing these two variations is indeed crucial for model selection in the empirical analysis of bundling decisions.