The Competitive Effects of Multi-Store Shopping: Uncovering Complementarities across Bundles

The nature of competition between entire retail storefronts has always been largely debated in antitrust. Looking at product offerings gives suggestive evidence to whether stores directly compete. I propose an estimator that uses consumer behavior to evaluate how much consumers treat entire storefronts as complements or substitutes. I develop the estimation strategy for the level of complementarity between two store-fronts, taking into account repeated purchase behavior over a long panel. I first set up the challenges of using a classical complementarity framework to estimate this. Then I propose the estimation strategy which has two major steps. The first step is a nested logit discrete choice framework to come up with preference parameters for consumers across different items. This step consists of estimating both an inner and outer nest. The next step of the estimation strategy is to use those preference parameters to construct a complementarity term across each pair of retailers in the sample, in line with our intuition about complementarity across bundles. Finally, I report these estimates as correlations between observable characteristics across the two stores and show that generally consumers that shop at multiple stores tend to value one store having high brand variety and might use this store as a focal store in their shopping trip that they might rely on for the coverage of variety.