Personalized Pricing and the Value of Past Purchase Histories: An Empirical Perspective

We investigate the profitability of information acquisition about consumers’ preferences and the implementation of personalized pricing in a multi-product monopolist market structure. Our analysis uses data on prices, aggregate quantities, and individual purchase histories from a large supermarket chain in the US and an empirical model to represent grocery shopping by consumers and the supermarket pricing strategies. We estimate a large set of demand systems and supermarket marginal costs consistent with the observed uniform price setting. With the estimated distribution of preferences in hand, we simulate the information acquisition by the supermarket from purchase histories, assuming the supermarket uses Bayes’ rule to update its priors about consumers’ preferences. We then evaluate how profitable it is to set personalized prices using the information contained in purchase histories. Our results show that the supermarket can reap between 60% to 80% of the potential gains from perfect price discrimination, depending on the product category. At the same time, aggregate consumer surplus decreases but some consumer types are presented with supra-uniform prices while others with infra-uniform prices. Keywords: personalized pricing, differentiated goods, demand calibration, price competition. JEL Classifications: L11, L13, L81