The Impact of Big Data on Dynamic Personalized Pricing: Theory and Evidence

Technologies that exist today allow firms to identify and track individual customers online. The possibility for firms to engage in personalized pricing have dramatically expanded with the advent of large datasets (Big Data) on consumers' individual characteristics, purchasing, and web-browsing behavior together with improved artificial intelligence (AI) techniques and the increasing role of customer feedback through social media. One concern about these developments is that firms in online markets with access to granular sensitive information about individual consumers may exploit this advantage through advanced algorithms to discriminate across consumers, by charging higher prices to those who are identified to have a higher demand for their products. This project investigates the impact of Big Data and AI on firms' incentives to acquire and disseminate information about market demand, the circumstances under which Big Data and AI intensify or hinder firm competition, and how the intensity of competition affects pricing and consumers' welfare.

The project first develops a dynamic model of competitive price discrimination in the presence of rich information that enables firms to implement sophisticated pricing strategies. The research demonstrates that unlike in a monopoly setting with first-degree price discrimination, in the setting considered in this research, novel forms of inefficiencies can arise. Hence, new types of policy interventions to restore efficiency may be appropriate. The project shows that as long as multiple firms have access to a given consumer's data for the same product, competition between firms can make consumers better off than if no firm had access to such data. The project then applies the developed model to the online sales of smartphones and tablets to quantify the effects of personalized pricing in this market. This exercise is a pilot study of a large and important market for which the pricing strategies considered are natural. The project then applies developed model to new data from a large retail marketplace in order to characterize the policy implications of the new environment for many more industries and products.