Personalized Pricing on Campus

While first-degree price discrimination is a stable of theoretical research and economics pedagogy, its study is rare because it is hard to implement and harder to observe. Personalized pricing is common in higher education, however, providing a useful testing ground as well as an economic context of interest in its own right. I use data on over 70,000 undergraduate students admitted to a major public U.S. research university over multiple years. Exploiting a source of exogenous variation in pricing arising from tuition differences between reciprocal and non-reciprocal states, I am able to credibly identify the price sensitivity parameter. I also make use of machine learning techniques to exploit a large number of potential explanatory variables to include in the demand model without overfitting. I find that price discrimination could raise revenue by 5 percent above revenue-maximizing uniform pricing and 15 percent above current practice, holding enrollment – and costs – constant. I also explore the welfare weights attached to different sorts of applicants that are implicit in the school’s current pricing policy.