Measuring Substitution Patterns and Firm Conduct in Differentiated Product Industries

Project Outcomes Statement

Intellectual merit: This project has developed an approach to estimate the substitutability of products using aggregate data on consumer choices and product characteristics. This is central to analysis of markets with imperfect competition and antitrust regulation, as well the evaluation of policies designed to improve consumer choices (health care, school, insurance).
The measurement of substitution is complicated by the presence of a simultaneity problem: When two products have correlated demands, is it because consumers view their attributes as substitutes, or because they have correlated unobserved attributes? The main theory result demonstrates the existence of a set of instruments that get around this problem and uses available data efficiently. The parameters determining the elasticity of substitution between products are identified by exploiting exogenous variation in the location and availability of rival products with similar attributes. The strength of the relationship between these measures of product isolation and observed market shares determine the weakness of the instruments, and ultimately the precision of the estimates. This leads to a new data-driven approach to the evaluate the strength of the instruments.
Broader impacts: The proposed is now widely used by practitioners. The main working paper has accumulated a large number of citations, and several prominent academic papers use the approach to estimate models of product differentiation in a wide range of context. Second, this research had an important impact on teaching and student training. The method is taught in many elite graduate programs in economics departments and business schools.