Research Productivity in a System of Universities

Abstract

This paper considers research performance of U.S. universities for eight science fields. At the aggregate level we find that research output follows a constant returns to scale process. However, for individual universities we find evidence of diminishing returns. We offer two explanations for these differing results. First, data errors are more important at the individual level. Second, research spillovers exist between universities and fields that are captured only at the aggregate level.

We thank the Mellon Foundation for financial support, and Jian-Mao Wang for excellent research assistance. We thank two referees as well as seminar participants at the University of Florida, the National Bureau of Economic Research, and the Strasbourg Conference for helpful comments.