Panel Paper: Postsecondary Supply and Unemployment Dynamics

Thursday, November 8, 2018
Marriott Balcony B - Mezz Level (Marriott Wardman Park)

*Names in bold indicate Presenter

Rajashri Chakrabarti, Federal Reserve Bank of New York and Michael F. Lovenheim, Cornell University; National Bureau of Economic Research


We examine the contribution of postsecondary supply conditions – the number and composition of higher education institutions – on the dynamics of unemployment in a local area. Adverse labor demand shocks cause workers to enroll in college (Christian 2007; Hershbein 2012; Aguiar, Hurst and Karabarbounis 2013; Long 2014), and their decisions of where to enroll are heavily influenced by the composition of postsecondary institutions in their area (Armona, Chakrabarti and Lovenheim 2016). There is a large amount of heterogeneity across institutions in focus, the type of degree and certificate programs, resources and productivity. As a result, the postsecondary institutions in some areas will be better positioned to retrain workers than others, potentially creating a link between higher education supply conditions and unemployment dynamics. This link has not been examined in prior research, but it is critical from a policy perspective to support optimal higher education investments by local, state and federal governments.

Our analysis begins with a descriptive examination of how unemployment dynamics vary with the postsecondary environment, which has not been previously explored by researchers. At the CBSA level, we estimate how unemployment rates as well as the frequency, length and magnitude of recessions vary with higher education supply conditions. This descriptive analysis is critical for establishing broad correlational patterns, but it is not causal because postsecondary supply is not randomly assigned across space.

We then implement an empirical strategy to tease out the mechanisms underlying this relationship. We use shift-share local labor demand shocks (Bartik 1991) that interact the local industrial mix in a base year with national changes in demand in each industry as a labor demand instrument. Using this measure combined with IPEDS data on the types of degrees produced by each institution, we estimate how higher education supply conditions in the base year affect the number and types of degrees produced. Do certain supply conditions facilitate more skill upgrading, and are the skill increases in areas that maintain higher labor demand? Using CPS and ACS data, we then estimate how the interaction of local labor demand shocks and postsecondary supply influences the occupational and industrial composition of workers in an area. Because we can pinpoint the industries driving the labor demand reduction in the shift-share instrument, we can see whether certain higher education supply conditions facilitate a shift from low-demand to high-demand industries and occupations. Finally, we calculate the implications of these effects for unemployment dynamics, thereby linking the first descriptive part of the paper with our causal estimates.