Complementarities in College and High School Investments

Abstract
We estimate the complementarities between multi-dimensional abilities, high school investments, and college investments in labor market returns. Using administrative data from Sweden, the analysis includes a rich set of observables; latent cognitive, grit, and interpersonal abilities; high school specialization in vocational, academic, and STEM tracks; and college major choices. First, we provide non-parametric evidence of sorting and the strong complementarities between abilities, high school investments, and college investments. Second, we estimate a generalized Roy model that accounts for additional unobserved heterogeneity using within-school-across-cohort variation in specialization choices and discontinuities in college admission. We use the estimated model to construct counterfactuals that compare (i) marginal and average returns and (ii) policies that induce marginal individuals to change high school or college investments. We find evidence of comparative advantage for most high school tracks and college majors. Nevertheless, marginal returns can be as high as average returns. We find that specialization of investments can lead to either positive or negative dynamic complementarities. Investing in more challenging tracks in high school increases college enrollment and graduation, but not necessarily wages. Specializing in STEM in high school would benefit 80% of marginal individuals and lead to higher wages compared to a more balanced academic track. Particularly for those who never go to college, pursue shorter college degrees, or pursue engineering or health sciences degrees in college. In comparison, the average marginal return to encouraging more STEM college applications is small. Our results suggest that earlier STEM investments may be more effective than later STEM investments. Joint paper with John Eric Humphries and Gregory Veramendi.