Association for Public Policy Analysis & Management

This paper uses rich administrative data and a quasi-experimental design to study how public higher education expenditures affect labor market outcomes for students, and how this varies by institution type and major. The motivation for this exercise stems from two stylized facts. First, per-student resources have fallen substantially at public and particularly public non-selective higher education institutions in recent years, and this decrease has coincided with a decline in academic outcomes such as graduation rates (Bound et al. 2010; Deming and Walters 2017). Second, these spending declines are heterogeneous by field of study, with some of the largest declines coming in high-return STEM majors (Altonji and Zimmerman 2017).

Our approach relies on unique data on academic and labor market outcomes for the population of students in the Arkansas higher education system from the early 1990s through 2010s. We link these records at the classroom level to data on instructors, including instructor salaries. These two data sources allow us to measure the instructional resources allocated to each course section.

We combine these data with exogenous variation in resources at the institution by major level generated by the interaction between changes in state policy that affect (and generally reduce) funding at the institution level and differences in departments’ capacity to adjust faculty inputs. The intuition is that in the short- to medium-run, budget cuts reduce instructional resources more in departments where more faculty have short-run contracts. This approach let us trace out how changes in resources within institution and field affect persistence, graduation, and labor market outcomes.