Policymakers at the state and federal levels are increasingly pushing to hold institutions accountable for the labor market outcomes of their students. There is no consensus, however, on how such measures should be constructed or how the choice of measure may affect the resulting institutional ratings. Using state administrative data that links postsecondary transcripts and in-state quarterly earnings and unemployment records over more than a decade, the authors of this NBER working paper construct a variety of possible institution-level labor market outcome metrics. They then explore how sensitive institutional ratings are to the choice of labor market metric, length of follow-up, and inclusion of adjustments for student characteristics.
The paper also examines how labor market metrics compare to the academic-outcome-based metrics that are more commonly incorporated into state accountability systems. The authors conclude that labor market data provides information that is distinct from students’ academic outcomes. Institutional ratings based on labor market outcomes, however, are sensitive to the specific metric. The choice of metric and length of follow-up appear to matter much more than compositional adjustments. The findings suggest a cautious approach: While a mix of feasible labor market metrics may be better than none, reliance on a single metric, especially if measured very early, may undermine policymakers’ ongoing efforts to accurately quantify institutional performance.