This analysis uses intra-state differences in unemployment rates during a person’s teenage years as an instrumental variable to address potential selection bias problems in estimating the effects of schooling on adult health outcomes. A higher unemployment rate during a person’s teenage years leads to greater educational attainment because lower wages and fewer jobs reduce the opportunity costs of attending school. From two-stage probit models, a year of school is estimated to reduce the probability of having a worklimiting health condition by 2.6 percentage points and to reduce the probability of requiring personal care by 0.67 percentage points. Both estimates are statistically significant and exceed the estimated beneficial effects of schooling from the corresponding standard probit models. For one other health measure, having a mobility limitation, the two-stage model shows no impact of schooling.
Arkes, Jeremy, Does Schooling Improve Adult Health?. Santa Monica, CA: RAND Corporation, 2003. https://www.rand.org/pubs/drafts/DRU3051.html.
Arkes, Jeremy, Does Schooling Improve Adult Health?, Santa Monica, Calif.: RAND Corporation, DRU-3051, 2003. As of July 15, 2021: https://www.rand.org/pubs/drafts/DRU3051.html