Is educational attainment a cause of better health? A test of conventional wisdom

Naomi Duke, University of Minnesota
Ross Macmillan, Università Bocconi

Research routinely finds a strong association between educational attainment and better health. The conventional interpretation of this association is causal, premised on basic ideas of education and human capital enhancement. An alternative perspective views educational attainment as somewhat endogenous given cognitive and non-cognitive skills that are formed early in the life course. By implication, this perspective would view the association between educational attainment and health as spurious. Using data from the NLSY97 and dynamic measures of both educational attainment and self-rated health, we evaluate these two perspectives. Specifically, we fit conventional ordinary least squares and maximum likelihood, fixed effects regression models where the latter can control for time-stable, unmeasured heterogeneity such as cognitive and non-cognitive skills. Contrary to conventional wisdom, results provide little support for the human capital and causation interpretation. Specifically, once controlling for unmeasured heterogeneity, the effects of education are either eliminated or reduced such that they would be deemed trivial to small. These conclusions are reinforced when we include a set of time-varying covariates that are robust predictors of health and when we examine such effects for six race-sex subgroups. We conclude by discussing the implications for future research on socioeconomic stratification and health.