| A large body of existing literature explores the relationship between education and health, and finds strong correlations between the two. However, a causal interpretation is undermined by the likely existence of unobserved variables that could affect this relationship. Using panel data on high school graduates and their siblings from the Wisconsin Longitudinal Study, I adopt a sibling-differencing strategy to obtain unbiased estimates of the health return to education. I find that higher levels of education are positively related to self-reported health after taking into account childhood health endowments, unobserved effects and measurement errors in education. Controlling for the unobserved effects in the regression of self-reported health on education lowers the estimated health returns to education by a significant amount. An additional year of schooling improves the probability of reporting excellent or very good health by 2.3 percentage points for men and by 2.1 percentage points for women. These results provide further support for the literature's contention that education positively affects health outcomes. |
