We estimate the impact of unemployment on health using the French database “Enquête Santé et Protection Sociale” (ESPS) over the period 2010-2014. We apply a difference-in-difference propensity score matching approach to identify the health effect of unemployment. By combining both methods, we minimize selection bias and remove unobserved individual fixed effects that are time-invariant as well as common period effects. Our results suggest that unemployment has no significant effect on health. We then use a bivariate Probit model to analyze the two-way causality between health and employment status. Our estimates confirm that unemployment has no significant causal impact on health. However, health significantly affect employment status.
- propensity score matching
- dynamic bivariate Probit model