Estimating the causal effects of private health insurance in Brazil: Evidence from a regression kink design.

Despite having free access to the public health system, 25% of the Brazilian population have a private insurance plan, which is subsidized by the government by means of an income tax rebate. This paper explores this rebate to tackle the potential endogeneity between private insurance and the demand for health services, using the fiscal incentive as a source of quasi-experimental variation in insurance prices. We estimate the average effect of private insurance for the marginal individuals at the kink points by means of a nonseparable nonparametric regression model. Our data allow us to disentangle moral hazard from adverse selection effects and the results indicate that private insurance has a positive impact on the use of preventive services, health outcomes, physical exercises, and smoke quiting, and does not impact the use of nonpreventive health services, such as inpatient services and surgeries.