Abstract: Based on the date of China Family Panel Studies 2018, this paper chooses the Bivariate Probit model and the instrumental variable Probit model to test the correlation between health risks and insurance and to study the selection effects in the supplementary medical insurance markets. The results show that the “health risk-insurance”correlation is significantly negative in the supplementary medical markets. The groups self-assessed as relatively healthy people are more willing to participate in the supplementary medical insurance. This paper confirms the robustness of the empirical results by making a robustness analysis with the more objective proxy variables and the adjusted hospitalization costs as the proxy variables of health risks. The results of the heterogeneity analysis show thatthe health risks and risk aversion jointly make clear that the selection effect is advantageous selection in China's supplementary medical insurance markets. In the relatively immature supplementary medical insurance markets, personal information is multi-dimensional; the relative risk aversion can exert a significantly impact on the decision to participate in the supplementary medical insurance.

Key words: supplementary medical insurance, positive selection, adverse selection, private information