Based on the data of the 2023 U.S. Medical Survey

this paper empirically analyzes the causal effect and distribution heterogeneity of employer's medical insurance on the medical expenditure of the elderly group by using the instrumental variable method and quantile regression model. The results show that: firstly

the instrumental variable method shows that the traditional OLS overestimates the promoting effect of insurance

and controls the actual reduction of insurance expenditure after adverse selection. Secondly

quantile regression reveals that the marginal effect of insurance on the low-spending group is significantly higher than that of the high-spending group reflecting the coexistence of the release of prevention demand and the constraint of the upper limit of protection. Finally

the marginal effect of chronic diseases decreases with the increase of expenditure quantiles

highlighting the value of early intervention. The results show that medical insurance has the dual effects of demand release and moral hazard inhibition

and the policy design should weaken adverse selection through actuarial pricing and target the low-spending group to optimize the supply of preventive services.