The efficient moral hazard effect of health insurance: Evidence from the consolidation of urban and rural resident health insurance in China

Ex post moral hazard, the additional healthcare utilization induced by health insurance, can be decomposed into an efficient portion generated by the income effect and an inefficient portion caused by the substitution effect, which has been discussed theoretically, but few studies have provided evidence of the efficient moral hazard. In 2016, the Chinese government launched the consolidation of urban and rural resident health insurance at the national level. After the consolidation, insurance benefits for nearly 800 million rural residents got improved. This paper uses a nationally-representative sample of 30,972 individuals from the China Health and Retirement Longitudinal Study (2011-2018) and adopts a 2-step empirical approach with the difference-in-differences method and the fuzzy regression discontinuity design to estimate the efficient moral hazard in the consolida-tion among rural residents. We find that the price shock contained in the consolidation increases inpatient care utilization, and the corresponding price elasticity is between-0.68 and-0.62. Further analysis shows that the efficient moral hazard resulting in welfare gains accounts for 43.33%-66.36% of the additional healthcare uti-lization. These findings highlight the necessity of evaluating the efficient moral hazard when analyzing the cost -benefit of health insurance reform.

Automated summary

This study examines the efficient moral hazard in the consolidation of health insurance among rural residents in China. Using a nationally-representative sample and a 2-step empirical approach, the study finds that the consolidation increases inpatient care utilization and that the efficient moral hazard accounts for a significant portion of the additional healthcare utilization. These findings underscore the importance of evaluating the efficient moral hazard when analyzing the cost-benefit of health insurance reform.