Abstract
This paper exploits the discrete increase in reimbursement rates at the statutory retirement age as a quasi-natural experiment to examine ex post moral hazard in inpatient care in China. Using nationally representative administrative data covering more than 450,000 hospitalization records across 29 provinces, we implement a cohort difference-in-differences design to identify the causal effect of insurance generosity on healthcare utilization. We find that a 5-percentage-point increase in reimbursement rates increases length of stay by 1.73 days and total inpatient expenditures by RMB 709.98. The additional spending is primarily financed by the insurance pool, while patients' out-of-pocket payments decline, consistent with moral hazard predictions. Decomposition results show that the spending increase is concentrated in diagnostic services, pharmaceutical use, and inpatient care components such as bed and nursing services, rather than high-intensity treatments. Further analysis highlights the role of supply-side institutional constraints: under the Diagnosis-Intervention Packet (DIP) payment system, utilization responses are largely muted, whereas under non-DIP regimes they remain substantial. In addition, moral hazard is concentrated in secondary hospitals. Overall, these findings suggest that moral hazard in inpatient care reflects demand-side price responsiveness and manifests differently across supply-side institutional environments. This institutional heterogeneity also underscores the potential importance of provider payment reform in health insurance policy design.