Do benefits in kind or refunds affect health service utilization and health outcomes? A natural experiment from Japan.

Although the payment systems of public health insurance vary greatly across countries, we still have limited knowledge of their effects. To quantify the changes from a benefits in kind system to a refund system, we exploit the largest physician strike in Japan since the Second World War. During the strike in 1971 led by the Japan Medical Association (JMA), JMA physicians resigned as health insurance doctors, but continued to provide medical care and even health insurance treatment in some areas. This study uses the regional differences in resignation rates as a natural experiment to examine the effect of the payment method of health insurance on medical service utilization and health outcomes. In the main analysis, aggregated monthly prefectural data are used (N=46). Our estimation results indicate that if the participation rate of the strike had increased by 1% point and proxy claims were refused completely, the number of cases of insurance benefits and the total amount of insurance benefits would have decreased by 0.78% and 0.58%, respectively compared with the same month in the previous year. Moreover, the average amount of insurance benefits per claim increased since patients with relatively less serious diseases might have sought health care less often. Finally, our results suggest that the mass of resignations did not affect death rates.