动态激励与最优医保支付方式

Dynamic Incentives and Optimal Payment Systems for Health Insurance

本文构建医院、患者、医保机构三方信息不对称条件下的重复博弈模型,研究了医保机构对医院的最优支付方式(最优价格管制)及其决定因素,研究发现:(1)无论医保机构实行预付制或后付制,为激励医院诚实诊疗,最优价格管制应保证医院获得正利润。(2)最优预付制价格可以得到显式解且形式简单,而有效规避医院道德风险的后付制价格组合则依赖于医保机构目标函数、患者收入分布的具体形式。(3)若患者自付比例足够低,则预付制优于后付制;若治疗成本随疾病严重程度变化足够大或医院不重视未来收益,则后付制优于预付制。模型解释了中国与发达国家医保支付方式改革中的几个标准化事实,并提示了我国当前实施医保预付制应注意的一些要点。

Summary： How to control rising medical expenses by reforming the health insurance payment system is a question of concern in China in recent years. This paper studies social health insurance＇s optimal payment systems to hospitals. By viewing medical services as credence goods, we constructed a game-theoretical model with information asymmetry to discuss the repeated games among hospitals, patients, and social health insurance institutions. Social health insurance institutions and hospitals are long-run players in the game, and patients are short-run players. Social health insurance institutions and patients face moral hazard problems related to hospitals： overtreatment, undertreatment, overcharge, or undercharge. With patients heterogeneous in income, the social health insurance institution pays part of the medical expenses for patients and is in a monopsony position, determining payment system and price levels to hospitals. Our conclusions are as follows： （1） To encourage hospitals to be honest in prescription and treatment, optimal payment systems should permit hospitals to obtain profits. （2） The optimal price in a prospective payment system （PPS） has a simple and explicit form, while efficient prices in a retrospective payment system （RPS） depend on specific forms of the social insurance institution＇s objective function and patient income distribution functions, putting a higher requirement on the social health insurance institution＇s ability to collect information. （3） PPS will be better than RPS if the patient＇s copayment/coinsurance is low enough, while RPS will be better than PPS if the variation of treatment costs is large enough, or if hospitals are too short-sighted. Our model interprets phenomena of health insurance payment systems in China and developed countries. Patients＇ copayments/coinsurance are very low in developed countries, which makes PPS more suitable, and it could be implemented for all diseases. China＇s basic medical insurances are organized by governments. With the gradual decrease in the coinsurance ratios of China＇s basic medical insurances, there has also been a trend towards PPS in China. However, Chinese patients＇ coinsurance ratio with basic medical insurance is high compared with that of the developed countries; thus the relative optimality of PPS and RPS may depend on specific types of diseases and medical services, and a special mixed payment system has emerged in China. Our model is also consistent with the following criterion for selection of diseases for PPS adopted by basic medical insurance institutions in China： ＂less complications, mature technology for diagnosis and treatment, quality and cost stability＂. Our model also explains the trend of increasing numbers of diagnosis-related groups （DRGs） in Medicare with the cost variation within a group decreasing. If the product quality is difficult to observe in advance, hospitals may reduce service quality under the fixed price of PPS. The health economics literature uses altruism to circumvent this problem partly, assuming that the quality of medical services is a direct element in the utility function of physicians. Nearly all theoretical studies based on the altruism assumption conclude that some form of mixed reimbursement （for the same case/patient） is optimal. Yet mixed reimbursement is rarely observed in practice. Our dynamic model explains the mechanism of PPS from the perspective of repeated games and reputation, without the altruism assumption. Policy implications of this work include that due to information asymmetry, the social health insurance agencies should not set PPS price down to the average cost of treatments, but should use a reputation premium to avoid hospitals＇ moral hazard. In addition, PPS price should remain relatively stable. Finally, most physicians in China are employees of hospitals. We ignore the difference between physicians and hospitals in the basic model. However, our conclusions are robust if we model explicitly such differences in repeated games among physicians, hospitals, patients, and the social health insurance institution.