Moral hazard issues arising from health insurance have been heatedly discussed among researchers and policy-makers. The focus of the existing literature has been on so-called “ex post moral hazard”, i.e., conditional on one’s health status, the responsiveness of an individual’s demand for healthcare to the out-of-pocket prices (Pauly, 1968). By contrast, another type of moral hazard (ex ante moral hazard), namely health insurance alters the behavior of an individual’s health investments, has received less attention in studies (Einav and Finkelstein, 2018). Following the mainstream literature, this paper also abbreviates “ex postmoral hazard” as “moral hazard”. In theory, there is no clear prediction on whether or not health insurance will increase the utilization and the costs of healthcare. A series of randomized experiments [e.g., Aron-Dine et al. (2013), Finkelstein et al. (2012), Manning et al. (1987), etc.] as well as some quasi-natural experiment studies [e.g., Chandra et al. (2010), Chandra et al. (2014), Shigeoka (2014), etc.] have provided compelling evidence of the existence of moral hazard in health insurance. Meanwhile, some studies also tend to characterize the magnitude of moral hazard by the price elasticity of demand for healthcare [e.g., Cutler and Zeckhauser (2000), Feng et al. (2020), Zhao et al. (2015), etc.]. Their estimates of price elasticity are about -0.1 to -0.7.
Therefore, it can be seen that the evidence about the magnitude and nature of moral hazard in health insurance is still limited, though its existence has reached a consensus. In developed countries, such as the U.S. and Europe, largescale health insurance expansion shocks have been rare after the 1980s. In China, however, emerging health insurance reforms and unique data provide researchers opportunities to answer new questions about moral hazard in health insurance. Based on this, this paper shows how to use China’s scenarios and data to define and discuss four research questions, which include: (1) whether monetary costs are the only source of healthcare costs; (2) whether the monetary costs of healthcare (out-of-pocket prices) faced by individuals are marginal prices(spot prices) or average prices (future prices); (3) how to distinguish moral hazard from income effects; (4) how large is the general equilibrium effect of health insurance expansion.
When talking about healthcare costs, the existing literature tends to focus only on monetary costs, without attention to time costs. In order to explore whether the time cost will affect the demand for healthcare (Question 1), we use the fact that the statutory retirement age for men in China is 60 and Urban Employee Basic Medical Insurance (UEBMI) data of male workers aged 50-70 in a city, to construct regression discontinuity models. Preliminary results show that at age 60, there is a signifi cant increase in the rate of hospitalization for men. This implies that the reduction of time costs resulted from retirement will increase the individual's demand for medical care, and thus the researchers should incorporate time costs into analyses to gain more accurate estimate of the price elasticity of demand for healthcare.
Question 2 is raised from nonlinear health insurance contracts. In order to alleviate the overuse of medical care, typical health insurance contracts often incorporate deductibles, cost limits, and other components, and also reset reimbursement annually based on a calendar year, which make patient cost sharing become a nonlinear function of the cumulative amount of healthcare spending (over the covered year). Because of it, when estimating the price elasticity, researchers need to consider whether the insured response to the spot price (i.e., current out-of-pocket price) or the future price (i.e., expected yearend price). At present, few studies can discuss the patients’ response to spot price and future price at the same time as well as compare their magnitudes. We fill this gap by designing two quasi-natural experiments, with the help of the policy for reaching the deductible in outpatient services and the accounting year reset policy of the UEBMI insured in a China’s city. In Experiment 1, only the spot price changes without changes in the future price, while in Experiment 2, both the spot price and the future price change simultaneously, which helps us jointly estimate these two price elasticities. Results show that under the nonlinear contracts, individuals respond differently to different prices: the spot price elasticity of the retired workers is about -0.09 to -0.14, whereas the future price elasticity is about -0.92 to -1.01. This suggests that researchers should incorporate individuals, dynamic incentive responses into the analysis of health insurance.
When observing the phenomenon that falling out-of-pocket prices lead to the increase in healthcare spending, most current studies generally treat it as the evidence of moral hazard in health insurance. However, the impact of price reduction on the consumption includes substitution effects and income effects. The so-called moral hazard in health insurance is the substitution effect, that is, the social deadweight loss due to price distortions. Instead, the income effect might relax the individual's liquidity constraint, which is social benefits. In order to distinguish the moral hazard effect from the potential income effect (Question 3), we exploit the changes in inpatient reimbursement and the changes in pension to construct two quasi-natural experiments, and thus decomposing the traditional price elasticity of demand for healthcare into moral hazard (deadweight loss) and the income effect (social benefit).Results from regression discontinuity design and event study approach show that, the substitution effect accounts for about 58% of the price elasticity, while the income effect accounts for about 42%. This implies that the traditional price elasticity of healthcare demand might overestimate the magnitude of moral hazard in health insurance. When assessing the welfare effects of health insurance reform, researchers should take both the social costs of moral hazard and the benefits of income effects into consideration.
In Question 4, this paper discusses the general equilibrium effect of health insurance. The existing literature often uses a partial equilibrium framework, that is, given the entire medical market remains unchanged, to discuss the impact of health insurance expansion or changes in the reimbursement rate on various outcomes. However, health insurance reforms might affect individuals’ decision making (demand side) as well as providers’ behavior (supply side). We evaluate the impact of China’s New Cooperative Medical Scheme (NCMS) expansion, and find that it not only affects individuals’ healthcare utilization and health outcomes, but also has significant impacts on the public hospitals’ income and expenditure, and increases the number of start-ups and patent applications in the pharmaceutical industry. These preliminary results suggest that the impacts of health insurance on providers’ behavior cannot be ignored. In future studies, researchers should broaden their analytical perspectives and further explore the general equilibrium effects of health insurance.
We believe that Chinese researchers can dig out newer research questions as well as fresher analytical perspectives by combining Chinese reality with big data and the latest empirical approaches. China’s stories and experiences will boost the development of literature on moral hazard in health economics and provide valuable lessons for the improvement of the health financing system in China and other countries.