Measuring Adverse Selection in China's New Cooperative Medical Scheme

In 2003, the Chinese government implemented a subsidized voluntary health insurance program called the New Cooperative Medical Scheme (NCMS) to alleviate rural population from medical impoverishment. Studies have found NCMS failed to provide sufficient financial protection. The most cited reasons include its limited benefits (e.g., large deductibles, low ceilings, and high coinsurance rates) and poor design of benefit packages. The central government conditioned NCMS subsidy contribution to poor provinces on a minimum of 80% participation level, and require the program premium to be fixed across households, then left the rest of detail benefit designs to local governments. As a result, local governments faced a trade-off between increasing NCMS benefits to meet the program participation prerequisite and minimizing the variable program costs given the fixed program premium per person. One obvious cost minimization way is to distort NCMS benefit plans to avoid high-risk participants, which is the classic adverse selection problem. However, this counters the initial program goal of minimizing medical impoverishment. To improve program effectiveness, it is crucial to understand the degree of distortion to inform program modification. This study fills that gap and identifies those health services that are under- or over-covered due to NCMS adverse selection. We also explore possible ways to improve the NCMS’s design through building in incentives targeting to address adverse selection problem.

Using principal-agent theory, we characterize how the amount of NCMS benefits are distorted from their optimal efficiency level using service-specific shadow prices. A shadow price is defined as a threshold of government assessed marginal valuation that a household has to exceed to qualify for receipt of a service. NCMS participation decisions were modeled at the household level based on household’s expected benefits. Then the local government chooses optimal shadow prices for different benefits based on its assessment of households’ contributions and health spending. Under perfect and then imperfect information flow, we show how the information asymmetry between households and local governments distorts the NCMS benefits and how shadow prices change when the premium follows the risk-adjustment systems.

We measure shadow prices using the China Health and Nutrition Survey data. Under additional assumptions regarding homogeneous preference and valuation function, the estimated shadow prices mainly depend on household expected benefits of NCMS and the risk premium paid. We use a two-part model to estimate the expected benefits. Although the actual NCMS risk premium is fixed across individuals, we investigate whether and how better risk adjustment algorithm can improve the program efficiency.

Preliminary results suggest that as the information asymmetry increases, local governments appear to under-provide coverage for hospitalization services, while over-provide coverage for preventive health services, perhaps because of the incentives to avoid high-cost enrollees or attract relatively healthy participants. This finding confirms the concerns that in a short run NCMS failed to provide financial protection to rural populations even with high participation rates. Another implication is that the efficiency of NCMS can be improved if its risk premiums are adjusted based on individual demographic characteristics and disease history.