توسعه بیمه سلامت برای جمعیت روستایی: بررسی تاثیر طرح جدید پزشکی تعاونی چین

In 2003, China launched a heavily subsidized voluntary health insurance program for rural residents. We combine differences-in-differences with matching methods to obtain impact estimates, using data collected from program administrators, health facilities and households. The scheme has increased outpatient and inpatient utilization, and has reduced the cost of deliveries. But it has not reduced out-of-pocket expenses per outpatient visit or inpatient spell. Out-of-pocket payments overall have not been reduced. We find heterogeneity across income groups and implementing counties. The program has increased ownership of expensive equipment among central township health centers but has had no impact on cost per case.

Several developing countries have recently used tax revenues to subsidize health insurance for informal-sector (usually rural) workers and their families, or at least the poorer ones among them. In Colombia, the Philippines and Vietnam, for example, the poor are enrolled in the national social health insurance scheme at the taxpayer's expense. The rest of the informal sector either have the option of enrolling (in the cases of the Philippines and Vietnam) or are required to enroll (in the case of Colombia). In all three countries, the household enrolls at its own expense though the contribution paid by non-poor voluntary enrollees is sometimes subsidized (it is, for example, in the case of Vietnam). In China and Mexico, by contrast, households not covered by formal-sector programs (albeit only rural households in China) have the option of enrolling in a separate subsidized public health insurance program. In both countries, the contribution is to some degree linked to household income, with poor households having their contribution paid entirely by the taxpayer, and non-poor households either paying a subsidized flat-rate contribution (the case in China) or an income-related contribution (the case in Mexico).1 Thailand recently opted for a third route, which was to enroll at the taxpayer's expense all those not covered by the various programs for formal-sector workers.2 This paper reports the results of an impact evaluation of China's scheme. The program, which began in 2003 and is being rolled out on a staggered basis with all rural county-level jurisdictions (hereafter counties3) to be covered by 2008, replaces China's old village-based rural health insurance program, known as the cooperative medical system or CMS.4 That scheme all but disappeared following the collapse of the commune system in the early 1980 s when China embarked on its market-oriented economic reforms.5 As of September 2006, an estimated 406 million people were enrolled the new scheme, which was up and running in over half (1433) of China's rural counties. The establishment of the new CMS or NCMS, as the new program is known, was a response to accumulating evidence that high and rapidly rising user charges were causing widespread poverty and deterring families—especially poor ones—from using health facilities.6 The program—which unlike its predecessor operates at county rather than village level, and exhibits variations in design and implementation across counties—is financed in part through flat-rate household contributions (the poor and certain other groups have their contributions subsidized) and in part through government subsidies, with central government helping county governments in China's poorer provinces with the local government contribution. One concern with the program is that its budget is too small to make a significant dent in households’ out-of-pocket spending. The revenue per enrolled is around only one-fifth of total per capita rural health spending, and copayments in the scheme are high, reflecting large deductibles, low ceilings, and high coinsurance rates. It is, in fact, possible that because the scheme is likely to encourage people to seek care who would not otherwise have done so, and because providers in China are paid fee-for-service through a price schedule that results in higher margins on drugs and high-tech care than on ‘basic’ services (Liu et al., 1999), insurance may result in people getting more expensive care, and this—together with any impact on utilization rates—may result in increased levels of out-of-pocket spending; this appears to have happened in China's urban scheme ( Wagstaff and Lindelow, 2008a). Concerns have also been expressed that the scheme may do little to increase utilization of health services among poor households because of the high copayments. Indeed, it has been suggested that these costs may reduce the benefits of the scheme to the poor to such a degree that they may be less likely to enroll. Concerns have also been expressed that the scheme may not attract the relatively good risks, and may therefore suffer from adverse selection. This paper attempts to shed light on these and other issues, and in the process to contribute to the more general literature on the impacts of subsidized health insurance programs aimed at informal-sector workers.7 Our focus is on the 189 counties that began implementing NCMS in 2003. We look not only at the impacts on a large sample of households in ten of these counties, but also at the impacts on township health centers and county hospitals in all 189 counties. We also investigate the issue of how the characteristics of different NCMS schemes—their generosity and which services are reimbursable—affect their impact. The paper is organized as follows. Section 2 provides a brief description of the NCMS. Section 3 outlines our methods. Section 4 presents our data. Section 5 presents the results of the matching exercise and shows how far we are able to reduce biases due to differences in observables. Section 6 presents our estimates of the program's impacts, and the final Section 7 contains a summary and discussion.

Our results suggest that, despite its relatively short life and limited financing, the NCMS has had appreciable impacts. It has increased the likelihood of people seeking outpatient and inpatient care, as well as the volume of care provided. Partly because of this increase in utilization, household out-of-pocket spending on health care does not appear to have been reduced by NCMS. There is an additional reason, however, namely that while the cost of at least one relatively well-defined item of health care (a delivery) appears to have been reduced by NCMS, the cost of a typical outpatient visit does not. In other words, NCMS appears to have resulted in people receiving more expensive health care per visit; this is consistent with the results of Wagstaff and Lindelow (2008a) derived from earlier (and mostly) urban health insurance schemes in China. There are differences, however, across income groups in this study, and across counties. The impact of NCMS on the use of outpatient care has been lower among the poor in terms of use of services at higher level facilities but higher in terms of use at lower level facilities. Partly because of this, the upward pressure on household out-of-pocket spending is much less pronounced among the poor. NCMS impacts also appear to have varied across counties. In one, the utilization effect has been muted and the scheme appears to have put considerable downward pressure on out-of-pocket payments per visit and overall. In others, the utilization effect has been strong and any change in spending per visit has been upwards. In these, out-of-pocket spending appears to have been increased by NCMS. Quite what explains these differences in impact across counties is unclear: it does not appear to be the size of the NCMS budget or the types of services that are reimbursed by the scheme. The results from the supply-side data are broadly consistent with those of the household data, and also show that the NCMS has had impacts on bed-occupancy, staffing, and capital investments, at least among township-level providers. One important difference between the demand- and supply-side estimates, which may be due to the different geographic coverage of the two samples, concerns where the extra utilization is occurring: the household data suggest that county hospitals have seen increased outpatient and inpatient utilization, but the former at least is not corroborated by the facility data. Our finding that NCMS has increased utilization of services is not especially surprising, and is consistent with the previous literature on subsidized health insurance programs and health insurance programs more generally; by contrast, our finding that NCMS has not reduced out-of-pocket spending is somewhat surprising, and is at odds with the same literature.27 What does not typically come through in these other studies—at least in those that control for unobserved heterogeneity—is the possibility that insurance may increase out-of-pocket spending. 28 The reason for the difference seems likely to lie on the supply-side—the fact that in other countries insurance schemes do not provide high-powered incentives at the margin to providers, while in China they are paid by fee-for-service and face a fee schedule that strongly encourages demand shifting to drugs and high-tech care on which the margins are higher ( Liu and Mills, 1999). 29 Seen in light of the broader evidence on the impact of health insurance, what are the policy implications of the findings reported in the paper? In and of itself, the findings that the NCMS has increased utilization and left out-of-pocket payments unchanged tell us little about the welfare implications of the policy change. The aim of health insurance is to reduce risk exposure and to make necessary health care affordable. This is achieved by reducing the direct cost of care to patients, which we would expect to induce greater use of health services. However, theory suggests that the welfare gains in terms of access and risk reduction that come from reducing the cost of care must be weighed against the potential welfare losses that arise from demand- and supply-side moral hazard. While the data used in paper cannot shed light on the extent of unnecessary care resulting from moral hazard, there are reasons for concern in the Chinese context. In 1998–1999, a study conducted in 4 township health centers and 8 village clinics in Wuxi County of Chongqing and Min County of Gansu concluded that less than 2% of drug prescriptions were ‘rational’; in the case of village clinics, only 0.06% of drug prescriptions were reasonable on medical grounds (Zhang et al., 2003). Another study found that 20% of hospital expenditures associated with the treatment of appendicitis and pneumonia were clinically unnecessary (Liu and Mills, 1999). In the case of TB, providers have delivered additional care to that in the free DOTS30 package, because doing so generates additional revenues for them (Zhan et al., 2004). This involved treating patients for longer than the recommended six months, and providing non-standard tests and medicines on top of those in the DOTS package. The fact that our study finds that NCMS has increased stocks of expensive equipment at in central THCs is potentially worrisome in this regard insofar as patients may be getting tests and treatment that are medically unnecessary, or which the THC is insufficiently skilled to deliver. Further research is required to investigate further the issue of whether the extra utilization NCMS has encouraged is medically necessary or not. In comparing the findings to those of other studies, and in thinking about the policy implications of the findings, it is important to keep the limitations of the study in mind. First, given the short life of the program and limitations of the baseline data, we focus on a limited set of outcome variables. Most notably, we do not consider the impact of the NCMS on health outcomes. Second, we have been able to shed only limited light on the question of how the impact of the scheme varies with design and implementation characteristics. This question is obviously of considerable policy interest. However, it could not be answered satisfactorily in the present study due in part to the limited number of counties in the sample, and the wide variation in design and implementation features. Indeed, the policy of ‘letting a thousand flowers bloom’ in the piloting of NCMS has much to commend it in terms of encouraging innovation, but it makes pinpointing the secrets of success very hard. Third, even after matching, there are still likely to be some biases in this study due to a lack of complete balancing on observables. In future work, even greater attention needs to be paid at the design phase to ensure greater similarity between the control and the ‘treatment’ counties. Finally, we must be careful in generalizing for China as a whole from the findings reported in the paper. This is not only because the sample of NCMS counties is not a random sample of NCMS pilots, but also because of non-random program placement. Our estimation method will help remove the influences of unobservables, but some unobserved heterogeneity likely remains, and we may be ascribing to NCMS what are really the effects of unobserved time-varying heterogeneity. As the NCMS is rolled out to other counties, its impacts may be different from those found in this paper. For example, it is possible that the impact on service utilization may be more muted due to weaker implementation and a less responsive supply-side. For the moment, though, the scheme appears to have achieved one goal (increased utilization of care) but has done less well on the other (greater financial protection)—the limited success on the latter owes much to the success of the scheme on the former.