Benefit equity of social health insurance in China and its provinces (2014–2020): implications for universal health coverage

Data source

We used data from the nationally representative CFPS,28 which uses a Probability Proportional to Size method to collect data biennially at individual, household and community levels on society, economy, demographics, education and health.29 30 The CFPS oversamples the population in five provinces, each from a geographical region of the country: Liaoning (Northern), Shanghai (Eastern), Henan (Central), Guangdong (Southern) and Gansu (Western), so the subsamples for these five provinces were representative at the provincial level.31 Our analysis used adult data from 2014, 2016, 2018 and 2020.

Study variables

Statistical analysis

Descriptive statistics

Table 1 presents the characteristics of study participants over the four survey waves. Average SHI benefit per person increased from 806 CNY, about 124 US$, in 2014, to 934 CNY in 2016 and 1052 CNY in 2018, with a slight decrease in 2020 (906 CNY, 139 US$). Most participants (over 70%) were covered by the resident scheme, which meant either NCMS or URBMI until 2016, or URRBMI after 2016. Between 12% (in 2014) and 15% (in 2020) were covered by UEBMI. The rest were covered by the alternatives of insurance (government or supplementary) or were uninsured. More detailed analysis showed that the distribution of insurance types and other variables was generally stable across the waves nationally and in the five provinces (online supplemental appendix tables 3–6).

Of note, people from the lower consumption quintiles tended to be more likely to be beneficiaries of the URRBMI (online supplemental appendix table 7), which means narrowing gaps between the benefits of resident and employee schemes benefited relatively more those with a lower level of consumption.

Distribution of SHI benefit at the national level from 2014 to 2020

The value of SHI benefit varied by consumption quintile. The CI for SHI benefit improved from 0.262 (p<0.001) to 0.133 (p<0.001). This improvement was further confirmed by the results of the dominance test (table 2 and figure 1). Dominance tests showed that the concentration curves were consistently positioned below the line of equality, indicating that the equity of SHI benefit remained pro-rich during this period.

Request permission

Allocation of SHI benefit according to health need

Figure 2 compares the distribution of SHI benefit by consumption quintile, with health need represented by self-assessed existence of ill health. Across the four waves, the richest quintile experienced around 15% (range 15.18%–15.78%) of the total prevalence of ill health, yet their share of SHI benefit was above 29% (range 29.54%–38.76%). Meanwhile, the poorest quintile experienced over 25% (range 25.07%–26.56%) of the total prevalence of ill health and their share of SHI benefit was small (range 8.52%–15.93%). Of note, while the distribution of ill health was essentially unchanged over the four waves, there was an evident shift of SHI benefit from the richest quintiles (down by 9.22 percentage points) to the poorest two quintiles (up by 7.41 percentage points) from 2014 to 2020. Such a shift of benefit to the poorer quintiles, however, was not obvious when using proportion of people who reported having a chronic disease to represent health need (online supplemental appendix figures 1 and 2).

Request permission

Heterogeneity in benefit equity at sub-national levels

Subnational analysis showed both similarities and differences (table 3). In 2014, all five provinces were better than the national average for CI of SHI benefit (0.262), with CI values ranging from the lowest (Shanghai) at 0.116 to the highest (Henan) at 0.254. In 2016, the CI values in Shanghai and Guangdong were better than the national average (0.281). Additionally, by 2018, the CI of Henan had also experienced an improvement (CI=0.165), placing it below the national level (0.184). In 2020, Gansu was the only province with a CI value (0.172) above the national average (0.133), and the only province with pro-richness that passed the dominance test. The values of concentration indices for Shanghai, Henan and Guangdong were the closest to 0 (CI 0.063 to 0.013, and 0.077) among the five provinces.

Comparing the changes across provinces, Shanghai and Guangdong, where the gap in per-enrollee expenditures between the employee and resident schemes was the narrowest (see online supplemental appendix table 1), was also generally more equitable in SHI benefit between 2014 and 2020. The CI of Shanghai steadily declined from 2014 to 2018, although dominance tests indicated that the change was not significant. For Guangdong, dominance tests indicated that the improvement of the SHI benefit distribution from 2016 to 2020 was significant. Notably, Henan showed significant improvement during this period, transitioning from a relatively high CI value of 0.254 in 2014 to the lowest CI value, of 0.013, among the five provinces in 2020. CI values of Liaoning and Gansu, where the gaps between per-enrollee expenditures were the biggest, exhibited fluctuations between 2014 and 2018, before reducing in 2020, though the changes did not pass the dominance test (table 3). Figure 3 plots the concentration curves of all provinces, by survey years. It shows that the overall trend was a gradual approach towards the line of equity.

Request permission

Decomposition of change in the inequity of SHI benefit

The CI of SHI benefit decreased from 0.262 in 2014 to 0.133 in 2020. Table 4 shows the decomposition results of change in benefit inequity. Consumption was not the main variable of interest here, since it was the ranking variable. The largest contributor to reduced inequity of SHI benefit was membership of the resident scheme as compared with the employee scheme (accounting for 44% of the total reduction). The change was not driven by change in the CI of insurance status, but rather by reduced benefit gaps between resident and employee schemes, as reflected in change in the elasticity of insurance status against benefit in comparison with employee scheme members. The second largest contributor to reduced inequity of SHI benefit was self-reported use of hospitals (as compared with clinics) as the usual source of care (15%). This was mainly driven by changes in the CI of relying on hospitals as usual source of care, as reflected in the term ∆Cη (which amounted to −0.012, much larger than the term ∆ηC at −0.007). This was generally consistent with the results of the annual CI decomposition (online supplemental appendix tables 8–11).

The top three contributors to increased inequity from 2014 to 2020 were age, higher education status and the status of being uninsured. The contribution of age (−14%) was mainly driven by older members being increasingly likely to receive SHI benefit (∆ηC=0.016). Second, the status of college degree or above also contributed substantially to increased inequity (−18%). This was largely driven by the more educated being increasingly likely to receive SHI benefit (∆ηC= 0.023). Being uninsured (as compared with employee scheme enrolment) was the third largest contributor to increased inequity in benefit (−11%). This was driven by people with lower income being increasingly likely to be uninsured (as compared with those enrolled in the employee scheme), reflecting both changes in the distribution of uninsured status and that lower benefits were likely to be received by the uninsured. For the decomposition results of change in benefit inequity by province, see online supplemental appendix tables 12–16.

Strengths and limitations

Our study had several limitations. First, our calculation of SHI benefit used self-reported total healthcare expenditure and out-of-pocket expenditure per year, which relied on a long recall period that may exacerbate recall bias. The conventional approach to benefit incidence analysis estimates benefits using unit costs calculated by dividing aggregate expenditure by the weighted sum of utilisation reported in the survey data, multiplied by the number of visits or admissions. However, because the less well-off tend to use lower cost services, even healthcare expenditures on a visit or hospitalisation at a similar facility can vary substantially,41 so the use of unit costs might mask the likely lower overall gaps in both expenditure and benefit allocated.35 Such systematic bias is less likely using our approach. Second, caution is needed in interpreting the results as we used consumption as measure of living standards. The expansion of SHI benefits may boost consumption by reducing the need for individuals to save for future healthcare needs. Resident scheme members, who were generally poorer, experienced greater benefit expansion relative to employee scheme members, so our analysis might underestimate benefits to poorer groups, as some of them would have moved up the consumption gradient. Third, missing data for explanatory variables and healthcare expenditure meant the exclusion of a portion of the original sample during data analysis. This exclusion may have introduced selection bias and impacted the generalisability of our findings to the entire population. Finally, the dataset used did not include detailed information on healthcare utilisation, such as the different types of healthcare providers and whether reimbursement was for outpatient or inpatient care.27 To address this limitation, we used usual source of care to assess the potential effects of provider choice on benefit concentration indices. According to data from the China Health Statistical Yearbook 2018,42 the actual proportion of care-seeking at hospitals was 45.84%, which is relatively close to the proportion of patients who self-reported use of hospitals as usual source of care.

Interpretation in relation to previous studies

First, our finding that the CI of SHI benefit reduced particularly after 2016, along with the essentially stable distribution of ill health, suggests that consumption-related distribution of SHI benefit became more equitable nationally and between and within its provinces in general. This finding is consistent with previous national and international studies regarding expansion of SHI or harmonising benefit policies between rural and urban populations. Within China, integrating URBMI and NCMS has been shown to facilitate access to health services and improve equity,43 particularly for the poor.44 Under Indonesia’s expansion of health insurance, those with lower income received a bigger share of public healthcare benefits as compared with the well off.45 In India, comprehensive healthcare-related strategies since 2008 have significantly alleviated the financial burden on the poor and advanced equity.46

Second, we have also provided novel evidence on key factors driving changes in SHI benefit equity. We have shown that the ratio of per-enrollee expenditures between employee and resident schemes tended to reduce (online supplemental appendix table 1). As the resident scheme beneficiaries tended to be poorer than employee scheme beneficiaries (online supplemental appendix table 7), the growth of resident scheme benefit reduced the relative gap between employee and resident schemes and contributed to the decline of overall consumption-related inequity of SHI benefits. This trend is further illuminated by the decomposition of the CI of SHI benefits, which underscores that the increase in benefits of the resident scheme was the main contributor to the improvement of SHI equity.

Another main contributor to improved equity was that the poor became more likely to use hospital services.43 44 In other words, the recent improvement in SHI benefit equity was driven by an increasingly hospital-centric model of care, which was confirmed by a noticeable increasing trend in the utilisation of hospitals as compared with primary care providers.22 Similar to China, several countries such as Vietnam47 and India48 have developed SHI schemes that prioritise hospital care. This is understandably given that hospital care tends to be expensive. However, while such policies are likely to contribute to enhanced benefit equity, they may increase demand for hospital services and associated out-of-pocket payment49 50 when primary or ambulatory care could have been more cost-effective. For example, in Vietnam, 73% of total health expenditure occurs in hospitals, current health services are heavily reliant on hospital-based care and out-of-pocket payments significantly exceed the global average.51 Our findings and relevant evidence from other LMICs warn that hospital-centric models of SHI benefits may not be efficient or sustainable.52

Third, we observed large subnational differences at any one-time regarding equity of SHI benefits. This finding is consistent with a previous study that revealed subnational disparity in incidence of catastrophic health expenditure and medical impoverishment.53 Among the five provinces we studied, the difference between employee and resident schemes in Shanghai had always been relatively small, corresponding to relatively equitable allocation of SHI benefits there. Meanwhile, gaps between schemes were continuously larger in Gansu (Western) and Liaoning (Northern), so the distributional equity of SHI benefits of these two provinces continuously lagged behind those of Guangdong (southern) and Shanghai (Eastern), corresponding to the relative lag of Gansu and Liaoning in terms of economic development. Noticeably, Henan (Central), with a large proportion of people outside the formal employment sector, achieved the greater improvement with a narrower benefit gap between schemes, despite its lag in economic development relative to Liaoning. Overall, subnational heterogeneity of benefit equity suggests the importance of paying attention to the western and northern provinces and, once again, supports the case for narrowing the benefit gap between schemes.

Morever, our study revealed important remaining inequities in the distribution of SHI benefits, which adds weight to previous findings that public health benefits remained pro-rich, highlighting that increased financial inputs to SHI alone are insufficient to ensure equitable distribution of healthcare benefit.14 54 On one hand, the fact that poorer members have greater perceived health problems but less SHI benefit requires attention. This situation has persisted despite China’s high health insurance coverage and rising premium levels. Drawing from the ‘Inverse Care Law’ proposed by Julian Tudor Hart in 1971,55 we may dub this phenomenon the ‘Inverse Benefit Law’.

While China has notably propelled the UHC Service Coverage Index to a 79-point score by 2019, the country’s incidence of catastrophic health expenditure is higher than the global average, with a concerning upward trend, contrasting with countries like Malaysia in Asia that have either maintained low or lowered their incidence of catastrophic health expenditure.56 Our results suggest that seeking to enhance benefit equity through greater government subsidy without changing the hospital-centric service delivery model is unlikely to be sustainable. Indeed, while increased subsidy made benefit allocation more equitable, poorer members would be exposed to greater out-of-pocket payments for direct medical costs to receive such benefit, in addition to indirect and non-medical costs, as well as an increased premium that led some to either remain uninsured or drop out from enrolment.57 Moreover, incentivising hospital use is not likely to mean good value for money or represent sustainable health financing.21 58

Implications for policymaking

We draw several policy implications from this study. First, it is necessary to further harmonise employee and resident schemes, as the gap between them remains substantial. Second, the central government needs to strengthen its support to local governments to increase the benefits for resident insurance relative to employee insurance in areas with greater resource constraints, which also see persistent higher inequity in SHI benefits, despite following the same national policy framework. For areas with a high proportion of the population in employee insurance, efforts should be made to investigate potential inefficient or even wasteful use of healthcare resources, in order to generate the fiscal space to expand benefits for the less well off. Third, China cannot rely predominantly on continuously increasing fund allocations to improve equity without adjusting the service delivery model, particularly given slower economic growth. While hospital expenditures need to be contained, concurrent development in service delivery to reduce reliance on the hospital-centric model of care is critical. This will require joint efforts by the National Health Security Administration and National Health Commission to leverage financing reform towards strengthening primary care. Fourth, there is a pressing need for risk-equalisation in the allocation of SHI funds, to address the different levels of health need between richer and poorer members.

Implications for future research

Future studies on benefit equity would be improved by linking claims data to household survey data. In addition, as the provinces, we studied exhibited different rates of change in equity, it will be important to study additional provinces and delve deeper into the underlying causes.