Descriptive analysis of the variables
STATA 15 was the main tool used for data analysis in this paper. The descriptive analysis of the variables is shown in Table 1. A total of 12.62% of the respondents with URBMI were hospitalized, while 11.68% of the respondents with NRCMI were hospitalized. The average reimbursement amount of URBMI was much higher than that of NRCMI. In the last two weeks, 28.81% of the respondents with URBMI were sick, while the percentage for those with NRCMI was 29.36%; in the last half year, 19.55% of the respondents with URBMI were sick, while the percentage for those with NRCMI was 16.17%. A total of 85.96% of the respondents with URBMI rated their health status as "Fair, good, excellent or perfect", while the percentage of the respondents with NRCMI who did was 83.56%. In addition, 69.75% of the respondents with URBMI and 59.82% of the respondents with NRCMI can be reimbursed by their medical insurance. The average hospitalization expenses of respondents with URBMI totalled 11,882.75 yuan, while the average hospitalization expenses of respondents with NRCMI totalled 17,272.32 yuan. More respondents with URBMI than with NRCMI chose to receive health care services in general hospitals or specialist hospitals. In addition, concerning the control variables, respondents with URBMI and NRCMI had the same characteristics in the following respects: the average age for both groups was approximately 46 years old; there were proportionally more males than females in both samples; and over 70% of respondents were married in both groups. The average years of schooling of respondents with URBMI was 8.50 years, 1.97 years higher than that of respondents with NRCMI. Compared with that of respondents with NRCMI, the family size of respondents with URBMI was smaller.
Descriptive analysis
Table 2 shows a significant difference in medical service utilization among different income groups. First, in terms of respondents with URBMI, total hospitalization expenses and income were positively correlated, with respondents in the lowest income group spending an average of 14,188.38 CNY/year and those in the highest income group spending 22,722.83 CNY/year. The same conclusion holds for NRCMI, with the lowest income group spending 10,335.58 CNY/year and the highest income group spending 12,637.06 CNY/year. Second, medical insurance reimbursement was positively correlated with both income and total hospitalization expenses. The reason for this may be that groups with higher income utilize more and higher-quality health care. Third, in both the URBMI and NRCMI samples, the medical insurance applicability of the lower income groups was worse, which may increase the inequity of medical insurance benefits.
If the positive correlations among medical insurance reimbursement, health care utilization and income are because people with higher incomes are unhealthier, then the conclusion that there is inequity in medical insurance benefits cannot be drawn. Table 3 shows the health status of the insured populations at different income levels; sick in the last 2 weeks, sick in the last half year and SRH are used to measure health status. In the sample of respondents with URBMI, from the lowest income group to the highest income group, the 2-week sickness rate dropped from 24% to 19%, and the half-year sickness rate dropped from 43% to 26%. The percentage of people with poor SRH dropped from 26% to 10%. In the sample of respondents with NRCMI, from the lowest income group to the highest income group, the 2-week sickness rate dropped from 20% to 13%, and the half-year sickness rate dropped from 36% to 25%. The percentage of people with poor SRH dropped from 25% to 10%. People with higher incomes are therefore healthier than their lower-income counterparts. Therefore, the assumption that high-income people receive more medical services due to poor health is not confirmed. Of course, to scientifically verify this conclusion, an empirical test is still needed.
Empirical test
- Test of the equity of medical insurance benefits
Table 4 reports the differences in medical insurance reimbursement for different income groups. Among respondents with URBMI, we found that there was no significant difference between the lowest income group and the second- and third-lowest income groups; however, the reimbursement rates in the fourth-lowest income group and the highest group are approximately 8.95% and 12.7% higher than that in the lowest group, respectively. Among respondents with NRCMI, the reimbursement rates of the second-, third-, and fourth-lowest income groups and the highest-income group were approximately 3.12%, 3.77%, 5.87% and 5.98% higher than that of the lowest income group, respectively. We also compared the differences between URBMI and NRCMI and found that the benefits equity of URBMI is better than that of NRCMI. The reason for this may be that the income gap in rural areas is wider than that in urban areas.
The descriptive analysis previously revealed that the reason that higher-income people are reimburse more by their medical insurance is not because people in this group are unhealthier, and the probit model will be used to ensure the robustness of this conclusion in this section (see Table 5).
Table 5 shows that among respondents with URBMI, the incidence of being sick in the last two weeks is significantly negatively correlated with income, but there is no significant correlation between the incidence of being sick in the last half year and income. Compared with the lowest income group, the highest-income group and the 4th-lowest income group have better health. Among respondents with NRCMI, the incidence of being sick in the last two weeks or half year and SRH were all significantly correlated with income. There was a significant positive correlation between income and health. In fact, according to the definition of equity, respondents with poorer health deserve more medical insurance compensation; therefore, the results indicate that equalized fundraising and reimbursement cannot guarantee the equity of medical insurance benefits and may even deepen health inequities.
Among respondents with URBMI, the lowest income group differed significantly from the second-lowest income group in terms of medical institution choice, but among respondents with NRCMI, this difference was not significant, indicating that more urban residents than rural residents can utilize more expensive and better health care, which leads to more inequitable benefits among respondents with NRCMI.
Mechanisms
This section analyses the potential mechanism of inequity in basic medical insurance benefits from three perspectives, and the regression results are shown in Table 6. The highest income group and 4th-lowest income group tended to be hospitalized in general or specialist hospitals at a higher rate than other income groups, and their total hospitalization expenses were significantly higher than those in other groups, indicating differences in health care utilization. In addition, from the perspective of medical insurance applicability, respondents with higher incomes had better medical insurance applicability; although the low-income group is covered by a medical insurance system, they are more likely to receive no reimbursement for their expenses. The reason for this may be that their total expenses do not reach the minimum level required.