Variables

The enrollment of the URBMI and the NCMS can be influenced by many factors. Considering data from the CHNS, the research objective and research needs, the study decides whether the individual enrolls in the medical insurance for the unemployed – whether the urban residents who do not enroll in the UEBMI and free health care enroll in the URBMI, whether the rural residents enroll in the NCMS or whether residents enroll in one of these two insurances–––to be the dependent variable. The dependent variable is a binary variable, which means that it has two values, enrolling as 1, not enrolling as 0. There totally 10 992 people are analyzed. The amount of people who enroll in the insurances is 8902, which shows the enrollment rate of the URBMI and the NCMS in 2011 was 81.0%, not a high number. Among them, the amount of urban residents is 3761 and there are 2291 people covered by the URBMI, leading to the enrollment rate of 60.9%; the amount of rural residents is 7231 and there are 6611 people covered by the NCMS, getting the enrollment rate of 91.4%. Generally speaking, the enrollment rates of the URBMI and the NCMS are rather different. Therefore, it is important for improving the basic medical insurance for the unemployed based on the voluntary enrollment to compare the enrollment performances of people enrolling in these two insurances and analyze the main causes of the differences.

The independent variables are classified into three categories: health consciousness and health status, controlled variables, and age. The variables in health consciousness and health status include whether the individual received health care in last four weeks, whether the individual got ill in last four weeks and whether the individual has chronic diseases. The controlled variables includes education, per capita income of their family for a year, whether the individual is on the jobs, living districts, gender, and nationality.

The health consciousness is measured by whether the individual received health care in last four weeks. It is acknowledged that people with higher health consciousness are more likely to avoid risk and choose to enroll in medical insurances. However, in the data of the CHNS, there is no direct indicator for preferences in risk avoiding. Referring to the conventional practices in international empirical literatures of health economics [26], the study uses individual health behaviors to reflect their preferences in risk avoiding, which means that people who want to avoid risk will pay attention to health care and purchase medical insurances. Finally, it chooses whether the individual received health care (such as physical examination, visual examination, blood test, test of hypertension, cancer screening, etc.) in last four weeks as the indicator.

The study regards whether the individual got ill in last four weeks and whether the individual has chronic diseases as the indicators of health status and predicts the expected size of medical needs by actual and hidden health problems. In the field of medical insurance, the adverse selection is shown as people with poor health status are more willing to enroll in the insurances, ie, their enrollment rate is higher. Because it was hard to know their health status and medical needs when respondents chose enrollment performances in the survey of CHNS, the study considers whether the individual got ill in last four weeks as the expected medical needs and whether the individual has chronic diseases as the actual medical needs. As for whether the individual has chronic diseases, according to the questionnaires, if an individual has been diagnosed as any one or more of hypertension, diabetes, myocardial infarction, stroke, neoplasm, fracture and asthma, he will be considered to have chronic diseases.

Among the controlled variables, the income is divided based on the quartile of the per capita income of their family for a year. Since the income of urban residents and of rural residents have a big difference and the individual financing levels of the URBMI and of the NCMS are also different, the study divides the per capita income of their family for a year of urban residents and of rural residents into quartiles and does regression analyses respectively. As for district, the Jiangsu Province, the Shandong Province, the Beijing City and the Shanghai City are classified as the East; the Henan Province, the Hubei Province and the Hunan Province are classified as the Middle; the Guangxi Zhuang Autonomous Region, the Guizhou Province and the Chongqing City are classified as the West; the Liaoning Province and the Heilongjiang Province are classified as the Northeast.

The evaluation method of the variables and descriptive statistical results are shown in Table 1 .

On the basis of characteristics of the URBMI and the NCMS, any urban residents without formal employment, including children, the elderly, and other unemployed urban residents, can enroll in the URBMI, while the rural residents, taking family as a unit, can enroll in the NCMS voluntarily. When doing a further analysis, among the group covered by the URBMI, 26.4% of them are minors; 34.2% of them being older than 55, 72.2% of them having no jobs; and for rural residents, the limits of natural resources and economic conditions restraint their medical needs to some degree. Most of these residents are the vulnerable group and their health should be secured by medical insurances.

In the social medical insurance system, the relative departments of the government will not investigate individuals’ health status like commercial insurance companies. Therefore, the health information of the insured cannot be observed by the insurers, which exacerbates the information asymmetry. Under the policy of voluntary enrollment, if the individuals having different health risks enjoy the same insurance price, when the insurer offers insurances with different security levels, the individuals having higher health risk will choose to enroll in the insurance with higher security level (with low out–of–pocket). Because both the URBMI and the NCMS are designed as enrolling voluntarily and are of single security mode, if there are severe adverse selections in these two medical insurances, the results of the logistic regression analyses will perform as the enrollment rate of the high–risk individuals are higher and whether the individuals enroll in the insurance is related with the level of risk significantly. If age influences enrollment greater than health status, then under the condition of other fixed variables, when put age into the model, the coefficients of health status will decrease much, and the influences of health status on enrollment will not be significant even.

Cross–tabulation analysis of health status and age, and enrollment

To reflect the relationship between whether the individual enrolls in the URBMI as well as the NCMS and health status as well as age, Table 2 shows the cross–tabulation analysis of health consciousness, health status, age and whether the individual enrolls in the medical insurances.

It can be seen from Table 2 that the enrollment rates of people with different health consciousness, different health status and different ages are distinct. As for the URBMI which has a rather low enrollment rate, the enrollment rate of people who got ill in last four weeks, have chronic diseases, received health care in last four weeks or are older than 55 are higher than it of people who did not and are not; as for the NCMS which has a high enrollment rate, the enrollment rate of people with poor health status and it of good health status have little differences, but people older than 55 are more willing to enroll in the insurance. Merging data of the URBMI and the NCMS, the difference is smaller than the data only of the URBMI. What should be paid attention to is that the data in Table 2 also report that whether received health care in last four weeks, as the variables reflecting health consciousness, influences enrollment rate slighter than health status, while age influences enrollment rate greater than health status.

The results of cross–tabulation analysis in Table 2 only explain that there are differences in enrollment rates of people with different health consciousness, health status and ages from the perspective of percentage. It need further analysis by logistic regression model to explore whether there is connection between enrollment and health status, health consciousness and age, how strong the connection is and which variable influences the insurance enrollment greater.

Influences of whether the individual got ill in the last four weeks on insurance enrollment

In the logistic regression model, the likelihood–ratio is applied to test and backward stepwise manner is applied to filter variables. If the variable's significance level is smaller than 0.05, it will be analyzed continually, but if its significance level is larger than 0.10, it will be excluded. The influences of whether the individual got ill in last four weeks on insurance enrollment are shown in Table 3 . Model I only estimates influences of whether the individual got ill in last four weeks and whether the individual received health care on insurance enrollment and broadens the presupposition of the model gradually. Model II adds controlled variables that are whether the individual is on the job, education, income, gender, nationality, district, etc. Model III adds the variable of age on the basis of Model II.

It can be seen from Model I and Model II in Table 3 that as for people insuring the URBMI and merging people insuring the URBMI and the NCMS, not considering other independent variables, whether the individual got ill in last four weeks influences insurance enrollment significantly. In addition, the enrollment rate of people who got ill in last four weeks is higher than it of people who did not get ill in last four weeks, while whether the individual got ill in last four weeks does not influence people just involving in the NCMS much.

The regression results of Model III show that after adding the variable of age, whether the individual got ill in last four weeks, which has influenced insurance enrollment significantly in Model I and Model II, does not influence it much yet. That is to say, age influences individuals on enrollment greater than whether the individual got ill in last four weeks.

Influences of whether the individual has chronic diseases on insurance enrollment

Using the same method, the results of influences of whether the individual has chronic diseases on insurance enrollment are shown in Table 4 . Model I estimates influences of whether the individual has chronic diseases and whether the individual received health care in last four weeks on insurance enrollment. The variables put into Model II and Model III are totally the same as the variables in last part.

Model I and Model II in Table 4 show that whether the individual has chronic diseases influences whether the individual enrolls in the insurances significantly. People with chronic diseases are more willing to enroll in the insurances than people without them.

Similar to the situation happened to whether the individual got ill in last four weeks, when adding the variable of age, the previous significant variable in Model I and Model II – whether the individual has chronic diseases – are not significant anymore. However, regarding merging residents, whether the individual has chronic diseases still has a statistically significant influence at the 10% level. Age influences enrollment greater than the individual’s actual health status (whether the individual has chronic diseases) toward urban unemployed and rural residents.

Summary

Analyzing Table 3 and Table 4 , the results show that the adverse selection exists in the URBMI and the NCMS. Compared with the adverse selection in the NCMS, the adverse selection in the URBMI is more evident but not severe. Whether the individual got ill in last four weeks and whether the individual has chronic diseases influence the enrollment of the URBMI significantly and people having gotten ill in last four weeks or having chronic diseases are more willing to enroll in the insurance; whether the individual got ill in last four weeks has no significant influence on the NCMS, but whether the individual has chronic diseases has, which is performed as the enrollment rate of rural residents who have chronic diseases is higher than those who do not have.

What should be noticed is that when we put the variable of age in the analysis, the influences of whether the individual got ill in last four weeks on the enrollment rates of urban residents, rural residents or urban and rural residents together are not significant, and the influences of whether the individual has chronic diseases on the enrollment rates of urban residents and rural residents are also not significant, but the level of influences on the enrollment rates of urban and rural residents together is still in 10% which is significant. Whether to take actual health status or to take expected health status as the variable indicating health status, the influences of age on insurance enrollment are similar and are greater than the influences of health status. In another word, in the market of the URBMI and the NCMS, there are not severe adverse selection. In addition, the results of Table 3 and Table 4 show that the enrollment rate will increase with age growing.

Above all the samples in this study, 20% of urban samples and 13% of rural samples were students, in addition, some samples were preschool children. For these samples, their enrolment decisions were made by their parents, thus it may have some influences on the final results and conclusion. Although this study cannot reveal the influences of parents on the enrolment willingness of these samples, it verified that age and health status were the main reasons that influence students and preschool children’s enrolment willingness.

Easing adverse selection and changing enrollment performances of residents by inclusive policies

According to the data from the CHNS of 2011, under the policy of voluntary enrollment, the URBMI and the NCMS in China cover a great range of people. Though there are problems of adverse selection in these two basic medical insurance markets, they are not severe. Age influences insurance enrollment greater than health status (whether the individual got ill in last four weeks and whether the individual has chronic diseases). The older the group are, the higher the enrollment rate is. The study considers that it is related to the inclusive design of the system for one thing, and changes of enrollment performances of the urban unemployed and rural residents for another thing.

In the market of the URBMI and the NCMS, the central government offers large subsidies. After the start of the new health–care reform in 2009, governments at all levels committed 850 billion yuan to medical field in the first three years, mainly to support five reforms including construction of basic medical security system. As for the construction of medical security system for the whole population, the government has spent over 680 billion from 2009 to 2012 to subsidy the URBMI and the NCMS. The increase of fiscal subsidy per capita in the URBMI are always larger than the increase of individual premiums, leading to a higher proportion of government subsidies to financing per capita, from 60.8% in 2009 to 79.3% in 2014. The ratio of government subsidies to individual premiums is close to 4:1, which means that the actual individual financial burdens are not heavy. The massive fiscal supports stimulate people’s willingness to enroll in the medical insurances and develop the coverage of the insurances effectively, which may be the reason why there is no severe adverse selection in the market of basic medical insurances in China.

Analyzing Table 1 and Table 2 , at present, the self–care consciousness of the urban unemployed and rural residents is still very low. The percentage of the urban unemployed who received health care in last four weeks is only 9.4%, while it of rural residents is only 5.1%. The percentage of individuals who received health care in last four weeks is similar to the percentage of individuals who enroll in the health care. Because of weak self–care consciousness, when the urban unemployed and rural residents decide whether they insure or not, their health status and medical needs will not influence their decisions. They pay more attention to their ages and potential health risks. In the process of developing basic medical insurances in China, under the background of large financial subsidies from Chinese government and high general enrollment rates of the insurances for the urban unemployed and rural residents, the enrollment performances of them may change. With age growing, the potential health risks and medical needs increase and the urban unemployed and rural residents are more sensitive to age. Therefore, they choose to enroll in the medical insurances.

Optimizing the structure of insuring group and increasing anti–risk ability of medical insurance funds

The results of the study show that although age influences insurance enrollment greater than health status, the enrollment performance that the enrollment rates increase with age growing will go against optimizing the structure of insuring group. With the growing aged tendency of population, the general enrollment rate of Chinese basic medical insurances for the urban unemployed and rural residents can maintain stable or even higher, but it also results in an unreasonable structure of insuring group. If the reimbursement level of these insurances raises continually, the burden of the medical insurance funds must be increased, which brings about a danger to them. To optimize the structure of insuring group, there must more young adults be appealed to enroll in the insurances, and then, the basic medical insurances exert the effect of social assisting and risk sharing under the law of great number.

Learning from the practices in China, the great input of government cannot keep the sustainability of the basic medical insurances. After the new reform of medical system, medical expenditures in our country increase rapidly, causing the growth rate of expenditures of medical insurance funds higher than the growth rate of income of it. In some provinces, the income of their medical insurance funds cannot pay out the cost of the funds and the insurance funds are dangerous. To keep the insurance funds from risks, the Ministry of Human Resources and Social Security of the People’s Republic of China and the Ministry of Finance of the People’s Republic of China has investigated in 2009 and suggested that only the fund can satisfy the financial demand of six to nine month, can the medical insurances be well operated. The data from the Ministry of Human Resources and Social Security of the People’s Republic of China show that compared with it of 2013, the length of payable months of 2014 in 22 provinces become shorter. There are six provinces that the length of payable months is shorter than six months. The medical insurance funds in many regions are in danger and the sustainability of social medical insurances face a great challenge. In addition, the policies of the insurance that integrate s the URBMI and the NCMS for urban unemployed and rural residents will follow the principle: the financing standard should be the lower one; the reimbursement level should be the higher one; the range of pharmaceuticals should be the larger one, which increases the expenditures of the medical insurance funds [27]. Therefore, it should keep the fund resources stable, optimize the financing structure, enlarge the coverage of the insurance, enlarge the medical insurance fund, and increase the anti–risk ability of the fund.