2.1. Impacts of the Health Insurance System on Individual Health
Previous studies showed that health insurance affects individual health status both directly and indirectly. Directly, the financial support provided by health insurance may affect an individual’s psychological stress level [
2], which directly affects the individual’s health status. There are three main ways in which health insurance may affect health status indirectly. First, because health insurance provides access to medical care, regular physical examinations, prophylactic treatment, and higher-quality health services, it may contribute positively to an individual’s health [
3,
4,
5]. Second, health insurance may affect health status by influencing personal behaviors. On the one hand, because being enrolled in a program reduces personal expense for medical services, people may actually continue engaging in unhealthy behaviors such as smoking or drinking [
6,
7]. On the other hand, since people who are enrolled can enjoy more prophylactic health care, they may reduce or quit such harmful behaviors [
8]. Finally, health insurance may indirectly affect health status by reducing the uncertainty one feels about future expenses for medical services, thus affecting savings in medical care [
9,
10].
Many previous studies also focused on correlations between health indicators in people in different age groups from different regions and the insurance programs they enroll in. Most of these studies found that participation in health insurance was significantly associated with lower mortality (or disease risk) and higher self-rated health [
11,
12]. At the same time, however, some studies indicated no significant correlation between them [
13]. Although most of these studies used multivariate regression models to estimate correlations, they had little or no consideration for the interactive relationship between health care and health. They also did not find a causal relationship between the two. Thus, the significance of these studies is limited.
Studies of the causal relationship between health insurance and health are mainly conducted by means of natural experiments. A considerable amount of such literature is focused on analyzing the US health insurance system. Medicare is free health insurance provided by the US government to citizens over the age of 65. Finkelstein and Mcknight (2008) [
14] found that Medicare did not significantly reduce mortality in people over 65 during the first 10 years of enrollment. Card, Dobkin, and Maestas (2009) [
15] used a regression discontinuity design to study the effects of Medicare on mortality of inpatients over age 65 in the emergency department within seven days. They found that Medicare reduced mortality by 20%.
Medicaid is the general term for health insurance provided by the US government to low-income families. The forms of Medicaid vary in different states, but it is usually provided by the government for free. In 1982, the government of California passed a bill that narrowed the range of application of Medicaid in order to cut down on public expense. Lurie et al. (1986) [
16] took this as an opportunity for a natural experiment to compare the health status of two groups of people one year after the bill took effect. One group of people remained enrolled in the program and the other group ceased to be enrolled. The results showed that the average blood pressure of the Medicaid group was higher, but there was no significant difference between the two groups regarding self-reported health status. Currie and Gruber (1996) [
17] studied Medicaid as its range of application expanded from 1984 to 1992, using the proportion of children enrolled in each state as the instrumental variable (IV). The results showed that child mortality was reduced by 5.1%. Another study conducted by Goodman showed that the expansion of Medicaid reduced infant mortality and the occurrence of low birth weight [
18].
Finkelstein et al. (2012) [
4] studied the impact of the Oregon Health Plan (OHP) Standard on the health status of individuals ages 19 to 64. OHP Standard is a form of Medicaid in Oregon that is provided free of charge to low-income citizens of the state. Because the number of applicants exceeds what the project budget can afford, the state government determines who is ultimately eligible to join the program by random selection. Utilizing this special enrollment policy, the study used two-stage least squares (2SLS) to estimate the impact of Standard OHP on health status. They defined the random number assigned by the government as the IV. The results showed that OHP Standard significantly improved self-reported physical and mental health status.
In addition to research in the United States, Hanratty (1996) [
19] studied the impact of Canadian universal health care on infant health. From 1962 to 1972, Canada gradually implemented its universal health insurance system nationwide. Hanratty took this gradual process of implementation as an opportunity for a natural experiment and found that universal health insurance reduced infant mortality by 4% and low birth weight by 1.3%. King et al. (2009) [
20] studied the impact of universal health insurance in Mexico on the health of the entire population. Aggarwal (2010) [
21] used the trend score matching method to examine community-based health care programs implemented in the Yeshavini area of India and found that community health insurance positively affected the health of the enrolled residents by increasing utilization of public medical resources.
2.2. Impacts of Health Insurance System on Individual Health in China
Regarding the impacts of the health insurance system in China, several scholars took advantage of the fact that the NRCMS was gradually implemented in different regions to study its impact on the health of rural residents. Two studies [
22,
23] found that the effect was not significant on the health of rural residents, but Jiang, You, Li, Wei, and Mainstone (2018) [
24] found new evidence that NRCMS improved the health status of rural residents. Bai and Wu (2014) [
25] analyzed the impact of NRCMS on the nutritional status of the insured elderly based on CLHLS data. Although NRCMS did not significantly improve the dietary diversity of the elderly, researchers found that it did promote the consumption of meat and fish and played a role in the balance of diet and improving the health status of the elders enrolled. The effect was more obvious among men and poor elders. Xie et al. (2018) [
26] used panel data of 2000 and 2014 in Jiangxi Province to empirically analyze the impact of NRCMS on the health status of the rural middle-aged and elderly population. The results showed that the implementation of NRCMS had an obvious impact on hospitalization expenses. Wang, Yip, Zhang, and Hsiao (2010) [
27] adopted the experimental research method to compare areas where NRCMS had already been implemented with areas where it was yet to be implemented, using random sampling. They found that NRCMS significantly improved the health status of rural residents. This included reducing the rate of self-reported pain and anxiety among rural residents and improving the operational ability of individuals over 55 years of age.
Some studies focused on the impact of BMISUR on the health of urban residents. Luo (2008) [
28] used data of the 2012 Chinese Household Income Project (CHIP) survey to study factors influencing the health status of urban residents and patterns of their medical expenses. The results showed that implementation of the insurance system had a significantly negative impact on the health status of urban residents and no significant influence on their self-reported health. Lou ascribed this result to endogenous independent variables. Hu and Liu (2012) [
29] utilized propensity score matching and difference-in-difference methods to evaluate the effect of BMISUR. They pointed out that BMISUR improved the health status of low-income sub-healthy urban residents. Pan et al. (2013) [
30] also analyzed the impact of BMISUR on the health status of urban residents. They found that participating in BMISUR improved the utilization of urban medical resources and did not cause a financial burden for urban residents. It had a greater impact on people of lower socioeconomic status.
Some studies have focused on the impact of BMISUE on the health of urban residents. Chen and Deng (2016) [
31] used data from the 2009 China Health Nutrition Survey (CHNS) to empirically analyze the impact of BMISUE on both the short-term and long-term health status of individuals enrolled. The results showed that BMISUE improved the short-term health status of participants to a certain extent while significantly improving their long-term health. Specifically, it reduced the prevalence of cardiovascular and cerebrovascular diseases. At the same time, participating in BMISUE increased the actual medical expenses of the individuals enrolled.
Other studies have paid particular attention to the impact of the health insurance system on the urban elderly and children. For example, two studies [
32,
33] focused on the impact of health insurance on mortality among urban elderly. Huang and Wu (2009) [
33] identified the average enrolled rate by province, gender, age, and education as the IV, and found that enrolling in a health insurance program reduced the three-year interval mortality rate among the elderly by 25.3%. Under the assumption that insurance status is an exogenous variable, Huang and Gan (2010) [
32] found that the risk of death for the elderly enrolled in an insurance system was reduced by 19% compared with those who did not enroll. B. Li and Hu (2010) [
34] studied the impact of health insurance on the health of the elderly, using the average community enrollment rate as the IV. They did not find the impact to be significant. Liu, Meng, and Han (2016) [
35] used data from the 2006, 2009, and 2011 CHNSs to evaluate the impact of health insurance on children’s health status using the two-stage least squares method. The results showed that health insurance could improve children’s health. Mou and Zhou (2017) [
36] used the same data and applied indicators such as health status in the past four weeks, self-rated health status, and body mass index (BMI) to study the impact of NRCMS and BMISUR on children. They found that neither NRCMS nor BMISUR significantly improved the short-term health of children, but NRCUMS significantly reduced the risk of obesity among children. Li and Fang (2018) [
37] used data of the 2012 and 2014 China Family Panel Surveys (CFPSs) to analyze the impact of health insurance on children’s health and found that social health insurance significantly improved the self-rated health of children. Enrollment in an insurance program also increased the utilization of medical resources, measured by the number of clinical visits in the past month.
Based on the existing literature, it can be seen that scholars widely agree that the fragmentation in social health insurance schemes in China has produced variations between different groups in terms of utilizing medical services and health status [
1]. There are three problems related to the impact of China’s medical insurance system on residents’ health. First, because this insurance program is aimed at a specific group of people, most of the existing research was focused on the impact of a certain insurance program on the health status of its target population. These studies did not give specific attention to the possibly different impacts of different programs. For example, there is no research specifically dedicated to analyzing whether there is a difference in impact of the two main insurance programs in China (BMISUE and BMISURR). Second, the current Chinese health insurance system has not achieved central planning. In some regions, the basic pooling unit is prefecture-level cities and in other regions it is county-level cities. At present, China has 333 prefecture-level administrative divisions (excluding Hong Kong, Macao, and Taiwan) and 2856 county-level administrative divisions (excluding Hong Kong, Macao, and Taiwan). When the place where people enroll in a program is different from the place where they seek medical help, the rates of reimbursement may also be different, thus affecting the utilization of medical resources and the health status of individuals enrolled. Third, in the studies of the health status of specific groups of people, such as the elderly population and children, no research has been conducted to analyze the impact of the health insurance system on the health status of the floating population.
This article has focused on senior floating population based on the three following reasons. First, because of rapid urbanization in recent years, the size of the floating population in China has grown at an alarming rate. According to data released by the Chinese government, the total number of migrants in China continually increased from 230 million in 2011 to 245 million in 2017. The floating population was made up of 17.72% of Chinese in 2017. As a constituent part of China’s floating population, senior floating population’s health at place of settlement and whether migration has affected their health deserves scholarly attention. Second, from a medical or physiological perspective, the senior population has a higher risk of various disease than younger populations, so examining the senior floating population’s health status and its relationship with insurance enrollment may yield more typical results. Third, more importantly, the current Chinese health insurance system offers multiple combinations of insurance programs and place of enrollment, providing an opportunity to compare the effects of different insurance programs on individual health.
In China, insurance programs are designed for specific social groups, such as urban employees, urban residents and rural residents, so that people of each social group can enroll in insurance programs according to their social identity. Yet the senior floating population, as a special social group that moves between different regions and between different occupations, are qualified to enroll in more than one kind of insurance program. This offers us an opportunity to compare the effects of different insurance programs and of the same program in different regions.
Under the current health insurance system, where prefecture-level and county-level cities serve as the basic pooling units, there are five possibilities for the floating population: (1) Enroll in BMISUE at the place of settlement. (2) Enroll in BMISURR at the place of settlement. (3) Enroll in BMISUE at a place other than the place of settlement. (4) Enroll in BMISURR at a place other than the place of settlement. (5) Do not enroll in any insurance program. ‘Place of settlement’ in this article refers to the city to which the senior migrants have immigrated, where they presently live and work.
It should be noted that, as mentioned in the introduction, since 2015, some regions have started to consolidate BMISUR and NRCMS to form the new BMISURR. Thus, the distinction between NRCMS and BMISUR was no longer applicable in 2015 when the data were collected. Therefore, in this paper, when the respondents report themselves as enrolling in BMISUR, NRCMS, or BMISURR, their responses are regarded as enrolling in BMISURR. That is to say, enrollment in BMISURR, as defined in this paper, includes enrollment in BMISUR, NRCMS, or BMISURR. This paper uses 2015 China Migrants Dynamic Survey data collected by the National Health Commission to study the health status of the elderly migrant population and analyzes the impact of place of enrollment within the Chinese health insurance system. The floating population studied in this paper is defined as those whose current place of work and residence is different from their place of residence registration (or “Hukou”). Under the special registration system currently implemented in China, each citizen is registered in the place where he or she was born (this can be a city, a prefecture or a county). When a citizen migrates to other places, he or she can either change their place of residence registration to their new place of living or remain to be registered in their place of birth. However, in practice, because of the complexity and triviality of the bureaucrat procedures, most migrants, especially seniors, will not change their Hukou after migrating. In this study, the “floating population” is defined as a person who lives in a place of residence for one month or longer without a local residence registration.