Introduction

Health inequality has long been a core topic of research in health economics. In addition to traditional socio-economic factors such as income, education, and urban-rural divides, the impact of air pollution on health inequality has received growing global attention(Chen et al., 2023; Ebenstein et al., 2017; M. Liu et al., 2017; Lu et al., 2017). According to a 2021 UN report, 92% of the global population lived in areas where PM2.5 levels exceeded the WHO guideline of 10 μg/m3, in 2019. The 2019 Global Burden of Disease report identified air pollution as the world’s fourth leading cause of death, responsible for 1.85 million deaths. In China, rapid economic growth since 1978 has led to serious environmental degradation. In 2020, the national annual average AQI was 80.49 μg/m3, far above WHO standards. Key economic regions such as Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta had even higher levels—up to nearly nine times the WHO recommendation. These figures highlight the urgency of addressing air pollution as both an environmental and a social justice issue.

In 2017, a substantial proportion of age-standardized disability-adjusted life years (DALYs) from chronic diseases—40.0% for lower respiratory tract infections, 35.6% for diabetes, 26.1% for lung cancer, 25.8% for ischemic heart disease, and 19.5% for stroke—were attributed to air pollution (Yin et al., 2020). Environmental health concerns driven by severe air pollution intersect with socio-economic health disparities related to income, education, and registered residence. This issue is particularly salient for migrant populations, whose health is more adversely affected by sustained exposure to polluted environments. Long-term exposure, particularly to nitrogen dioxide (NO2), significantly impacts migrants’ health, increasing risks of hypertension and diabetes(Downward et al., 2022). Vulnerable groups, such as men and individuals over 30, exhibit heightened sensitivity to pollution, which not only compromises their health but also weakens their willingness to settle in polluted areas (Zhao et al., 2021).

In order to address health inequality, China established three medical insurance systems in 1998, 2003, and 2007, covering urban workers, farmers, and other urban residents, achieving near-universal coverage. However, due to regional enrollment restrictions, migrants often register their insurance in their registered permanent residence (RPR), limiting access to medical services in inflow cities. Cross-city medical treatment is constrained by complex referral and reimbursement procedures, resulting in delays and reduced healthcare accessibility for migrants across different medical insurance pooling regions.

Against this backdrop, the Cross-Regional Medical Insurance System (CRMIS) was introduced, allowing the insured to choose enrollment in either their RPR or current location. This concept first appeared in the 1998 State Council Decision, which enabled sectors with the characteristic of cross-regional work—such as railways and maritime transport—to centralize insurance in work locations. In 2010, the Interim Measures for the Transfer and Continuation of Basic Medical Insurance for Migrant Workers permitted rural migrants with stable employment to join urban insurance schemes. Article 32 of the 2011 Social Insurance Law further supported transferring insurance relationships across regions. Following the 2016 integration of urban and rural schemes, CRMIS reforms promoted unified enrollment and enhanced portability (Cheng et al., 2015; Yang, 2007). Nevertheless, health inequality driven by air pollution continues to undermine these achievements, highlighting the need for further system improvements to safeguard health equity.

Literature review and research hypothesis

Impact of air pollution on health and health inequality

There is a large literature on the impact of air pollution on health, with only a few on health inequality. Rising air pollution levels lead to a significant decline in the mental health of the population (Buoli et al., 2018; Yang et al., 2021). Air pollution also leads to a decline in individual subjective well-being (Zhang et al., 2017). Increased air pollution leads to increased mortality among infants, adults, and the elderly (Cesur et al., 2018; Chay & Greenstone, 2003; Heft-Neal et al., 2018). The increase of air pollution will boost the number and prevalence of respiratory diseases (Schlenker & Walker, 2016).

The health damage consequences caused by different air pollution levels vary among different economic status, regions, races, nationalities, genders, and division between urban and rural (Collins et al., 2017; Evans & Kantrowitz, 2002; Gu et al., 2013). First of all, air pollution is obviously pro-poor (Dell et al., 2012; Kumar & Khanna, 2019), and groups with lower socioeconomic status are more likely to be exposed to pollution. In areas with low economic development, the health economic burden of air pollution is heavier (Finkelstein et al., 2005; Grineski et al., 2013). The results of Richardson et al. (2013) showed that mortality from circulatory diseases in low-income areas of Eastern Europe and respiratory mortality in males in low-income areas of Western Europe were more likely to be affected by air pollution. Only in the poorest areas, the reduction of air pollution is likely to reduce the degree of health inequality (Richardson et al., 2011). The health damage caused by air pollution further exacerbates health inequality across income groups and increases the extent to which income inequality affects health inequality (Yang & Liu, 2018). Secondly, the difference in air pollution suffered by racial and ethnic disparities will also lead to health inequalities (Hackbarth et al., 2011). Woo et al. (2019) concluded that racial and ethnic minorities are exposed to significantly higher levels of air pollution compared to whites, and addressing this uneven environmental burden may be a promising way to improve population health and reduce racial and ethnic disparities among them. Thirdly, there are significant urban-rural and gender differences in health inequality caused by air pollution (Azimi et al., 2019). Rural men aged 30–34 years are more likely to die of cancer, while rural women aged 15–19 years have a 47% higher death rate and 44% higher cancer death rate than urban women (Liu et al., 1999).

The limitations of current research and the marginal contributions of this article

To sum up, while existing studies have made significant progress in exploring the relationship between air pollution and health inequality, several limitations remain. Firstly, due to the limitation of research indicators, most studies analyze the health consequences of air pollution from the perspective of the differences between urban-rural areas, regions, and other populations, while there are few research results on the direct effects of air pollution on health inequality. The second is the lack of explanation from the perspective of medical insurance policy, which is hidden behind the health inequality problem. In light of the problems above, this article conducts research. The marginal contributions of this paper are as follows. First, this paper employs the relative deprivation index to construct an individual-level health inequality index, and based on this, the quantitative analysis reveals that air pollution exacerbates health inequality—an increase in air pollution further intensifies disparities in individual health inequality. Second, it examines the effect of the CRMIS targeting migrants in mitigating health inequality induced by air pollution. The findings indicate that for migrants across different medical insurance pooling regions in China, enrolling in medical insurance in the inflow region rather than in their registered residence region can effectively reduce the health inequality caused by air pollution.

Building upon the existing literature and addressing the above-mentioned research gaps, this study attempts to further explore the mechanisms through which air pollution contributes to health inequality. In particular, we draw upon the theory of relative deprivation to provide a more nuanced understanding of how disparities in environmental exposure translate into unequal health outcomes across individuals.

Relative deprivation theory

In the concept of relative deprivation, health inequality means that in a group, the lower the individual’s health level, the higher the relative deprivation degree will be in the accumulation of health disadvantage, and the higher the degree of health inequality will be. The idea behind the concept, which has been used by a wide range of scholars (Smith et al., 2012; Wilkinson, 1997), is that a person’s health or health-related behaviors depend on both his own resources and his relative position with respect to those resources (Caner & Yiğit, 2019). Differentiated air pollution exposure will lead to differences in the relative degree of individual health deprivation. Compared with those at low levels of air pollution (Bell & Ebisu, 2012), individuals at high levels of air pollution will face a higher risk of worse health. When individuals’ access to goods, services, and social activities is limited (Kuo & Chiang, 2013), the relative degree of individual health deprivation caused by air pollution will be deepened; thus air pollution will aggravate the health inequality among individuals (Currie et al., 2009; Currie et al., 2011). Therefore, we put forwarda hypothesis:

H1: The more severe the air pollution, the higher the relative health deprivation of individuals caused by air pollution, and the higher the level of health inequality will be.

The moderating effect of CRMIS

The choice of CRMIS will greatly affect the availability and accessibility of individual health care services. Individuals with local medical insurance are more likely to obtain timely medical intervention to avoid health deprivation caused by air pollution, thus reducing the harm of air pollution on health inequality (Chen et al., 2013; Currie & Neidell, 2005). By increasing the probability of individuals living in air pollution areas accessing health care services, reducing the economic burden of disease, and avoiding the level of health deprivation of migrants, the CRMIS alleviates the harm of air pollution on health inequality (Meng et al., 2018; Onwujekwe et al., 2010). Therefore, this paper puts forward a hypothesis:

H2: The harm of air pollution on the level of health inequality among individuals is regulated by the location of medical insurance enrollment. Compared with the migrants insured in the permanent registered residence, the migrants insured in the inflow region suffer less harm from the relative deprivation of health caused by air pollution, and the level of health inequality is also lower.

Methodology

Material, methods, and data sources

The data come from the China Health and Retirement Longitudinal Survey (CHARLS). The survey is a large-scale interdisciplinary survey project conducted by the National School of Development at Peking University and the China Social Science Survey Center. CHARLS conducted surveys and interviews in 150 counties and 450 communities (villages) in 28 provinces (autonomous regions and municipalities directly under the Central Government) in 2011, 2013, 2015, and 201,8 respectively. By the time the national follow-up was completed in 2018, its sample had covered 19,000 respondents from a total of 12,400 households. Considering that the core explanatory variables (AQI) before 2014 are seriously missing, this paper chooses the data of 2015 and 2018 in CHARLS as samples for research. After deleting the missing values of health inequality and the sample observations under 45 years old, we obtain 32479 valid samples from 7305 households in 28 provinces, and 28752 valid samples after deleting the missing values of main control variables. Finally, after taking the moderating variable of medical insurance participation into account in the measurement model, there are 26,526 valid samples. In addition, the data of AQI, PM10, NO2, and CO were collected from the National Urban Air Quality real-time publishing platform of the China Environmental Monitoring Station and the China Environmental Statistical Yearbook of the National Bureau of Statistics and the Ministry of Environmental Protection of China. PM2.5 in the air pollution data comes from the Atmospheric Composition Analysis Group (ACAG) of Washington University (St. Louis).

Variable measurement

Dependent variable

The dependent variable in this paper is health inequality index. Firstly, the physical health indicator was selected with reference to the disease classification of the CHARLS survey, which measures health from the dimension of physical health, considering both acute and chronic shocks. These physical health indicators are closely related to air pollution, and the more serious the air pollution is, the more serious the harm of diseases such as respiratory diseases, cardiovascular diseases, and liver fibrosis will be (C. Liu et al., 2017; Zheng et al., 2015). Secondly, referring to the construction method of physical health index by Wang and Luo (2020), the entropy weight method is used to calculate the health index (Wang et al., 2025; Zhang & Jiang, 2021), and the index and weights are shown in Table 1. Finally, Kakwani (1984) proposed the concept of relative deprivation to convert physical health into an index of physical health inequality. The Kakwani index satisfies both dimensionality and normalization characteristics, and the weighted average of relative deprivation among all individuals is the Gini coefficient, which has good properties in fitting income, consumption, and health distributions(Fukushige et al., 2012). According to the notion of relative deprivation, in a group, the lower the individual’s health level, the greater the health disadvantage, and the higher the degree of relative health deprivation, which means the higher the level of health inequality. Suppose Y represents a reference group, and the sample size is n. We rank the health level y of individuals, and the overall health distribution is obtained as a vector \(Y=({y}_{1},{y}_{2},{y}_{3},\ldots ,{y}_{n-1},{y}_{n})\), where\({y}_{1}\le {y}_{2}\le {y}_{3},...,\le {y}_{n-1}\le {y}_{n}\). Then, compared resident i with another resident j, the relative deprivation index \({RD}\left({y}_{j},{y}_{i}\right)\) of resident i is expressed as:

$${RD}\left({y}_{j},{y}_{i}\right)=\left\{\begin{array}{ll}{y}_{j}-{y}_{i}\qquad{if}\,{y}_{j} > {y}_{i}\\ 0\qquad\qquad{if}\,{y}_{j}\le {y}_{i}\end{array}\right.$$
(1)
Table 1 Construction of health indicators.
Full size table

On the basis of the formula (1), the average value of \({RD}\left(y,{y}_{i}\right)\) received by resident i can be expressed as:

$${RD}\left({y}_{j},{y}_{i}\right)=\frac{1}{n{\mu }_{Y}}({n}_{{y}_{i}}^{+}\times {\mu }_{{y}_{i}}^{+}-{n}_{{y}_{i}}^{+}\times {x}_{i})$$
(2)

Where \({\mu }_{Y}\) is the average health level in the reference group Y, \({n}_{{y}_{i}}^{+}\) is the number of individuals whose health index exceeds \({y}_{i}\) in the group Y, and \({\mu }_{{y}_{i}}^{+}\) is the average health level of those individuals mentioned above.

Independent variable

The air pollution indicators were measured using the Air Quality Index (AQI) for the year prior to the CHARLS2015 and CHARLS2018 survey, respectively. According to the Ambient Air Quality Standards of the People’s Republic of China (GB3095-2012), the ambient air quality index (AQI) is calculated from the concentrations of six pollutants in the air, namely SO2, NO2, PM10, PM2.5, carbon monoxide, and ozone. The corresponding pollutant air quality sub-index (IAQI) is calculated based on the monitored average concentrations of the six air pollutants over a certain period of time, where the maximum value of the pollutant air quality sub-index (IAQI) is the ambient air quality index (AQI) for that period of time. AQI has good representativeness and applicability (Hu et al., 2015; Lu, 2020). Secondly, in order to verify the robustness of air pollution indicators, we adopt the specific values of PM2.5, PM10, and CO to measure air pollution.

Moderating variable

This paper mainly discusses the CRMIS, and the selected index is the location of enrollment in medical insurance. This index is mainly measured by the question “ place of medical insurance enrollment”, and the main answers are 3 options: the inflow location, the registered residence place, and other places. The insurance in the inflow region means the migrants are insured in their settlement where they work and live, instead of their registered residence places, and there is no overlap with the two kinds of places. Since the proportion of the insured samples in other places is very small, less than 1%, we only discuss the different degrees of the moderating effect caused by the difference of insured place between the inflow region and the registered residence region for the migrants.

Control variable

Referring to the previous studies (Lavaine, 2015; Yang & Liu, 2018), the control variables selected in this study include the characteristics of individual residents’ socioeconomic status and health behavior. The socio-economic status characteristics mainly include gender, age, marital status, education level, registered residence, place of settlement, social medical insurance, commercial medical insurance, household per capita income, and region. Health behavior characteristics mainly include smoking, drinking, and exercise. Table 2 reports the descriptive statistics of main variables. Family income and Ventilation Coefficient(VC) were logarithmized in the regression equation.

Table 2 Statistical description of main variables.
Full size table

Measurement methods

This study aims to explore the impact of air pollution on health inequality, and uses the two-stage least squares (2SLS) method for estimation, which is widely used in the research on air pollution, health and its inequality (Anwar et al., 2021; Kim et al., 2003; Wang et al., 2021). The 2SLS regression model is set as follows:

$${{Pollution}}_{{ij}}={\alpha }_{0}+{\alpha }_{1}{{VC}}_{{ij}}+{\alpha }_{2}{X}_{{ij}}+{\varepsilon }_{{ij}}$$
(3)
$${{Inequality}}_{{ij}}={\beta }_{0}+{\beta }_{1}{{Pollution}}_{{ij}}+{\beta }_{2}{X}_{{ij}}+{\omega }_{{ij}}$$
(4)

Where \({{pollution}}_{{ij}}\) is the AQI and six kinds of air pollution indicators of prefecture-level administrative region j where resident i is located, indicating the degree of air pollution; \({{VC}}_{{ij}}\) is the Ventilation Coefficient of the prefecture-level administrative region j where resident i is located, which is used as the instrumental variable of air pollution in 2SLS regression. \({{Inequlity}}_{{ij}}\) is the degree of health inequality of resident i in area j; \({X}_{{ij}}\) is the control variables, which include a series of factors affecting health inequality at the individual, family, and regional levels; \({\varepsilon }_{{ij}}\) and \({\omega }_{{ij}}\) are the random term.

This study refers to the construction method of \({\rm{VC}}\) (Broner et al., 2012). The formula for calculating \({VC}\) is as follows, where \({WS}\) is the wind speed at a height of 10 meters and \({BLH}\) is the Boundary Layer Height.

$${VC}={WS}\times {BLH}$$
(5)

The \({VC}\) satisfies the constraints of correlation and exclusivity, meets the requirements of instrumental variable method, and can effectively solve the problems of reverse causality, measurement error, and omitted variables. The reasons are as follows: firstly, the wind speed determines the diffusion speed and range of pollutants, while the height of the boundary layer determines the height of diffusion. Therefore, ceteris paribus, regions with larger air mobility coefficients have stronger transmission, diffusion capacity of air pollution,n and lower concentrations of pollutants, thus satisfying the correlation hypothesis. Secondly, the \({VC}\) is determined by meteorological conditions such as wind speed and boundary layer height, and has nothing to do with economic activities such as economic growth, capita,l and technology (Broner et al., 2012), so it will not affect residents’ ability to invest in health. Moreover, the calculation results show that meteorological conditions such as wind speed will not affect the health inequality among residents by affecting daily activities or outdoor sports, so as to satisfy the exclusivity constraint. Data on wind speed and boundary layer height were obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF). In this study, ArcGIS is used to transform and extract global raster meteorological data, and the mean value of \({VC}\) one year before the survey is calculated asan instrumental variable.

Results

Regression results of two-stage least squares method

Table 3 shows the results of the first-stage regression based on two-stage least squares. Firstly, the validity of instrumental variables is examined, and the F-values are all greater than 10 and significant at 1% confidence level, indicating that there is no weak instrumental variable problem. In addition, the air mobility coefficient is significantly negative, indicating that the increase of air mobility coefficient favors the diffusion of air pollutants and reduces the concentration of pollutants, which is negatively correlated with each other. Columns 1 and 2 in Table 4 report the regression results by 2SLS without and with control variable, respectively. To verify the robustness, columns 3 and 4 of Table 4 report the regression results by the ordinary least squares method (OLS) without and with control variable, respectively. From the results, the results of 2SLS are better than those of OLS, and the OLS method underestimates the direct role of air pollution on health inequality. The regression of 2SLS needs to first test the validity of the instrumental variables. In the first-stage regression, the \({VC}\) is significantly negative, indicating that the increase of \({VC}\) is conducive to the diffusion of air pollutants and reduces the pollution concentration, and the two have a negative correlation. In addition, the F values are all greater than 10 and significant at the 1% confidence level, indicating that there is no weak instrumental variable problem. In the second-stage regression, the estimated coefficients of the inequality of AQI on physical health are significantly positive in both columns (1) and (2), indicating that air pollution has caused an increase in the inequality of physical health. Specifically, the regression coefficients in columns (1) and (2) are 0.0030 and 0.0040, respectively. which means that for every 1 unit increase in annual average AQI, the inequality of physical health will increase by 0.0030 and 0.0040 units, respectively. This result shows that air pollution will aggravate the degree of health inequality among individuals, which verifies Hypothesis 1.

Table 3 2SLS (first stage) regression results.
Full size table
Table 4 OLS and 2SLS(second stage) regression results.
Full size table

Robustness test

Replace self-rated health inequality indicators

This paper conducts robustness test from three aspects. Firstly, the health inequality indicator is reconstructed based on self-rated health, which is assigned 5, 4, 3, 2, 1, and 0 points in order of excellent, very good, good, fair, bad, and very bad for robustness test. And then the self-rated health indicator is transformed into health inequality index by relative deprivation index method. According to the results in Column (1) of Table 5, air pollution significantly increases the inequality of self-rated health among residents, and the regression results are still robust after replacing the index.

Table 5 The robustness test of self-rated health inequality index, replacing instrumental variable and using fixed effects models.
Full size table

Replace instrumental variable

We use the “inverse temperature” indicator as an instrumental variable of air pollution for robustness testing. The number of annual occurrences of inversions at the city level in China provided by NASA is calculated. The validity tests for the inverse temperature variable revealed that all F-statistics exceeded 10 and were statistically significant at the 1% level, indicating the absence of a weak instrument problem. Column (2) of Table 5 shows the regression results. It is found that a one-unit increase in AQI significantly increases physiological health inequality by 0.0026 units at the 1% significance level. This indicates that the effect of air pollution on worsening health inequality remains significant when using IV for air pollution, and the conclusions of this paper are robust.

Using fixed effects models

To verify the reliability of the instrumental variable (IV) approach, this study further employs a fixed effects model to analyze the panel data. The model incorporates both individual fixed effects and year fixed effects to identify the dynamic effects of air pollution on health inequality. As shown in Column (3) of Table 5 the direction of the regression coefficient for air pollution remains consistent with that of the IV approach. This suggests that the positive effect of air pollution on health inequality remains robust even after controlling for unobserved fixed differences across regions and time-varying confounders, further supporting the reliability of the study’s conclusions.

Select different pollutant indicators

PM2.5, PM10, and CO were selected to replace AQI as indicators to measure air pollution levels, because these are main air pollutants, which are very harmful to the human body. The results are shown in column 2, column 3 and column 4 of Table 6, respectively. If PM2.5, PM10, and CO increase by one unit, the health inequality of residents will increase by 0.0028, 0.0066, and 0.3603 units, respectively.

Table 6 The robustness test of different pollutant indicators method.
Full size table

Heterogeneity analysis

Heterogeneity analysis of different populations

To further investigate the heterogeneity in the impact of air pollution on health inequality, we introduce interaction terms between AQI and key individual characteristics, including gender, age, and self-rated health status. The heterogeneity analysis is also conducted using the two-stage least squares (2SLS) instrumental variable approach to ensure consistency with the baseline estimation and address potential endogeneity. As shown in Table 7, the interaction between AQI and gender is significantly negative (−0.0076, p < 0.01), indicating that air pollution increases health inequality for both men and women, with a stronger effect observed among women. For age, the sample is divided into a middle-aged group (45–59 years) and an elderly group (60 years and above). The interaction term is also significantly negative (−0.0025, p < 0.01), suggesting that the elderly are more vulnerable to the inequality-increasing effects of air pollution compared to the middle-aged. Regarding self-rated health, individuals with poorer health—classified as those who rated their health as “fair,” “poor,” or “very poor”—are found to be more severely impacted, as indicated by the significant interaction term (−0.0026, p < 0.01). These results collectively demonstrate that air pollution exacerbates health inequality more substantially among women, the elderly, and those with poor health status.

Table 7 Heterogeneity analysis of the impact of air pollution on health inequality among different populations.
Full size table

Analysis of heterogeneity by socioeconomic status

To analyze the heterogeneity of the impact of air pollution on health inequality across different socioeconomic groups, Table 8 presents 2SLS regression results based on registered residence, income level, and educational attainment. In terms of registered residence, the interaction term between AQI and rural registration is significantly negative (−0.0116, p < 0.01), indicating that air pollution significantly increases health inequality among rural residents, while the effect for urban residents is relatively weaker. For income level, using the average annual household income of CNY 39,162 as the threshold, results show that air pollution significantly exacerbates health inequality in the low-income group (−0.0007, p < 0.01), but the effect is not significant for the high-income group. Regarding education, individuals with lower educational attainment experience a greater worsening of health inequality due to air pollution, as shown by a significantly negative interaction coefficient (−0.0003, p < 0.01), whereas the effect is less pronounced for those with higher education levels.

Table 8 Heterogeneity analysis of the impact of air pollution on health inequality among different socioeconomic status populations.
Full size table

Moderating effect

In order to further explore the mechanism of air pollution on health inequality, we use the “place of medical insurance enrollment” as the moderating variable for analysis.

The regression results in columns 1 and 2 of Table 9 show that air pollution worsens health inequality. The impact of the interaction term between air pollution and the insured place on health inequality in column 1 and column 2 is significantly negative. The coefficient of the interaction term in column 2 is −0.0031 and passes the significance test at the level of 1%, which is consistent with Hypothesis 2. This result shows that the health inequality caused by the air pollution suffered by the insured individuals in the inflow region is lower than that of the insured individuals in registered residence areas. This study reveals that the choice of medical insurance enrollment location (particularly migrants opting for coverage in their inflow regions) significantly moderates air pollution’s impact on health inequality. By enhancing healthcare accessibility and reimbursement rates (compared to hometown-based insurance), inflow-region enrollment enables timely medical interventions, reduces out-of-pocket costs, and alleviates household financial burdens, thereby mitigating health disparities caused by pollution.

Table 9 Test results of moderating mechanism.
Full size table

Discussion

First of all, this study found that air pollution (AQI) can increase the degree of health inequality, and different pollutants (PM2.5, PM10, CO) have a similar effect. Differentiated air pollution exposure will lead to differences in individual health inequality, and high air pollution will aggravate individual health inequality through relative deprivation. This finding further verified previous studies (Azimi et al., 2019; Bell & Ebisu, 2012), and Hypothesis 1 is verified. This paper provides new insights into the mechanisms behind these results, arguing that the reasons for this result are through differentiated air pollution exposure levels related to socioeconomic status and lack of adequate health care system. High air pollution exposure is highly associated with disadvantaged groups of socioeconomic status (Gray et al., 2014). High-income groups generally live in areas with low air pollution exposure, and air pollution control policies have a significant positive impact on low exposure and middle-income groups (Shen et al., 2020). Low-income people are more likely to be exposed to the working and living environment with high air pollution (Carson et al., 1997; Landrigan, 2017), which will aggravate the degree of health inequality among individuals. The lack of medical insurance system, uneven distribution of medical health resources, and poor accessibility of medical services are also important reasons for the aggravation of health inequality (Matus et al., 2012; Yang & Liu, 2018). Basic medical insurance will reduce the individual’s medical expenditure (Dou et al., 2018), and participating in medical insurance will aggravate the inequality of the poor’s medical care utilization (Zhang et al., 2023). When individuals suffer from the damage caused by air pollution, health inequalities can be alleviated through timely treatment, improving health status, and reducing the economic burden of individuals. This study adds to the existing literature by linking air pollution exposure to structural barriers in healthcare access.

Secondly, this study also finds that the impact of air pollution on health inequality is heterogeneous, and the air pollution impact on the degree of health inequality is greater for female groups. The reason why women are more vulnerable to air pollution may be due to the influence of physiological and social factors. Previous studies have also shown a similar view that women’s lungs and immune system may be more sensitive to air pollution, making them more vulnerable to the pollutants (Ghosh et al., 2007). Moreover, women may engage in more household and community activities, thus increasing their exposure to indoor air pollution (Parikh et al., 1999; Pitt et al., 2005). These research conclusions are consistent with this paper that air pollution has a greater impact on women’s health inequality. Moreover, the air pollution impact on health inequality is greater for the group over 60 years old, rural registered residence, low income, and low education level, respectively. This paper further advances previous findings by highlighting these subgroup disparities. Due to the decline of physiological function, weakened immune system, and chronic diseases, air pollution may further reduce the health level of the elderly, leading to health inequality (Brook et al., 2010; India State Level Dis, B (2019); Pope et al., 2002). Low-income groups and those with low education levels generally tend to live in areas with high air pollution, and when faced with air pollution, they take fewer relative measures and suffer more health damage (Bowe et al., 2018; Chakraborty, 2021). This study sheds new light on how environmental inequality and social vulnerability interact to reinforce health disparities, using the relative deprivation index approach to measure health inequality.

Finally, the moderating effect in this study finds that the place of medical insurance participation can moderate the relationship between air pollution and health inequality. Specifically, the migrants choosing the enrollment locations where they inflow into and settle down can reduce the health inequality caused by air pollution. Although the moderating effect of medical insurance has been documented by many studies (Lee et al., 2019; Lee et al., 2012; Yang & Hu, 2022), previous studies mainly focused on the relationship between length of residence, chronic diseases, and health expenditure, rarely addressing the moderating effect of medical insurance in the relationship between air pollution and health inequality. This study innovatively finds that choosing the inflow region for medical insurance enrollment can play a positive role in moderating the health inequality caused by air pollution. The reason may be that compared with the insurance in the registered residence, the insurance in the inflow region can improve the convenience of individual medical treatment and the reimbursement ratio. Through timely medical treatment, individuals can reduce the health damage caused by air pollution, reduce the personal medical expenses, and reduce the family burden, so as to increase the relative investment in health, improve their own health level, and reduce the degree of health inequality. Those research conclusions verify Hypothesis 2.

Conclusions

Air pollution remains one of the most significant environmental threats to human health, and the health inequality it induces—particularly among migrants—requires urgent attention. Drawing on longitudinal micro data from the 2015 and 2018 waves of the CHARLS, combined with official air pollution data, this study employs a 2SLS estimation approach to identify the causal effects of air pollution on health inequality. The findings confirm that air pollution significantly exacerbates health inequality, particularly among women, older adults, and socioeconomically disadvantaged populations. Furthermore, the CRMIS demonstrates a mitigating effect, alleviating the severity of pollution-induced health disparities among migrants who are enrolled in local insurance programs.

This paper contributes by showing that air pollution worsens individual health inequality using a relative deprivation index, and that the CRMIS helps mitigate this effect for migrants enrolled in inflow-region insurance. A limitation of this study is that the temporal scope of the AQI data is limited, which prevents us from capturing the relationship between air pollution and health inequality prior to 2014. Future research could extend the time frame to explore the long-term impacts of air pollution on health inequality more comprehensively.

Focusing on air pollution in China and its impact on the health of internal migrants, this study offers insights that are relevant to other countries confronting comparable challenges. The findings highlight the critical importance of incorporating health equality into environmental policy objectives and provide practical recommendations for policymakers to mitigate pollution-induced health disparities.