Does air pollution affect public health and health inequality? Empirical evidence from China

Accepted Manuscript Does air pollution affect public health and health inequality? Empirical evidence from China Tingru Yang, Wenling Liu PII:

S0959-6526(18)32591-5

DOI:

10.1016/j.jclepro.2018.08.242

Reference:

JCLP 14026

To appear in:

Journal of Cleaner Production

Received Date: 19 April 2018 Revised Date:

1 August 2018

Accepted Date: 22 August 2018

Please cite this article as: Yang T, Liu W, Does air pollution affect public health and health inequality? Empirical evidence from China, Journal of Cleaner Production (2018), doi: 10.1016/ j.jclepro.2018.08.242. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Does air pollution affect public health and health inequality? Empirical evidence from China Tingru Yanga,b, Wenling Liua,b,c,d* Center for Energy and Environmental Policy Research, Beijing Institute of Technology, Beijing 100081, China

b

School of Management and Economics, Beijing Institute of Technology, Beijing 100081, China

c

Sustainable Development Research Institute for Economy and Society of Beijing, Beijing 100081, China

d

Beijing Key Lab of Energy Economics and Environmental Management, Beijing 100081

SC

RI PT

a

Abstract: Air pollution and its effects on public health have received considerable attention in China. This paper extends discussion of the relationship between

M AN U

pollution and health by focusing on the effects of certain socioeconomic factors from an equality perspective. Hierarchical linear regression models are used to analyze the effects of environmental pollution on the health of residents and explore the inherent mechanisms through which environmental and economic factors contribute to increases in health inequality. The main results indicate that pollution poses

TE D

significant risks to health. Individual income and life satisfaction are found to be significantly positively related to health and this relationship is intensified by the effects of environmental pollution. The results show that health inequality is prevalent throughout China and is more severe in rural areas. The damage to health caused by

EP

pollution further increases the levels of health inequality to varying degrees in groups with different income levels. Pollution also increases the impact of income inequality on health inequality. Overall, the results to some extent validate the concept of an

AC C

“environment-health-poverty trap.” Key words: Health; Health inequality; Environmental pollution; Income inequality; Environment-health-poverty trap

1. Introduction As the world’s leading primary energy consumer, in 2016, China accounted for 62% of the world’s coal consumption (BP Statistical Review of World Energy, 2017). The high level of energy consumption has led to a series of negative environmental effects (Dangerman, 2013), including air pollution, especially in the Central and 1

ACCEPTED MANUSCRIPT Eastern regions of China (Niu et al., 2016). In addition to increasing the public awareness of environmental and health issues, the severe environmental problems in China may further exacerbate the gap between the rich and the poor (CCICED, 2013). In recent years, China has witnessed consistent increases in environmental pollution, the associated health effects have accelerated the depreciation of human capital and

RI PT

affected labor productivity and economic growth. Moreover, the health problems and inequality related issues caused by environmental pollution have become an important contributor to the “middle-income trap” (Zhang, 2013), which is a crucial problem in contemporary China. A number of studies have examined this issue. For example,

SC

Yang et al. (2013) found that environmental pollution in China accounted for 8%-10% of actual GDP, and the substitution effect of economic growth on the health of residents was far greater than the income effect, which in general reduced the level of

M AN U

health. Therefore, it is vital to pay more attention to the effects of pollution on health, especially in relation to poverty reduction and the promotion of sustainable and equitable development (Jennifer, 2010).

Methods including the health production function and Exposure-Response function are commonly used to estimate the health effects of air pollution. These studies have ever contributed a main body for academic evaluation on the effects of

TE D

air pollution on public health and generally confirmed statistically significant relationships between levels of pollutants such as PM and ozone in ambient air with mortality (Pope et al.,1995; Arceo et al., 2016) and other cardiopulmonary (Pope et al., 2004), cardiovascular and respiratory (Beatty and Shimshack., 2014) health outcomes.

EP

However, the health production function or Exposure-Response function cannot explain individual-level differences in health effects in the same exposure context.

AC C

Concretely, a population group with a specific socioeconomic status might be more sensitive than other groups to air pollution. Meanwhile, residents of an area with greater pollution do not necessarily have a higher exposure risk, because they might, for example, adopt preventive countermeasures. Because of disparities in socioeconomic status, the probability of exposure to pollution varies by group and individual. In addition, income inequality and other socioeconomic factors may cause and even aggravate the effects of pollution on health. Moreover, studies have shown that the combined effects of environmental pollution on health, poverty, and inequality can create an “environmental-health-poverty trap” (Qi and Lu, 2015). However, few studies have examined the intrinsic mechanisms of health inequality or the socioeconomic effects of pollution. 2

ACCEPTED MANUSCRIPT The lack of researches in these areas has hindered the formulation of effective environmental health policies because an economic perspective is needed to convert our understanding of the endogenous mechanisms of health inequality into practical policies. This paper may contribute to the literature from three aspects: (1) examining the issues of health and health inequality from a socioeconomic perspective and analyzing the health effects of environmental pollution; (2) focusing on the issue of

RI PT

inequality, analyzing the intrinsic relationships between health inequality, income inequality, and environmental pollution; (3) taking the method of econometric models to treat the audience objectives as being heterogeneous, considering the individual differences exposed to air pollution as far as possible.

SC

The remainder of the paper is as follows. Section 2 reviews the literature on environmental pollution and its effects on health. Section 3 outlines the methodology and the dataset used in this paper. Section 4 evaluates the situation of health inequality

M AN U

in China and further analyzes how environmental pollution affects public health and health inequality. Section 5 discusses the implications of the findings in relation to the environmental-health-poverty trap. Section 6 concludes the paper. 2. Literature review

Numerous studies have examined the effects of environmental pollution on

TE D

health and health inequality. These studies either focused on environmental pollution and its effects on health or estimated the levels of health inequality in different environmental and socioeconomic contexts. The early studies on the health effects of environmental pollution regarded public health as a form of economic capital. In a

EP

pioneering study, Grossman (1972) proposed the health production function model. Subsequent studies have sought to integrate economic methods in the field of

AC C

environmental health science (e.g., Alberini et al., 1997; Kumar and Rao, 2001; Dasgupta, 2004; Dzikuć and Piwowar, 2016). Because the effects of air pollution on public health are largely dependent on the risk of exposure to pollution (Coneus and Spiess, 2012), numerous studies have focused on the Exposure-Response (ER) functions. The ER functions quantify how many health-end outcomes or change in the death rate caused by a unit increase in a pollutant’s concentration level (Matus et al., 2012). For example, studies have evaluated how the environment affects health by calculating the probability of exposure to air pollution (Pan, 2007; Mestl, 2007) whereas others have sought to estimate the co-benefits of human health (Ezzati and Kammen, 2002; Spalding-Fecher, 2005; Milner et al., 2012). Assuming that the other 3

ACCEPTED MANUSCRIPT conditions are fixed, higher levels of exposure to environmental pollution can be considered to lead to greater health risks and health hazards. Accordingly, numerous studies have evaluated the effects of air pollution on mortality, life expectancy, respiratory disease, and hospitalization rates by combining the exposure response functions with the OLS and GAM metrological models (Chen et al., 2013; Zhang et

RI PT

al., 2014). Some studies considered the endogenous issues relating to the treatment of pollution and health (Qi et al., 2015) by, for example, examining the relationships between various pollution factors and the health of different groups, and issues such as education quality, labor supply, productivity, and economic growth.

SC

It should be noted that in these studies, the health production function or the exposure-response relationship is determining to the estimation results and thus their accuracy appears to be very important. And, these methods treat the audience

M AN U

objectives as being homogenous, but actually the environmental pollution indicators such as concentration of pollutants cannot represent the level of exposure of the target, and the individual exposure time at a certain level of pollution is unknown. Different from the above methods, the econometric model is also widely used to estimate the health effects of environmental pollution. As an examination on causal relationship, an econometric estimation is not necessary to be based on the influence mechanism

TE D

such as the exposure-response relationship and thus is capable to avoid the impacts of the ER function inaccuracy. For instance, recent studies have particularly discussed the health consequences of sustained exposure to air pollution by exploiting the Huai River policy in China. Chen et al (2013) found that Chinese space heating policy had

EP

dramatic impacts on pollution and human health with an econometric model, by examining the concentrations of total suspended particulates (TSP) for 90 cities from

AC C

1981 to 2000.

Besides confirming the health effects of pollution, health inequality caused by

environmental pollution has also become an issue of wide concern. Socioeconomic factors such as income, education level, occupation, and family size have been shown to have a significant influence on health (Ji et al., 2015; Kravitz-Wirtz, 2016; Jiménez et al., 2016; Bakhtsiyarava and Nawrotzki, 2017). Moreover, socioeconomic status was proved to be associated with the varying risk of exposure to environmental pollution among different groups. For instance, Schoolman and Ma (2012) found that townships with a higher percentage of rural migrants are more likely to be exposed to high levels of air and water pollution. Yang et al. (2013) concluded that the cost of 4

ACCEPTED MANUSCRIPT environmental pollution in developed areas is significantly higher than that in underdeveloped areas. Environmental pollution can cause certain diseases, which may affect the sufferers’ socioeconomic status and lead to increased income inequality. Alternatively, income inequality and other related social inequalities can be directly hazardous to

RI PT

health, thus causing the environment-health-poverty trap (Qi and Lu, 2015). A number of studies have examined the effects of income inequality in different geographical zones including state, metropolitan, and county levels. Charafeddine and Boden (2008) used a hierarchical logistic regression to examine the association between general

SC

self-reported health and fine particulate pollutants and proposed that income inequality is a modifier that controls for the usual confounding effects. Finally, income inequality was found to modify the association between public health and air

M AN U

pollution. Studies have examined the association between income inequality and health at the state level the U.S. and found that income inequality is linked to higher all-cause mortality risk, lower self-rated health, higher prevalence of depressive symptoms, and more adverse health-related behaviors (Subramanian and Kawachi, 2006). In a review of 155 studies on the association between income inequality and population health, Wilkinson and Pickett (2006) found that the studies conducted in

TE D

large areas are more supportive of the association between inequality and health. However, no studies have verified whether this association applies to the urban or community levels in China.

Although numerous studies have examined the relationship between the

EP

environment and health from the perspective of fairness, most of the studies only focused on the health inequalities caused by pollution and failed to address the

AC C

socioeconomic effects. However, the impact of pollution on health inequality can further affect the distribution of income and social welfare. Although a few studies have explored the effects of income inequality on health (Deaton, 2003; Org, 2003), the mechanisms driving these effects have not been effectively interpreted or tested with empirical data. Hence, this paper establishes econometric models to avoid the inaccuracy of ER function, analyzes the factors that affect the health status of residents who are exposed to environmental pollution, explores the inherent mechanisms driving health inequality, and examines the socioeconomic dimensions of these mechanisms to highlight the relationship between income inequality and health inequality. 5

ACCEPTED MANUSCRIPT 3. Methodology 3.1. Data sources and variables selection The data used in this paper came from the 2014 national survey of China Family Panel Studies (CFPS). CFPS is a nationwide, comprehensive, longitudinal social

RI PT

survey that is used for research on a large variety of social phenomena in contemporary China.1 The 2014 survey covered 25 provinces and nearly 16,000 households. The dataset used in this paper comprised 26,396 sets of valid individual data. Basic information and descriptive statistics on key variables are shown in Table

SC

1.

Table 1 Descriptive statistics on key variables Definition

Total

Mean

Income

Log form of per capita income in the past 12 months (Yuan)

House Number of owned houses

Vehicle ownership

Number of household

EP

Family size

Number of owned vehicles

resident population Age

AC C

Age of adults (>18)

Subjective Health

judgment

Self-rated scores of health status (1-5 points)

Medical

Log form of

expenditure

medical expenditure in the

SD

Mean

SD

Rural Mean

SD

3.96

0.43

4.11

0.39

3.84

0.41

0.20

0.49

0.24

0.52

0.16

0.45

0.17

0.38

0.21

0.41

0.14

0.35

4.23

1.88

3.87

1.74

4.53

1.94

46.49

16.46

46.41

16.59

46.55

16.34

2.82

1.31

2.80

1.26

2.83

1.36

3.80

0.49

3.93

0.46

3.69

0.48

TE D

ownership

Urban

M AN U

Variable

per capita

past 12 months (Yuan) # SD

standard deviation

In the CFPS survey, the respondents are asked to self-rate their health status. 1

Readers can obtain more information from the following website. http:// www.isss.edu,cn/cfps/ 6

ACCEPTED MANUSCRIPT Three main types of health indicators, namely, medicine, organism function, and subjectivity, tend to be used in the research on health evaluation (Xie, 2009). In this paper, the self-rated health status is used as a major indicator to represent individuals’ health status. Self-rated health has significant advantages with respect to integrity, availability, and robustness (Qi, 2006; Qi and Li, 2011). Although environmental

RI PT

factors are considered to have major effects on health and medical research suggests that more and more diseases are related to environmental factors, certain health indicators are unable to adequately reflect the general effects of environmental factors. Overall, self-rated health data can be consistent with the need to analyze the

SC

relationship between the environment and health (Qi and Lu, 2015).

Specifically, the self-rated health data are obtained by asking the participants to rate their health on a scale of 1 (“very unhealthy”) to 5 (“very healthy”). Given the

M AN U

subjective nature of the self-rated health data, the effects of the individuals’ sickness status are also taken into account. To calculate the health status of individuals, the negative effects of chronic diseases are combined, that is, the health score equals the self-rated health score plus the chronic disease score. The life satisfaction variable combines the measures of an individual’s satisfaction with his/her own life, confidence in the future, and satisfaction with family life. The measures for the other

TE D

variables are shown in Table 2.

Table 2 Outline of the variables

Item

Description and measurement

Education

The highest level of education

Income

Per capita household income in the past 12

EP

Category

House ownership

AC C

Basic Information

months Number of owned houses

Vehicle ownership

Number of owned vehicles

Occupation

Occupation of the respondents

Subjective health

Self-rated scores of health status (1-5 points)

judgment

Medical and health

Objective health

Scores of whether suffering from any types of

judgment

chronic diseases (assigning -1 point with one type, and so on)

Life satisfaction

[ Satisfaction with one's life + Confidence in one's future /2+ Satisfaction with family life] /2 7

ACCEPTED MANUSCRIPT (1-5 points) Medical resources

Sickbeds per million people

Medical expenditure

Per capita household medical expenditure in the past 12 months Concentrations of CO2, SO2, NOx and industrial

Gaseous emission

Industrial waste water

RI PT

dust

Types of pollutants

Volume of industrial wastewater discharge

3.2. Methods for measuring income inequality

SC

In line with previous studies, the Gini coefficient method was used to test the degree of regional income inequality (He and Hong, 2016). The Gini coefficient

M AN U

represents income inequality as a number between zero and one; the higher the number, the greater the level of inequality, as shown in the following equation: G=


 




∑
∑ 
  −  =   ∑
(2 −  − 1) 

Where: G: Gini coefficient;

(1)

TE D

n: the number of samples;

µ: the average income (Yuan); and x : the income of sample i (Yuan).

EP

This method is also used to calculate the inequality of health. 3.3. Methods to assess health inequality

AC C

There are two methods commonly used to measure the degree of health inequality. The slope indices of inequality (SII) (Rao and Yao, 1998), which are defined as the slope of the regression lines between the health status of each group and the order of the corresponding socioeconomic group, are used to reflect changes in the level of health. As shown in Figure 1, the samples are first divided into groups according to their socio-economic status, and the average health status of each group is calculated and the groups are classified according to their socioeconomic status. Because the SII can reveal changes of health from an economic perspective, it is able to reflect the effects of different socioeconomic factors on health inequality. The health concentration curve is used to measure the degree of health inequality 8

ACCEPTED MANUSCRIPT associated with socioeconomic status. As shown in Figure 1, the horizontal axis of the concentration curve represents the cumulative percentage of the population according to economic status, and the vertical axis is the cumulative percentage of the level of health or economic loss. If the concentration curve intersects with the diagonal line, this means that the level of health is evenly distributed among the socioeconomic

RI PT

groups. If the concentration curve is below the diagonal, this means that the lower socioeconomic groups have largely lower health, and the more the concentration

80%

Ⅴ

Ⅳ

Ⅲ

TE D

Average health

M AN U

Cumulative percentage of health status

100%

SC

curve deviates from the diagonal, the greater the degree of health inequality.

Ⅱ

Ⅰ Correlation rank

60%

40%

20%

0% 0% 20% 40% 60% 80% 100% Cumulative percentage of population (sorted by socio-economic condition)

EP

Fig.1. The SII method (left) and the health concentration curve method (right)

Therefore, the SII and the health concentration index satisfied the measurement

AC C

of health equality from different aspects (Rao and Yao, 1998). The relationship between the economy and health equality can be revealed by sorting and grouping the social and economic conditions in a region. Hence, to avoid instability in the sample size, the SII is combined with the health concentration curve to describe the current levels of health inequality. 3.4. Econometric analysis model The basic problem of sociology is to associate personal attributes with their surrounding groups and structures, which means that individual behavior is not only influenced by the characteristics of the individual but also by the environmental factors (Bryk and Raudenbush, 2002). Accordingly, hierarchical data structures are 9

ACCEPTED MANUSCRIPT widely used in the literature. The basic assumptions of the traditional linear model are linearity, normality, homogeneity of variance, and independence. The latter two hypotheses may not apply to the hierarchical data structure used in this paper, and explaining the data merely at the individual level would misrepresent the results (Zhang, 2005). The advantage of using a hierarchical linear structure is that it can

RI PT

decompose the overall health inequality into different levels, which cover both the micro level (individual) and macro level (social environment).

Previous studies have proved that socio-economic factors such as income, education level, occupation, and family size have significant influences on public

SC

health (Kravitz-Wirtz, 2016; Jiménez et al., 2016; Bakhtsiyarava and Nawrotzki, 2017). Meanwhile, individual subjective factors, such as life satisfaction could also affect the mental and physical health status (Easterlin et al., 2012; Knight and

M AN U

Gunatilaka, 2011; Gunatilaka, 2010). Therefore, this study mainly focuses on the effects of environmental pollution on health status and establishes the econometrical model combining the influence of socio-economic factors and individual subjective factors on health. These variables can be expressed in two levels: health and socioeconomic factors at the individual level and environmental pollution at the regional level.

TE D

Thus, using Raudenbush and Bryk (2002) as a reference, a hierarchical regression model is used to distinguish the multilevel mechanisms that influence health. The analysis mainly considers the effects of pollution on the relationship between health and certain socioeconomic factors. These two mechanisms affect the

EP

health of residents directly through the exposure to different levels of pollution and indirectly through the differences in the socioeconomic factors. However, because the

AC C

regional codes used in the CFPS survey are not publicly available, the individual data are unable to be matched with the pollution and medical resource data at the city level. Thus, the data are collected at the individual and province levels, as shown in the following equation: The first layer:

ℎ =  + 
× !" +  × #$% + & × ' + ( × )* + + × , + -

(2)

The second layer:  = - + -
(./%/) + 0 
= -
 + -

(./%/) 10

(3)

ACCEPTED MANUSCRIPT  = - + -
(./%/) …… + = -+ + -+
(./%/) Where:

RI PT

Healthij: health status of sample i in province j; Incij: per capita household income in the past 12 months of sample i in province j; Eduij: education level of sample i in province j; Genij: sex of sample i in province j; Ageij: age of sample i in province j; Satij: life satisfaction of sample i in province j; and

SC

Pollutionj: air pollutants (determining the weights of SO2, NOx, and industrial dust by the entropy method), greenhouse gas (CO2), and industrial waste water in region j.

M AN U

In addition, a hierarchical data structure is used to explore the inherent mechanisms of health inequality. The first layer estimates the effects of income and income inequality on health inequality at the county level, and the second layer considers the influence of pollution and medical resources at the provincial level. The equation is as follows: The first layer: (4) The second layer

TE D

ℎ1%23 =  +
× '3 +  × !"3 + -3

 = - + -
(./%/) +- (,"45$) ++0

(5)

EP


= -
 + -

(./%/)  = - + -
(./%/)

AC C

Where:

Health inequalityhj: the health inequality of county h in province j; Ginihj: Gini coefficient of county h in province j; Inchj: per capita household income in the past 12 months of county h in province j; Pollutionj: greenhouse gas (CO2) and pollutants (SO2, NOx, industrial dust, industrial waste water) in province j; and Sickbedj: the number of sickbeds per million people in province j. 4. Health inequality in China 4.1. Description of the health inequality in different areas In order to fully analyze the effects of environmental pollution on health and 11

ACCEPTED MANUSCRIPT health inequality, this paper describes the status of health inequality in different areas with the methods in Section 3.3 and focuses on the differentiation of health inequality in Eastern, Central, and Western areas. To calculate the SII for assessing health inequality, the samples are then divided into six groups and the groups are ranked in terms of per capita income, with the per capita income varying from 1000 to 100,000

RI PT

Yuan. The proportion of each group in the overall sample and the average health status of each group are then calculated. The SII in different regions are shown in Table 3. As the table shows, the SII of the three regions are positive, which indicates that there is an upward trend in the average health status in relation to the growth of

SC

income. Obviously, health inequalities exist within all three regions. Although the health of minority high-income residents is comparatively good, a large proportion of

M AN U

low-income residents are more likely to be in poor health.

Table 3 The SII in different regions

Proportion of total samples 65.62% 24.35% 7.16% 1.45% 0.83% 0.59% 0.038

Central

Average health status 2.83 2.85 2.76 2.91 2.67 3.18

Proportion of total samples 78.64% 17.14% 2.88% 0.75% 0.38% 0.21% 0.001

TE D

1 2 3 4 5 6 SII

Eastern

EP

Income groups

Average health status 2.86 2.84 2.87 2.84 2.94 2.82

Western

Proportion of total samples 86.44% 11.29% 1.71% 0.42% 0.11% 0.03% 0.228

Average health status 2.74 2.80 2.82 3.11 2.43 4.5

AC C

Comparing the three regions, the most serious health inequalities are found in

Western China, with the highest SII being 0.228. In Western China, the average health status of low-income residents (who account for 86.44% of the total population) is only 2.74, which is the lowest among the three regions, while the health status of high-income residents (who account for only 0.03% of the population) is as high as 4.5, well above the levels in the other regions. According to the economic statistics in 2014, the per capita disposable income in China was 20,167 Yuan, whereas the figures were 25,954 Yuan in the Eastern region, 16,868 Yuan in Central region, and 15,376

12

ACCEPTED MANUSCRIPT Yuan in the Western region.2 Previous research has shown that the gap between the rich and poor in China is more significant in the poor areas than the rich areas (Li and Xu, 2017). The results are consistent with this finding. The area with the lowest disposable income, Western China, has a relatively large gap between rich and poor, which further demonstrates the relationship between health inequality and income

RI PT

inequality. The health concentration curve shown in Figure 2 further proves the existence of health inequalities. The health concentration curve is below the diagonal, which indicates that lower health is concentrated in populations with lower income levels.

SC

With respect to the different regions, the concentrated curve of the West is slightly farther away from the diagonal. Although the difference is small, the curve indicates a negative relationship between income and health inequality, which is further analyzed

M AN U

100%

Western

80%

Eastern Central

TE D

60%

40%

20%

EP

Cumulative percentage of health status

in the next section.

0%

AC C

0%

20% 40% 60% 80% 100% Cumulative percentage of samples (sorted by per capita income )

Fig.2. The health concentration curve in different regions

To analyze the different levels of health inequality in urban and rural China, the

urban and rural samples are divided into six groups with incomes ranging from 1000 to 150,000 Yuan. As shown in Table 4, the degree of health inequality in the rural areas is higher than in the urban centers, where the SII is 0.34. The low-income residents account for 94.38% of the total population, whereas the high-income groups 2

Data source: 2015 China Statistical Yearbook. 13

ACCEPTED MANUSCRIPT only account for 0.01% and have average health levels as high as 5. According to the economic statistics for 2014, although the average per capita disposable income in China was 20,167 Yuan, the figures were 28,843 Yuan for the urban areas and 10,488 Yuan for the rural areas.3 These figures confirm that the rural and urban areas have different standards of living. Because people with a low disposable income may not the inequality of health. Table 4 The SII in urban and rural areas

0.12

0.34

TE D

SII

Rural Proportion of Average total samples health status 94.38% 2.83 4.90% 2.96 0.49% 2.88 0.19% 3.04 0.03% 3.25 0.01% 5

SC

1 2 3 4 5 6

Urban Proportion of Average total samples health status 80.17% 2.81 16.51% 2.77 2.22% 2.89 0.90% 2.91 0.14% 3.29 0.05% 3.33

M AN U

Income groups

RI PT

able to afford high medical bills, this income inequality is likely to be associated with

Similarly, the health concentration curves indicate the degrees of health inequality in the urban and rural areas. As shown in Figure 3, the concentration curves for both the urban and rural areas are below the diagonal, which means that lower

EP

levels of health are mainly concentrated in the lower socioeconomic areas. In addition, the curve for the rural areas shows a slightly greater deviation from the diagonal

AC C

compared with the urban areas, thus indicating that areas with lower incomes are more likely to suffer from health inequality.

3

Data source: 2015 China Statistical Yearbook. 14

100% Rural Urban

80% 60%

RI PT

40% 20% 0% 0%

20%

40%

60%

80% 100%

SC

Cumulative percentage of health status

ACCEPTED MANUSCRIPT

Cumulative percentage of samples (sorted by per capita income )

M AN U

Fig.3. Health concentration curves in urban and rural areas

In this section, the levels of health inequality in the different regions are quantified and health inequality is found to be more likely to be concentrated in poor areas. In the next section, the differences in health and health inequality are examined using econometric models, and the factors that influence health and health inequality,

4.2. Empirical results

TE D

and their relationship with air pollution and income inequality, are analyzed.

Before the hierarchical linear analysis, this paper conducted a correlation test and

EP

proved no existence of multicollinearity. As illustrated in the methodology section, the regression model has two different layers. The regression results are shown in Table 5, the regression coefficients in the first layer represent the effects of pollution and

AC C

individual socioeconomic factors on health status, and the regression coefficients in the second layer reflect the influence of the pollution variables on the effects of the socioeconomic factors. The second layer indicates whether the effects of the socioeconomic factors change under the context of pollution. The samples are classified into urban and rural areas and the synthetic indexes of SO2, NOx, and industrial dust are represented using the entropy method. To evaluate the effects of pollution on health status, industrial air pollutants, carbon dioxide, and industrial wastewater are used as indicators. The results indicate that increased pollution has significant and negative effects on individual health. Air pollutants and industrial wastewater are found to be associated with a larger decline in 15

ACCEPTED MANUSCRIPT self-rated health status in the rural areas than in the urban areas. The results confirm that income levels have a general effect on health, such that the higher the income level, the better the health of the population. This result is not surprising because low-income groups and rural residents are more likely to expose to environmental pollution. For instance, the floating populations, who face employment

RI PT

barriers in the labor market and lack knowledge and skills, are mostly concentrated in high intensity and long duration working areas, where safety and health cannot be guaranteed. Moreover, low-income groups and rural residents have relatively little knowledge about pollution and disease prevention, and are usually unable to pay for

AC C

EP

TE D

M AN U

SC

medical expenses and other healthcare costs.

16

ACCEPTED MANUSCRIPT

Dependent variable: Self-rated health status (Intra-Class Correlation =0.061 ) All samples (Mean VIF=1.58)

Urban (Mean VIF=1.57)

CO2

Industrial waste water

Air

CO2

Pollutant

ICC=0.021

ICC=0.021

Pollutant

ICC=0.020

AIC=93430.9

AIC=93430.9BIC=93526.0

AIC=93430.9

BIC=93526.0

Air

CO2

Industrial

ICC=0.01 3

waste water

Pollutant

ICC=0.022

waste water

ICC=0.013

AIC=38133.1

ICC=0.013

ICC=0.022

AIC=

ICC=0.022

AIC=38133.1

BIC=38273.7

AIC=38133.1

AIC=

46638.2

AIC= 46638.2

BIC=38273.7

46638.5

BIC=

BIC= 46721.7

BIC=

46721.7

BIC=38273.7

M AN U

BIC=93526.0

Estimation of intercept ( β0j ) [0.000]

[0.000]

-0.006***

(-01)

46782.2

2.578***

2.602***

2.604***

2.671***

2.713***

2.854***

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

-0.001**

-0.058*

-0.005***

-0.001

-0.003***

-0.003

0.000

-0.079*

[0.000]

[0.002]

[0.099]

[0.000]

[0.152]

[0.000]

[0.201]

[0.211]

[0.080]

Random

0.034**

0.038***

0.035**

0.010**

0.011**

0.010**

0.036**

0.039**

0.030**

Effects

[0.028]

[0.007]

[0.017]

[0.023]

[0.022]

[0.022]

[0.027]

[0.013]

[0.025]

(-00) Pollution

Estimation of income ( β1j )

TE D

2.534***

Intercept

EP

2.003***

Rural (Mean VIF=1.50)

Industrial

SC

Air

2.282***

RI PT

Table 5 Regression results for the factors influencing health

0.042**

0.061

0.001**

0.082

0.092

0.064

0.068

0.083

0.006*

(-01)

[0.017]

[0.427]

[0.072]

[0.198]

[0.116]

[0.307]

[0.487]

[0.333]

[0.096]

Pollution

-0.001

0.000

-0.010

0.002**

0.000

-0.000

0.000

0.000

-0.032

(-11)

[0.648]

[0.593]

[0.721]

[0.020]

[0.497]

[0.309]

[0.997]

[0.856]

[0.305]

Random

0.013*

0.013*

0.014*

0.002

0.002

0.003

0.018**

0.019**

0.017**

Effects

[0.098]

[0.082]

[0.095]

[0.556]

[0.448]

[0.655]

[0.021]

[0.024]

[0.025]

AC C

Income

Estimation of education ( β2j )

17

ACCEPTED MANUSCRIPT

-0.030

-0.014

-0.018

-0.032

-0.017

0.047*

0.009

0.016

0.014

-20) (-

[0.310]

[0.620]

[0.174]

[0.256]

[0.540]

[0.091]

[0.882]

[0.660]

[0.312]

0.000

0.000

0.008

0.000

0.000

0.000

0.000

0.000

0.030**

[0.615]

[0.892]

[0.559]

[0.617]

[0.989]

[0.721]

[0.849]

[0.647]

[0.002]

0.003***

0.003***

0.003***

0.002*

0.002

0.002

0.001

0.001

0.001

[0.001]

[0.001]

[0.001]

[0.081]

[0.727]

[0.655]

[0.372]

[0.386]

[0.356]

-0.035***

-0.036***

-0.034***

-0.022***

-0.023***

-0.023**

-0.025***

-0.026***

-0.027***

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

-0.002

-31) (-

[0.401]

[0.163]

[0.306]

[0.406]

[1.000]

[0.875]

[0.702]

[0.762]

[0.121]

Random

0.000**

0.000**

0.000**

0.000

0.000

0.000

0.000**

0.000**

0.000**

Effects

[0.036]

[0.045]

[0.026]

[0.788]

[0.994]

[0.654]

[0.026]

[0.026]

[0.015]

0.371***

0.369***

0.300***

0.126***

0.143***

0.123***

0.401***

0.398***

0.280***

(-40)

[0.000]

[0.000]

[0.000]

[0.000]

[0.001]

[0.000]

[0.000]

[0.000]

[0.000]

Pollution

-0.002

0.000**

0.004

0.001

0.000

0.001

-0.002**

0.000

-0.017

(-41)

[0.130]

[0.019]

[0.886]

[0.222]

[0.927]

[0.179]

[0.012]

[0.004]

[0.497]

0.004

0.003

0.003

0.000

0.000

0.000

0.002

0.002

0.002

[0.268]

[0.306]

[0.536]

[0.348]

[0.789]

[0.655]

[0.522]

[0.511]

[0.568]

0.248***

0.230***

0.027**

0.212***

0.231***

0.282***

[0.000]

[0.000]

[0.000]

[0.092]

[0.000]

[0.000]

[0.000]

-0.017

0.001

0.000

0.000

0.002***

0.000**

-0.035***

[0.262]

[0.169]

[0.829]

[0.739]

[0.003]

[0.014]

[0.000]

(-21) Random Effects

Pollution

Random Effects

Estimation of satisfaction ( β5j ) 0.294***

0.285***

(-50)

[0.000]

[0.000]

Pollution

0.001*

0.000**

(-51)

[0.094]

[0.020]

Satisfaction

AC C

Gender

EP

Estimation of gender ( β4j )

TE D

Age (-30)

M AN U

Estimation of age ( β3j )

0.351***

SC

Pollution

RI PT

Education

18

ACCEPTED MANUSCRIPT

Effects

0.001

0.001

0.001

[0.282]

[0.537]

[0.556]

0.002

0.001

0.001

0.000

0.001

0.001

[0.138]

[0.737]

[0.756]

[0.374]

[0.252]

[0.256]

RI PT

Random

AC C

EP

TE D

M AN U

SC

Note. *Significant at 10% level. **Significant at 5% level. ***Significant at 1% level

19

ACCEPTED MANUSCRIPT In addition to income, other individual related variables show significant effects on health. As expected, age is negatively correlated with individual health, meaning that older residents tend to have poorer health. Males generally have better health status than females. People with high life satisfaction also generally have better self-rated health, which to some extent reflects the impact of psychology, because a result is consistent with the findings of Zhang et al. (2017).

RI PT

positive attitude toward life tends to be associated with good physical health. This The interaction between the individual related variables and pollution variables is also estimated. The income and air pollutant variables show a significant and

SC

positive interaction in the models for the urban samples. This indicates that the effect of income on health in urban areas is more obvious in the regions with higher levels of pollution or worse environmental quality. In general, as levels of pollution increase,

M AN U

income has a greater influence on health. This result also suggests that income increases the effect of health inequality, because high-income groups usually have the economic capacity to avoid the health risks caused by pollution. In some of the models, the interactions of the pollution and personal life satisfaction items have significant positive effects on health. These results are not surprising because, in areas with higher levels of pollution, mental happiness is likely to have a more significant

TE D

effect on individual health.

The findings suggest that lower socioeconomic groups are generally exposed to higher health risks. The damage caused by pollution increases the levels of health inequality in different socioeconomic groups to varying degrees. Pollution is an

EP

important mechanism linking socioeconomic status and health inequality. Next, the internal mechanisms of the effects of income and income inequality on health

AC C

inequality are analyzed.

As shown in Table 6, the results of the regression model show that air pollution

can generate health inequalities. This effect on health inequality appears to be slightly different for the different kinds of pollutants. For example, income inequality has a relatively greater impact on health inequality when combined with air pollution than with industrial wastewater. When the concentration of industrial dust increases, the effect of income inequality on health inequality is further intensified. This result also suggests air pollution has negative effects on health. There is a significant negative correlation between income and health inequality. Low-income groups are more likely to expose to pollution because, as 20

ACCEPTED MANUSCRIPT abovementioned, the floating and rural populations are more likely to be concentrated in areas where safety and health cannot be guaranteed. Furthermore, low-income groups usually lack knowledge of pollution and disease prevention, and are unable to pay for medical expenses and other healthcare costs. The results also show a positive correlation between income inequality and

RI PT

health inequality, such that a high degree of income inequality in a region may lead to increased health inequality. The Gini coefficient represents the degree of income inequality, with higher numbers indicating high levels of inequality. Inequality is mainly reflected in the inequality of income distribution and the unequal access to

SC

public resources. Because high-income groups have greater access to medical and health resources, the inequalities associated with healthcare resources and medical affordability may contribute to increased health inequality. Moreover, pollution

M AN U

prevention depends on both private and public intervention and in areas with low levels of economic development, the government may have limited capacity to supply public infrastructure. For instance, the release of pollution information and the provision of environmental medical services are mostly concentrated in the urban areas, while rural residents lack access to information about pollution and medical

TE D

infrastructure.

Table 6 Regression results for the factors influencing health inequality Dependent variable: health inequality (Intra-Class Correlation =0.061; Mean VIF=1.63 ) Industrial dust

SO2

NOx

CO2

Industrial waste water

EP

Estimation of intercept ( β0j ) 0.234 ***

0.334 ***

0.381 ***

0.375**

0.220 ***

(-00)

[0.000]

[0.000]

[0.000]

[0.000]

[0.000]

Pollution

0.000**

0.022***

0.001***

0.000**

0.002*

(-01)

[0.005]

[0.007]

[0.006]

[0.026]

[0.056]

Sickbed

-0.098***

-0.001***

-0.059***

-0.019

-0.013***

(-02)

[0.000]

[0.137]

[0.000]

[0.291]

[0.001]

Random

0.002

0.001

0.001

0.001*

0.001

Effects

[0.155]

[0.147]

[0.125]

[0.060]

[0.123]

AC C

Intercept

Estimation of Gini ( β1j ) Gini

0.071*

0.773***

0.339**

0.656

0.056

(-01)

[0.100]

[0.003]

[0.023]

[0.133]

[0.156]

Pollution

0.000***

-0.049

-0.001

0.000

0.000

(-11)

[0.010]

[0.805]

[0.708]

[0.689]

[0.989]

Random

0.225

0.222

0.222

0.22

0.211

21

ACCEPTED MANUSCRIPT Effects

[0.156]

[0.124]

[0.133]

[0.136]

[0.126]

Estimation of income ( β5j ) -0.058**

-0.186***

-0.086***

-0.042*

-0.083**

-50) (-

[0.063]

[0.000]

[0.000]

[0.767]

[0.002]

Pollution

0.000

-0.085

-0.002

0.000

-0.001

(-51)

[0.964]

[0.190]

[0.157]

[0.186]

[0.960]

Random

0.018

0.018

0.018

0.018

0.025

Effects

[0.266]

[0.257]

[0.259]

[0.229]

[0.245]

RI PT

Income

Note. *Significant at 10% level. **Significant at 5% level. ***Significant at 1% level.

The medical infrastructure variables are also controlled and the medical

SC

infrastructure is found to have a significant impact on health inequality. Specifically, increased medical infrastructure can help reduce the level of health inequality, which

M AN U

suggests that governments should increase their investment in medical infrastructure. In summary, the results of the regression models show that pollution is one of the main environmental factors affecting health and that pollution can increase the level of health inequality. However, the levels of health inequality vary in regions with different income levels, with poorer populations suffering from higher levels of health inequality. Furthermore, the impact of income inequality on health inequity increases

TE D

in areas with higher levels of pollution, and lower income groups are generally exposed to higher health risks. The damage caused by pollution further increases the levels of health inequality to varying degrees among the groups with different income levels.

EP

5. Implications for the Environmental-Health-Poverty Trap A large body of literature has examined whether economic development and

AC C

environmental quality can achieve convergence (Wang et al., 2016; Onater-Isberk, 2016; Azam and Khan, 2016). This issue is increasingly being discussed in relation to the idea of the environmental-health-poverty trap. In an early study of the environmental-health-poverty trap, John et al. (1995) used a limited-time agent’s OLG model to examine the relationship between economic growth and environmental quality. They found that areas with sufficient capital and environmental quality experienced continuous improvement in growth and environmental governance, whereas areas with unfavorable initial conditions experienced low capital and low environmental quality. Xepapadeas (1997) examined how emissions reductions affect the relationship between economic growth and the environment. 22

ACCEPTED MANUSCRIPT As early as the 1990s, Chinese scholars began examining the coordinated development of the economy and the environment in poor areas, and proposed the cycle of low income-ecological destruction-low income, which, to a certain extent, is a precursor to the idea of the environment-poverty trap. However, subsequent research has not followed this path and focused on the internal mechanisms driving

RI PT

the interaction between economic development and environmental protection. Instead, studies have largely focused on the unidirectional relationship between economic development and environmental pollution (Chen, 2013; Jin et al., 2011). Qi and Lu (2015) recently expanded the definition of the environment-poverty trap to the

SC

environment-health-poverty trap and concluded that pollution increases the inequality between regions and urban-rural areas through its effects on health.

This paper focuses on the issues of pollution, health, and health inequality and

M AN U

the results support the concept of an environment-health-poverty trap to some extent. The findings indicate that pollution harms health and increases health inequality. In addition, income is found to be strongly correlated with health and health inequality. In particular, poorer regions have higher levels of health inequality, and the gap between the rich and the poor further increases this health inequality. Considering the interactions between pollution, health (inequality), and income (inequality),

TE D

low-income groups who face high pollution conditions and bad health are very likely to fall into the environment-health-poverty trap. 6. Conclusion

EP

This paper examines the health effects of environmental pollution and their implications for health inequality. Health inequality is found to exist in different

AC C

regions of China, and is largely concentrated in poor areas, such as the rural and Western areas. The quantitative analysis of health inequality shows that income (inequality) is associated with health (inequality) in the context of environmental pollution.

The findings indicate that increased pollution has significant negative effects on

health, and that air pollution contributes to increased health inequality. These effects on health and health inequality appear to be slightly different for different kinds of pollutants. Air pollution is found to cause a larger decline in self-rated health status than other pollutants. The findings also suggest that the health effects of pollution are inseparable from socioeconomic factors. For example, income is found to be positively correlated with health status, such that areas with higher average incomes 23

ACCEPTED MANUSCRIPT have better health. Moreover, income is found to have more noticeable effects on health in regions with higher levels of pollution or worse environmental quality. Other socioeconomic variables, such as age, gender, and life satisfaction, also have significant effects on health. The internal mechanisms of the effects of income and income inequality on

RI PT

health inequality are further analyzed. Air pollution mainly contributes to increased health inequality in low-income areas because the residents are unable to choose their conditions of life. Overall, the findings show that a high level of income inequality in a region may accordingly contribute to increased health inequality, which thus verifies

SC

the concept of an environment-health-poverty trap to some extent. Moreover, the findings suggest that low-income groups who face high pollution conditions and bad health are very likely to fall into the environment-health-poverty trap.

M AN U

In summary, this paper examines the interactions between pollution, health (inequality), and income (inequality). The findings suggest that pollution is an important mechanism driving the effects of socioeconomic status on health inequality. Lower income groups are more likely to expose to pollution and experience greater health effects. In the different regions of China, especially the impoverished rural and Western areas, environmental pollution may be imposing an even heavier health and

TE D

economic burden on the population. Moreover, the income inequality in a region may contribute to increased health inequality, which suggests that solutions to the problem of health inequality in China are urgently needed. Because high-income groups have greater access to medical and health resources, policies need to be devised to reduce

EP

the inequality in healthcare resources and medical affordability. The findings show that increased medical infrastructure can help reduce the level of health inequality,

AC C

which suggests that governments should increase their investment in medical infrastructure.

Finally, the paper has some limitations with respect to data selection. Because

the regional codes in the CFPS are not publicly available, the provincial level pollution and medical resource data are used. Future studies should try to match individual data with pollution and medical resource data at the city level. Acknowledgement: The authors acknowledge financial support received through China’s National Key R&D Program (2016YFA0602603), the National Natural Science Foundation of China (no.71521002 and no.71573016) and the Special Fund for Joint Development 24

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Program of Beijing Municipal Commission of Education.

25

ACCEPTED MANUSCRIPT Reference [1] Alberini A, Cropper M, Fu TT, Krupnick A, Liu JT, Shaw D, Harrington W. Valuing health effects of air pollution in developing countries: the case of Taiwan. Journal of Environmental Economics and Management 1997; 34: 107-126.

RI PT

[2] Arceo, E., Hanna, R., Oliva, P. Does the effect of pollution on infant mortality differ between developing and developed countries? Evidence from Mexico City. The Economic Journal 2016; 126: 257-280.

[3] Azam M, Khan A Q. Testing the Environmental Kuznets Curve hypothesis: A comparative

& Sustainable Energy Reviews 2016; 63:556-567.

SC

empirical study for low, lower middle, upper middle and high income countries. Renewable

[4] Bakhtsiyarava M, Nawrotzki R J. Environmental inequality and pollution advantage among

M AN U

immigrants in the United States. Applied Geography 2017; 81(1):60-69.

[5] Beatty, T.K., Shimshack, J.P. Air pollution and children's respiratory health: A cohort analysis. Journal of Environmental Economics and Management 2014; 67: 39-57. [6] BP Statistical Review of World Energy,

https://www.bp.com/zhcn/china/reports-and-publications/ bp 2017- html; 2017.

TE D

[7] Bryk A S, Raudenbush S W. Hierarchical linear models: applications and data analysis methods. Publications of the American Statistical Association, 2002. [8] CCICED (China Council for International Co-operation on Environment and Development), Summary report of China environmental protection and social development research group,

EP

http://www.cciced.net/zcyj/yjbg/zcyjbg/2013/201607/P020160708393872731595.pdf; 2013. [9] Chen Y, Ebenstein A, Greenstone M, et al. Evidence on the impact of sustained exposure to

AC C

air pollution on life expectancy from China's Huai River policy. Proceedings of the National Academy of Sciences of the United States of America 2013; 110(32):12936.

[10] Chen J, Zhao G, Pietikainen M. Health Impact Assessment of Beijing's Residents in Exposure of Chief Air Pollutants from 2010 to 2015 Based on Energy Consumption Scenarios. Procedia Environmental Sciences 2013; 18 (3): 277-282.

[11] Charafeddine R, Boden L I. Does income inequality modify the association between air pollution and health ? Environmental Research 2008; 106(1): 81-88. [12] Coneus K, Spiess CK. Pollution exposure and child health: Evidence for infants and toddlers in Germany. Journal of Health Economics 2012; 31: 180-196.

26

ACCEPTED MANUSCRIPT [13] Dangerman A T C J, Hans Joachim S. Energy systems transformation. Proceedings of the National Academy of Sciences of the United States of America 2013; 110 (7): 549-58. [14] Dasgupta P. Valuing health damages from water pollution in urban Delhi, India: a health production function approach. Environment and development economics 2004; 9: 83-106. [15] Deaton A. Health, Inequality, and Economic Development. Journal of Economic Literature

RI PT

2003; 41(1):113-158. [16] Dzikuć M, Piwowar A. Ecological and economic aspects of electric energy production using the biomass co-firing method: The case of Poland. Renewable & Sustainable Energy Reviews 2016; 55: 856-862.

SC

[17] Easterlin R A, Morgan R, Switek M, et al. China's life satisfaction, 1990-2010. Proceedings of the National Academy of Sciences of the United States of America 2012;

M AN U

109(25):9775-9780.

[18] Ezzati M, Kammen D M. Evaluating the health benefits of transitions in household energy technologies in Kenya. Energy Policy 2002; 30 (10): 815-826.

[19] Grossman M. On the concept of health capital and the demand for health. Journal of Political economy 1972; 80: 223-255.

[20] Gunatilaka J K R. The Rural-Urban Divide in China: Income but Not Happiness? Journal of

TE D

Development Studies 2010; 46(3):506-534.

[21] He B, Hong X. Review of calculation and decomposition methods of Gene coefficients. Statistics and Decision 2016 (14): 13-17. (in Chinese)

EP

[22] Jennifer Holdaway. Environment and Health in China: an introduction to an emerging research field. Journal of Contemporary China 2010; 19(63): 1-22. [23] Jiménez A M, Torrecilla R C, Domínguez M J V, et al. Assessing environmental justice

AC C

through potential exposure to air pollution: A socio-spatial analysis in Madrid and Barcelona, Spain. Geoforum 2016; 69: 117-131.

[24] Ji S, Cherry C R, Zhou W, et al. Environmental Justice Aspects of Exposure to PM2.5 Emissions from Electric Vehicle Use in China. Environmental Science & Technology 2015; 49 (24):13912-13920. [25] Jin Y, Li H, Wu B. Income inequality, consumption, and social-status seeking. Journal of Comparative Economics 2011; 39(2): 191-204. [26] Knight J, Gunatilaka R. Does economic growth raise happiness in China? Oxford Development Studies 2011; 39(1):1-24.

27

ACCEPTED MANUSCRIPT [27] John A, Pecchenino R, Schimmelpfennig D, et al. Short-lived agents and the long-lived environment. Journal of Public Economics 1995; 58(1): 127-141. [28] Kravitz-Wirtz N, Crowder K, Hajat A, et al. The long-term dynamics of racial/ethnic inequality in neighborhood air pollution exposure, 1990-2009. Du Bois Review Social Science Research on Race 2016; 13 (2): 237-259.

RI PT

[29] Kumar S, Rao D. Valuing The Benefits of Air Pollution Abatement Using a Health Production Function A Case Study of Panipat Thermal Power Station, India. Environmental and Resource Economics 2001; 20: 91-102.

[30] Li Y, Xu S. A Study on Residents Income Gap and Social Welfare in the Western Region of

SC

China. East China Economic Management 2017; 31(10): 70-76. (in Chinese)

[31] Matus, K., Nam, K.-M., Selin, N.E., Lamsal, L.N., Reilly, J.M., Paltsev, S. Health damages

M AN U

from air pollution in China. Global Environmental Change 2012; 22: 55-66. [32] Mestl H E S, Aunan K, Seip H M, et al. Urban and rural exposure to indoor air pollution from domestic biomass and coal burning across China. Science of the Total Environment 2007; 377(1):12.

[33] Milner J, Davies M, Wilkinson P. Urban energy, carbon management (low carbon cities) and co-benefits for human health. Current Opinion in Environmental Sustainability 2012; 4(4):

TE D

398-404.

[34] Niu S, Liu Y, Ding Y, et al. China’s energy systems transformation and emissions peak. Renewable & Sustainable Energy Reviews 2016; 58:782-795. [35] Org Z. Health, Inequality, and Economic Development. Social Science Electronic Publishing

EP

2003; 3: 113-158.

[36] Onater-Isberk E. Environmental Kuznets curve under noncarbohydrate energy. Renewable &

AC C

Sustainable Energy Reviews 2016; 64: 338-347.

[37] Pope, C.A., Thun, M.J., Namboodiri, M.M., Dockery, D.W., Evans, J.S., Speizer, F.E., Heath, C.W. Particulate air pollution as a predictor of mortality in a prospective study of US adults. American journal of respiratory and critical care medicine 1995; 151: 669-674.

[38] Pope, C.A., Burnett, R.T., Thurston, G.D., Thun, M.J., Calle, E.E., Krewski, D., Godleski, J.J. Cardiovascular mortality and long-term exposure to particulate air pollution. Circulation 2004; 109: 71-77. [39] Pan X, Yue W, He K, et al. Health benefit evaluation of the energy use scenarios in Beijing, China. Science of the Total Environment 2007; 374(2-3): 242.

28

ACCEPTED MANUSCRIPT [40] Qi Y, Lu H. Pollution, health and inequality -- a leap across the trap of "environmental health poverty". Management Word 2015; 9:32-51. (in Chinese) [41] Qi Y, Lu H. ‘Environmental poverty trap’ mechanism and China environmental knee. China Population Resources and Environment 2015; 10: 71-78. (in Chinese) [42] Qi L, Li Z. The liquidity of using income related health and medical services. Economic

RI PT

Research Journal 2011; 9: 83-95. [43] Qi L. Income, income inequality and health: the impact of urban and rural differences and occupational status. Economic Research Journal 2006; 11: 16-26.

[44] Rao K, Yao L. Methods of measuring health equity. Chinese Health Economics 1998; 12:

SC

30-32. (in Chinese)

[45] Raudenbush S W, Bryk A S. Hierarchical linear models: Applications and data analysis

M AN U

methods. Sage Publications; 2002.

[46] Schoolman E D, Ma C. Migration, class and environmental inequality: Exposure to pollution in China's Jiangsu Province. Ecological Economics 2012; 75 (2): 140-151. [47] Spalding-Fecher R. Health benefits of electrification in developing countries: a quantitative assessment in South Africa. Energy for Sustainable Development 2005; 9(1): 53-62. [48] Subramanian S V, Kawachi I. Whose health is affected by income inequality? A multilevel

TE D

interaction analysis of contemporaneous and lagged effects of state income inequality on individual self-rated health in the United States. Health & Place 2006; 12(2):141. [49] Wang Y, Han R, Kubota J. Is there an Environmental Kuznets Curve for SO2 emissions? A

EP

semi-parametric panel data analysis for China. Renewable & Sustainable Energy Reviews 2016; 54: 1182-1188.

[50] Wilkinson R G, Pickett K E. Income inequality and population health: a review and

AC C

explanation of the evidence. Social Science & Medicine 2006; 62 (7): 1768.

[51] Xepapadeas A. Economic development and environmental pollution: traps and growth. Structural Change & Economic Dynamics 1997; 8 (3): 327-350.

[52] Xie E. Research of income related health and medical services utilization. Economic Research Journal 2009; (2): 92-105. (in Chinese) [53] Yang J, Xu J, Wu X. Economic growth and health costs of environment and society. Economic Research Journal 2013; 12: 17-29. (in Chinese) [54] Zhang L, Lei L, Guo B. Applied multilevel data analysis. Educational Science Publishing House; 2005. (in Chinese) 29

ACCEPTED MANUSCRIPT [55] Zhang X, Zhang X, Chen X. Happiness in the Air: How Does a Dirty Sky Affect Mental Health and Subjective Well-being? Journal of Environmental Economics & Management 2017; 81-94. [56] Zhang Y, Wang S, Shang K et al. Influence of air pollutants on outpatient numbers of respiratory diseases in Beijing. Annual meeting of China Meteorological Society,

RI PT

2014:2401-2407. (in Chinese) [57] Zhang D. The mechanism of middle-income trap and the potential factors influencing China's

AC C

EP

TE D

M AN U

SC

economic growth. Economic Research Journal 2013; 9:17-29. (in Chinese)

30