Subjective Well-being and Environmental Quality: The Impact of Air Pollution and Green Coverage in China

Ecological Economics 153 (2018) 124–138

Contents lists available at ScienceDirect

Ecological Economics journal homepage: www.elsevier.com/locate/ecolecon

Analysis

Subjective Well-being and Environmental Quality: The Impact of Air Pollution and Green Coverage in China Liang Yuana,1, Kongjoo Shinb,c,

⁎,1

T

, Shunsuke Managib

a

Graduate School of Engineering, Kyushu University, Japan Department of Urban Engineering, School of Engineering, Kyushu University, Japan c Urban Institute, School of Engineering, Kyushu University, Japan b

A R T I C LE I N FO

A B S T R A C T

Keywords: Air pollution Air quality index Subjective well-being Life satisfaction Green coverage Subjective health evaluation China

Rapid environmental degradation is a well-publicized issue, particularly in rapidly developing countries. This study examines the impact of air pollution and green coverage on people's subjective well-being (SWB) in China using self-reported life satisfaction (LS) from survey data combined with the city-level air quality index (AQI) and green coverage data. The results show that air pollution and green coverage are significantly negatively and positively correlated with LS, respectively. The total effect of green coverage on life satisfaction constitute of a direct effect of green space itself and indirect effects through improving air pollution and health. The implicit monetary valuations of a 1-unit reduction in the AQI and a 1% increase in green coverage according to the respondent's annual gross individual income are approximately 239–280 USD (1.7%–2.0%) and 420–444 USD (3.0%–3.2%), respectively. The results also indicate that the average benefit from a 1% change in green coverage for people with a poor subjective health evaluation is almost 2 times higher than that for their counterparts.

1. Introduction Rapid economic growth and environmental degradation are environmental challenges. In order to assess the impact of environmental issues, non-market evaluations, conventional methods, and the life satisfaction (LS) approach that are also known as experienced preference methods, have received increased attention in the environmental economics literature (Welsch, 2007; Welsch, 2009; Welsch and Ferreira, 2014), and often have been used in environmental evaluations (Frey et al., 2010; MacKerron, 2012; Welsch and Ferreira, 2014). The LS approach has been used to investigate the environmental determinants of subjective well-being (SWB), which refers to an overall evaluation of one's life and is broader than “happiness” (Helliwell and Putnam, 2012; Welsch and Kühling, 2009). Previous studies have shown that the environmental surroundings, including biodiversity (Ambrey and Fleming, 2014b), noise (van Praag and Baarsma, 2005), water pollution (Israel and Levinson, 2003), air pollution (Smyth et al., 2011), environmentally friendly goods (Welsch, 2009) and green areas (Ambrey and Fleming, 2014c) influence residents, businesses, governments, and ecosystems.

In China, environmental degradation is a crucial issue given its wellknown severity. In the Environmental Performance Index: 2016 report published by Yale University, the air quality score for China ranked second to last (see “Global Metrics for the Environment” (2016)). According to the World Bank report, 16 of the world's 20 most polluted cities are in China (World Bank, 2006). In 2015, only 21.6% of Chinese cities met the Ambient Air Quality Standards (GB3095-2012), which is set by the Chinese Ministry of Environmental Protection (MEP). The Chinese president, Jinping Xi,2 said that “Air quality has directly affected the Chinese people's happiness” in an official statement in 2014. The air pollutants in China diffuse across the country's border and affect neighboring countries. Hence, the air pollution in China is both a domestic and international problem. Moreover, the demands for urban greenery to abate environmental problems have increased with an increasing awareness of various environmental problems (Cai et al., 2002; Chen and Jim, 2008a, 2008b; Jim and Chen, 2009). Introducing urban forests in China has been discussed since the 1990s; however, green coverage has only recently received attention in the public area (Li et al., 2005). China's State Council has addressed the importance of urban greening and the recent

⁎

Corresponding author at: 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan. E-mail address: [email protected] (K. Shin). 1 Co-authors contributed equally this work. 2 The report is available at http://www.gov.cn/xinwen/2914-03/07/content_2632820.htm. https://doi.org/10.1016/j.ecolecon.2018.04.033 Received 20 September 2017; Received in revised form 16 April 2018; Accepted 28 April 2018 0921-8009/ © 2018 Elsevier B.V. All rights reserved.

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

plans for the world's first pollution-eating “forest city” is a part of the government's efforts to address the negative externalities of rapid urbanization (Yu and Padua, 2007). Previous studies have extensively examined the impact of air pollution on SWB in China (Li et al., 2014; Smyth et al., 2008; Smyth et al., 2011; Zhang et al., 2017a, 2017b). However, previous researches have not examined the relationship between green coverage and SWB in China, combined with air pollution. Ambrey et al.'s (2014) recent analysis uses data from South East Queensland, Australia to examine the relationship between air pollution, green coverage and SWB. Nevertheless, the authors used the green coverage as a control variable and focused their discussion on the relationship between air pollution and SWB. This study contributes to the literature by providing additional evidence of the impact of air pollution and urban green coverage on LS using recent Chinese data, and provides detailed assessment of the relationship between environment and SWB. The analysis examines the possible interaction between air pollution and green coverage (i.e., green coverage may partially increase LS through reducing air pollution). In addition, we also consider the interaction effect of environmental determinants and health status on respondents' LS. We use originally collected data from a social survey that was conducted in 2016 and includes 281 major Chinese cities. To the best of our knowledge, this is the most recent SWB survey that was conducted in China with relatively complete coverage, especially at the city level. We use the air quality index (AQI) as the pollution measure (Zhang et al., 2017a, 2017b), which is calculated based on a health risk assessment that is associated with six major pollutants and is used by many countries to communicate with the public (Ferreira et al., 2013; MacKerron et al., 2009). To provide an extensive analysis on the relationships among air quality, green coverage and SWB, we also assess the interaction effects of pollution and green coverage as well as the interaction effects of both of these variables with the respondent's health condition. In addition, we provide implicit willingness-to-pay estimates for air quality and green coverage using the results from a regression analysis and compute the marginal rate of substitution of individual gross equivalised household income for air quality improvement as well as expanded green coverage. The remainder of this paper is organized as follows. Section 2 provides background and hypotheses and Section 3 provides a description of the data. Section 4 describes the estimation model. Section 5 provides the empirical results and discussion. Section 6 presents the conclusion.

1989),3 and some urban dwellers have ambivalent attitudes toward the urban green spaces (e.g., Bonnes et al., 2011; Carrus et al., 2004). In addition to reduce the negative sensorial perception, harmful ultraviolet radiation and noise, urban forests and trees also act as excellent filters of air pollution (Nowak et al., 2014). Urban green vegetation can absorb carbon and air pollutants that originate from three primary processes: wet deposition (e.g., the transfer of pollutants by falling rain/snow), chemical reactions (e.g., gas phase reactions in the atmosphere), and dry deposition (e.g., the transfer of gaseous and particulate pollutants to several surfaces, including trees) (Rasmussen et al., 1975). Given the challenges of a central government in imposing “top down” directives to reduce air pollution or to resolve other environmental problems (Economy, 2006), individual/community efforts to increase urban greening are highlighted as an effective measure to improve air quality (Manes et al., 2014; Bottalico et al., 2016a, 2016b). Moreover, the results of previous related studies suggest that air pollution may have a negative effect on LS through deteriorating the health (Mabahwi et al., 2014; Zhang et al., 2017a, 2017b); green space may have a positive influence on LS through health, which is a positive determinant of SWB (van den Berg et al., 2016). Also, previous studies found that the exposures to pollutants are associated with increases in mortality and hospital admissions due to respiratory and cardiovascular disease (Pope and Dockery, 2006; Brunekreef and Holgate, 2002). On the other hand, viewing nature through a window (Ulrich, 1984; Honold et al., 2014), living in environments with high share of green spaces (Maas et al., 2006) and having access to nearby green areas and parks (Cohen-Cline et al., 2015) are positively associated with the residents' health. Based on the abovementioned previous results, this study examines four hypotheses regarding the direct, indirect and interaction effects of AQI, green coverage and subjective health status on individuals' wellbeing.

2. Background and Hypotheses

3.1. Survey

The impact of environmental degradation, e.g., air pollution, is an important policy issue and research area in the literature (MacKerron and Mourato, 2009; Ambrey et al., 2014; Cuñado and de Gracia, 2013; Li et al., 2014; Zhang et al., 2017a, 2017b). Overall, the empirical evidence indicates that air pollutants, including NO2 and Pb (Welsch, 2006), SO2 (Ferreira et al., 2013; Luechinger, 2010), PM10 (Levinson, 2012) and PM2.5 levels (Du et al., 2018), has a significant negative impact on people's SWB. However, the magnitude of the impact varies by the specific pollutant type and the geographical region in each study. In contrast, green coverage has relatively limited evidence for its relationship with SWB. Previous studies have found that the biodiversity increases the psychological benefits that are associated with the “green” experience (Fuller et al., 2007); more access to greenspace is associated with higher levels of life satisfaction (Ambrey and Fleming, 2012; Ambrey and Fleming, 2014c; Carrus et al., 2015; Fleming et al., 2016; Krekel et al., 2016). On the other hand, some studies have found building green areas can induce negative feelings among the users (e.g., Gobster, 1994; Henwood and Pidgeon, 2001; Nassauer, 1995; Thayer,

This study uses a web-based social survey that was conducted during January and February 2016 in China; the respondents were recruited by posting the survey link on several websites, and the payment amount varied depending on the expected targeted respondent for each website. We constructed the matrix of age group (20s, 30s, 40s, and over 50s) and gender that reflected the general population and continued to recruit and collect responses until each cell was complete. Internet surveys can prevent interviewer biases that are caused by arbitrary factors, such as the appearance or gender of interviewer (Welsch

1. Air pollution has a negative impact on LS; green coverage has a positive impact on LS. 2. Green coverage partially increases LS through reducing air pollution. 3. Part of the effect of air pollution and green coverage on LS occurs through deteriorating and ameliorating subjective health status, respectively. 4. Poor subjective health status aggravates the effect of air pollution and enhances the effect of green coverage on LS. 3. Data and Variables

3 Henwood and Pidgeon (2001) reported that some people have ‘modern’ feelings and fears (such as fears of wildness, threats of darkness and ancient cultural meanings) toward trees and forests. Both Nassauer (1995) and Thayer (1989) argued that “the appearance of natural habitats transgresses American cultural norms for the neat appearance of landscapes. In this social context, natural ecosystems may be viewed as messy and untended.” Additionally, through his examination of children, Gobster (1994) observed that the spatial configurations that are preferred by humans “are consistent with the visual characteristics of vegetation of poor ecological quality.”

125

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

current pollution level and used to provide forecast information. The AQI aggregates the pollution levels of six air pollutants, and Table A2 shows the relationship between the AQI and the corresponding concentration limits of pollutants (see the “Technical Regulation on Ambient Air Quality Index” (HJ633-2012)). The magnitude of the AQI value corresponds with the level of health risk that is associated with local air quality in China. A higher AQI indicates poor air quality and a higher risk for inhabitants. As shown in Table A3, AQIs have health risk categories, but the ranges of pollution levels vary for each category. We use short-term and long-term averages of AQI. The short-term AQI is the average AQI for 3 months prior (November 2015 to January 2016) to the survey period for each respondents' resident cities, and the long-term AQI is the average AQI for one year (12 months) prior to the survey period in 2015. The analysis uses two AQI indicators for several reasons. The survey follows the popularly used framing of SWB questions and asks the respondents to report their life satisfaction for “these days.” Given the vague temporal implications of LS measure, we use the AQI measures that have a different temporal specification to identify and check the robustness of the relationship between SWB and air pollution. In addition, there is seasonality in air pollution level, especially in China; the air pollution is worse in the winter and spring (Guo and Chen, 2009; Liu et al., 2018). Since our survey was conducted during the winter, the short-term AQI reflects respondents' experience of air quality at the time of the survey, while the long-term AQI reflects respondents' general sentiment about air quality for residing in a specific city over a longer time period. Figs. A3 and A4 show the AQI distribution for the 281 Chinese cities. The short-term average AQI indicates higher overall air pollution across regions compared to the long-term AQI. For the geographical variation in the AQI, the northeastern and southern regions of China have relatively good air quality, while the eastern and central regions, including the Hebei, Shanxi, Henan and Shandong provinces, have relatively high levels of air pollution.

and Kühling, 2009). The survey was designed to collect self-reported life satisfaction as well as several personal and household characteristics: a total of 21,312 observations were recorded from 283 of the 293 Chinese prefecture-level cities. We eliminated the samples of respondents without household income and/or available air quality data in their residing cities, and 18,441 samples from 281 cities were retained for the analysis. Popular datasets that are used by other studies in the area of subjective well-being include the Chinese General Social Survey (CGSS), Chinese Household Income Project (CHIP) and China Family Panel Studies (CFPS). To the best of our knowledge, this dataset is the most recent and comprises the broadest coverage of city-scale social survey data in China that includes subjective well-being questions. The latest CGSS 2013 surveyed 11,438 respondents, and the latest CHIP 2013 surveyed 1948 households with 64,777 individuals in 126 cities. The latest CFPS 2014 consists of 13,946 sample households. We eliminated samples of respondents who did not provide gross equivalised household income and/or air quality data. A total of 18,441 respondents from 281 cities were retained for the analysis. Table A1 provides a comparison between the demographic distribution of our respondents and other available sources of social demographic statistics from the 2016 China Statistical Yearbook. We obtained variables that were not available in the statistical yearbook from CGSS 2013 and CHIP 2013 data (e.g., gross equivalised household income). The average annual gross equivalised household income of our respondents is higher than in the CGSS but lower than in the CHIP. Nonetheless, the more developed east coast region reported higher average LS scores than the central China, and there was large variation in the average LS scores among the inner regions of western China. CGSS included non-urban respondents; however, CHIP does not contain urban immigrants, who tend to be relatively poor. The respondents in our survey are skewed toward the urban population; consequently, our data does not fully represent rural areas compared to CGSS but provide a better overall representation of gross equivalised household income groups compared to the CHIP. The age and gender distributions of our survey data are consistent with the official statistics from the 2016 China Statistical Yearbook (National Bureau of Statistics of China, 2016).

3.4. Green Coverage Urban green coverage has a crucial role in abating the negative effects of air pollution (Roy et al., 2012), especially by mitigating the negative health impact on citizens that is caused by a broad range of atmospheric pollutants (Manes et al., 2014; Selmi et al., 2016). This study used city-level green coverage areas as the primary explanatory variable along with the AQI. Official statistics of green coverage below the city level are unavailable and it is difficult to obtain nationwide detailed satellite image data to distinguish between green space and residences at local levels. Nevertheless, previous studies have indicated that there are several recognizable characteristics of developing urban cities in China: roads that have limited interstitial spaces for greenery, high number of cities with no major green spaces, and major parks frequently located outside of the city centers (Jim, 2004; Tian et al., 2012; Sun et al., 2017). Hence, the variation among urban residents in the Chinese cities for green space access appears to be relatively. According to the “Urban Land Classification and Planning Land for Construction Standards” that were published in 1991 and the “Urban Greening Planning and Construction Indicators” that were published in 1993 by the Ministry of Construction of China, green coverage area is measured as the proportion of green areas relative to the total city area; green areas include public green space, residential green space and green spaces that are attached to building units, protected green land, productive plantation areas and scenic forest land (Carrus et al., 2015). The average proportion of the green coverage area in the sample cities was approximately 40% in 2013. Out of 281 cities, the share of green coverage area was < 30% in 29 cities, was between 30 and 40% in 110 cities, and 40% or above in 142 cities. The coverages of green areas range from 26 ha (Longnan) to 124,295 ha (Shanghai). Fig. A5 shows the distribution of the city-level green coverage areas in China.

3.2. Life Satisfaction Index Our dependent variable is an 11-point LS rating. We asked respondents “All things considered, how satisfied are you with your life as a whole these days?” and respondents were presented with options that ranged from 0 (completely dissatisfied) to 10 (completely satisfied). Fig. A1 shows the distribution of the LS ratings in our sample. Ratings of 7 and 8 were chosen by almost half of the respondents. The mean LS rating for the sample is 7.06 (SD = 1.81), which is consistent with other Chinese SWB data in recent analyses for all of China. For example, Asadullah et al. (2018) reported an average LS in 2010 of 3.77 based on a 1–5 scale, which is 7.54 based on an 11-point scale. Moreover, the Chinese SWB data used in Chen et al. (2015) had average urban LS scores of 4.88 in 2011, based on a 7-point scale, which is 6.97 on an 11point scale. Fig. A2 provides the geographical distributions of city average LS ratings; the apparent differences in LS ratings across the cities may be due to location-specific factors (Brereton et al., 2008). We do not observe clear geographical “clustering” of LS scores. Nonetheless, the relatively more developed east coast region has a higher LS compared to central region, and the variation of LS ratings among the inner regions of western China is relatively large. 3.3. Air Quality Index Air quality index (AQI) is used as a measure of air pollution. The AQI is used by the government agencies as an aggregate measure of the 126

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

According to MacKerron and Mourato (2009), the “OLS model provides similar results as the ordered probit” model, and is more straightforward than marginal effects across a large number of categories. Knight et al. (2009) and Du et al. (2018) also have used OLS model to examine the impacts of the determinants of LS. We used the estimation models in which LS was expressed as a function of: 1) environmental and subjective health (ES), which includes the air quality index (AQI), green coverage (GC), subjective health and interaction variables, 2) city characteristics (CC), and 3) personal characteristics (PC). We also controlled for the provincial dummy variables (Goetzke and Islam, 2017). ε is the error term that reflects the random factors. Air quality index (AQI), green coverage (GC), subjective health are the main explanatory variables in the estimation model. The following structural model (Eqs. (1)–(4)) examines the direct and indirect effects that are set out to be tested in the hypotheses presented in Section 2.

3.5. Other Control Variables We use several personal and household characteristics as control variables. We have constructed a subjective health evaluation dummy variable from a 5-point scale response to the question “All in all, how would you describe your state of health?” If a respondent chose poor subjective health status (options 3–5), then the dummy variable was coded 1 and was otherwise 0. Subjective health status adequately reflects the respondent's physical health condition with one question. However, it is crucial to note that the self-assessment may reflect respondents' mental health conditions (Au and Johnston, 2014). Hence, it is reasonable to view this variable as an aggregate of the respondents' general ‘vitality.’ In addition, we use the natural log of respondent's individual share of gross equivalised household income (hereinafter referred to as ‘individual income’) as an income measure. This variable was calculated based on Joung et al. (1997) and MacKerron and Mourato (2009). Annual gross individual income is the gross equivalised household income (from the survey data) divided by the household size equivalence factor (a + 0.7c)0.5, where ‘a’ is the number of adults and ‘c’ is the number of children in the household. The respondents selected the range of their gross equivalised household income and the available options included 1 = ‘0–6000 CNY’ to 21 = ‘340,000 CNY or above.’ We used the midpoint of the self-reported gross equivalised household income range as the gross equivalised household income measure (i.e., if the respondent selected 108,000–119,999 CNY, then that would be 114,000 CNY). We have also controlled for respondents' personal characteristics and attitudes toward socio-economic issues that are commonly used in SWB studies, such as trust, community attachment, activity participation, household income attitudes, age, marital status, gender, level of education, the situation of children and children age 6 or below in the household. It is important to denote possible self-selection in the survey participation and potential omitted variables that pose a threat to the validity of inferences drawn using observational data. Nevertheless, these potential sources of confounding are mitigated through covariate adjustment. We mitigate the abovementioned limitation by focusing on the urban population4 and by controlling for the city-level variables: the number of hospitals and health centers per capita, the ratio of consumption, investments in fixed assets and the number of public transportation vehicles. We also use the provincial dummy variables to control for the other potential unobserved provincial variations in climate, commodity prices, language, cultural traditions and the other geographical and socio-political characteristics. In the analysis, we use Beijing as the reference city. We used 2013 city-level objective data from the China City Statistical Yearbook 2014 (National Bureau of Statistics of China, 2014), which provides the most detailed official micro-level data that includes the green coverage information. Denote that using city level variables may result in the modifiable areal unit problem (MAUP) (Heywood et al., 1998), and may also undermine the individual level variation in the survey data. Thus, the accuracy and robustness of the analytical results can be improved with increased data availability by using more disaggregated and spatially referenced socio-demographic and environmental data in the future research. Table A4 provides the detailed description of the variables that are used in the analysis.

LS = f (AQI, GC, Health; X)

(1)

AQI = g (emissions, GC)

(2)

Health = h (AQI, GC)

(3)

X = I (CC, PC, provincial dummy variable)

(4)

We test the hypotheses with two sets of five regression models that include short-term AQI (M1–M5) and long-term AQI (M6–M10). M1/ M6 uses GC as an explanatory variable and does not include AQI and subjective health status; M2/M7 uses AQI as an explanatory variable and does not include GC and subjective health status; M3/M8 controls for both AQI and GC but does not include subjective health status; M4/ M9 controls AQI, GC and subjective health status; M5/M10 is the full model that includes AQI, GC, subjective health status and their binary interactions (interaction of AQI and subjective health status/interaction of GC and subjective health status). The interaction variables were centered to eliminate collinearity using the calculation (= variablei − variable) (See, Robinson and Schumacker (2009) for a detailed explanation as to why centering variables is required for examining interaction effects.) In addition, the age and age-squared variables were centered to eliminate collinearity. Furthermore, we used the natural log for all city-level control variables and per capita calculations for the city-level control variables. Correlated independent variables cause estimation bias due to multi-collinearity. To check the validity of our regression models, we examined the variable inflation factors (VIFs) for each model: the VIFs for all 10 models were < 7, which is lower than 10 (usually used as the rule of thumb), and lower than the VIFs of the models without centering or using the natural log of the variables. Thus, the centering and using natural log of some variables have reduced the total VIF of the models. Another way to assess the subjective effect of air pollution changes is through the monetary valuation (MV) between income and air pollution (Bayer et al., 2009). The MVs of calculated using the estimated coefficients indicate trade-off between annual gross individual income and both air pollution and green coverage, while holding individual LS constant. We use the delta method to estimate the standard error and confidence interval for the MV. For the models without interactions, we calculated the MV (Eq. (5)) based on Eqs. (1)–(4)

MV = −Incomei 4. Empirical Analysis

β1 β2

(5)

For the models with interactions, the equation (Eq. (6)) becomes

To analyze the relationships among air pollution, green coverage LS, we used ordinary least squares (OLS) as the estimation method.

MV = −Incomei

β1 + βi Ii β2

(6)

where Incomei is the respondent's individual income. β1 is the regression coefficient for the AQI or green coverage area, β2 is the regression coefficient of the log of the individual income, βi is the regression

4

The responses of rural residents were excluded from the analyses by checking their reported addresses and/or zip-codes. 127

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Table 1 Results from the OLS regressions; dep. variable: life satisfaction. M1

M2

M3

M4

M5

M6

AQI: short-term

M8

M9

M10

AQI: long-term

Without interaction

With interaction

Without interaction

With interaction

Only GC

Only AQI

AQI and GC

AQI, GC and health

AQI#health health#GC

Only GC

Only AQI

AQI and GC

AQI, GC and health

AQI#health health#GC

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

Coefficient

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

−0.0037⁎⁎⁎ (0.0008)

−0.0035⁎⁎⁎ (0.0008)

−0.0039⁎⁎⁎ (0.0008) −0.0057⁎⁎⁎ (0.0013)

−0.0044⁎⁎⁎ (0.0013) 0.3110⁎⁎⁎ (0.1000)

−0.0041⁎⁎⁎ (0.0013) 0.2770⁎⁎⁎ (0.0973) −0.9990⁎⁎⁎ (0.0288)

−0.0047⁎⁎⁎ (0.0013) 0.2380⁎⁎ (0.1030) −0.9870⁎⁎⁎ (0.0288) 0.0031⁎⁎⁎ (0.0012)

Environmental and subjective health (ES) AQI: short term −0.0044⁎⁎⁎ (0.0008) AQI: long term Log (green coverage)

M7

0.3970⁎⁎⁎ (0.0970)

0.2950⁎⁎⁎ (0.0997)

Poor subjective health evaluation Interaction between AQI and poor health Interaction between green coverage and poor health City characteristics (CC) Hospitals & HC per 0.0798⁎⁎ capita (0.0311) Log (consumption) 0.3110⁎⁎⁎ (0.1090) Investment per capita 0.0302⁎⁎⁎ (0.0115) 0.0010 Log (public (0.0333) transportation per capita) Personal characteristics (PC) Trust 0.6250⁎⁎⁎ (0.0409) Community 0.8100⁎⁎⁎ attachment (0.0258) Activity participation 0.0303 (0.0382) Household income 1.0130⁎⁎⁎ attitudes (0.0272) Log (individual 0.2160⁎⁎⁎ income) (0.0142) Age 0.0006 (0.0009) 2 Age −0.0001⁎ (0.0001) Female 0.1990⁎⁎⁎ (0.0232) Married 0.2090⁎⁎⁎ (0.0478) Children 0.1560⁎⁎⁎ (0.0460) Children under 6 −0.1140⁎⁎⁎ (0.0294) Own educational 0.0372 background (0.0291) Partner's educational 0.1620⁎⁎⁎ background (0.0264) Constant 0.0007 (0.6880) Observations 18,441 R-squared 0.2689 Adjusted R-squared 0.2670

0.2610⁎⁎⁎ (0.0966) −0.9990⁎⁎⁎ (0.0288)

0.2200⁎⁎ (0.0988) −0.9870⁎⁎⁎ (0.0288) 0.0002 (0.0009)

0.3970⁎⁎⁎ (0.0970)

1.3820⁎⁎⁎ (0.1510)

1.3070⁎⁎⁎ (0.1530)

0.0676⁎⁎ (0.0311) 0.1900⁎ (0.1100) 0.0286⁎⁎ (0.0113) 0.0425 (0.0337)

0.0732⁎⁎ (0.0311) 0.2260⁎⁎ (0.1100) 0.0283⁎⁎ (0.0115) 0.0287 (0.0339)

0.0698⁎⁎ (0.0301) 0.2240⁎⁎ (0.1070) 0.0312⁎⁎⁎ (0.0111) 0.0111 (0.0329)

0.0642⁎⁎ (0.0301) 0.2570⁎⁎ (0.1070) 0.0287⁎⁎⁎ (0.0111) 0.0145 (0.0329)

0.0798⁎⁎ (0.0311) 0.3110⁎⁎⁎ (0.1090) 0.0302⁎⁎⁎ (0.0115) 0.0010 (0.0333)

0.0706⁎⁎ (0.0311) 0.2180⁎⁎ (0.1100) 0.0346⁎⁎⁎ (0.0113) 0.0168 (0.0332)

0.0761⁎⁎ (0.0311) 0.2550⁎⁎ (0.1100) 0.0331⁎⁎⁎ (0.0115) 0.0063 (0.0334)

0.0726⁎⁎ (0.0301) 0.2520⁎⁎ (0.1070) 0.0357⁎⁎⁎ (0.0112) −0.0103 (0.0323)

0.0681⁎⁎ (0.0301) 0.2770⁎⁎⁎ (0.1070) 0.0343⁎⁎⁎ (0.0112) −0.0112 (0.0325)

0.6300⁎⁎⁎ (0.0408) 0.8110⁎⁎⁎ (0.0258) 0.0343 (0.0381) 1.0170⁎⁎⁎ (0.0272) 0.2170⁎⁎⁎ (0.0141) 0.0005 (0.0009) −0.0001⁎ (0.0001) 0.1990⁎⁎⁎ (0.0232) 0.2130⁎⁎⁎ (0.0477) 0.1420⁎⁎⁎ (0.0459) −0.1110⁎⁎⁎ (0.0294) 0.0426 (0.0291) 0.1660⁎⁎⁎ (0.0263) 2.4470⁎⁎⁎ (0.5800) 18,510 0.2693 0.2674

0.6240⁎⁎⁎ (0.0409) 0.8140⁎⁎⁎ (0.0258) 0.0284 (0.0382) 1.0150⁎⁎⁎ (0.0272) 0.2140⁎⁎⁎ (0.0141) 0.0005 (0.0009) −0.0001⁎ (0.0001) 0.2000⁎⁎⁎ (0.0232) 0.2100⁎⁎⁎ (0.0478) 0.1510⁎⁎⁎ (0.0460) −0.1120⁎⁎⁎ (0.0294) 0.0395 (0.0291) 0.1640⁎⁎⁎ (0.0264) 1.1300 (0.7350) 18,441 0.2697 0.2677

0.3020⁎⁎⁎ (0.0407) 0.6120⁎⁎⁎ (0.0257) 0.0363 (0.0370) 0.8680⁎⁎⁎ (0.0267) 0.2060⁎⁎⁎ (0.0137) 0.0002 (0.0009) −0.0002⁎⁎ (0.0001) 0.2210⁎⁎⁎ (0.0225) 0.2080⁎⁎⁎ (0.0463) 0.1820⁎⁎⁎ (0.0446) −0.1550⁎⁎⁎ (0.0285) 0.0442 (0.0282) 0.1170⁎⁎⁎ (0.0256) 2.3410⁎⁎⁎ (0.7130) 18,441 0.3144 0.3126

0.3170⁎⁎⁎ (0.0406) 0.6140⁎⁎⁎ (0.0257) 0.0391 (0.0369) 0.8710⁎⁎⁎ (0.0266) 0.2050⁎⁎⁎ (0.0137) 0.0001 (0.0009) −0.0002⁎⁎ (0.0001) 0.2170⁎⁎⁎ (0.0225) 0.2080⁎⁎⁎ (0.0462) 0.1900⁎⁎⁎ (0.0445) −0.1550⁎⁎⁎ (0.0285) 0.0476⁎ (0.0282) 0.1160⁎⁎⁎ (0.0255) 2.3680⁎⁎⁎ (0.7170) 18,441 0.3176 0.3156

0.6250⁎⁎⁎ (0.0409) 0.8100⁎⁎⁎ (0.0258) 0.0303 (0.0382) 1.0130⁎⁎⁎ (0.0272) 0.2160⁎⁎⁎ (0.0142) 0.0006 (0.0009) −0.0001⁎ (0.0001) 0.1990⁎⁎⁎ (0.0232) 0.2090⁎⁎⁎ (0.0478) 0.1560⁎⁎⁎ (0.0460) −0.1140⁎⁎⁎ (0.0294) 0.0372 (0.0291) 0.1620⁎⁎⁎ (0.0264) 0.0007 (0.6880) 18,441 0.2689 0.2670

0.6300⁎⁎⁎ (0.0408) 0.8100⁎⁎⁎ (0.0258) 0.0357 (0.0382) 1.0160⁎⁎⁎ (0.0272) 0.2180⁎⁎⁎ (0.0141) 0.0005 (0.0009) −0.0001⁎ (0.0001) 0.1990⁎⁎⁎ (0.0232) 0.2140⁎⁎⁎ (0.0477) 0.1430⁎⁎⁎ (0.0459) −0.1110⁎⁎⁎ (0.0294) 0.0425 (0.0291) 0.1670⁎⁎⁎ (0.0264) 2.5110⁎⁎⁎ (0.5950) 18,510 0.2690 0.2671

0.6240⁎⁎⁎ (0.0409) 0.8130⁎⁎⁎ (0.0258) 0.0296 (0.0382) 1.0140⁎⁎⁎ (0.0272) 0.2150⁎⁎⁎ (0.0141) 0.0005 (0.0009) −0.0001⁎ (0.0001) 0.2000⁎⁎⁎ (0.0232) 0.2110⁎⁎⁎ (0.0478) 0.1520⁎⁎⁎ (0.0460) −0.1130⁎⁎⁎ (0.0294) 0.0392 (0.0291) 0.1640⁎⁎⁎ (0.0264) 1.062 (0.7590) 18,441 0.2694 0.2674

0.3020⁎⁎⁎ (0.0407) 0.6110⁎⁎⁎ (0.0257) 0.0374 (0.0370) 0.8660⁎⁎⁎ (0.0267) 0.2070⁎⁎⁎ (0.0137) 0.0002 (0.0009) −0.0002⁎⁎ (0.0001) 0.2210⁎⁎⁎ (0.0225) 0.2080⁎⁎⁎ (0.0463) 0.1830⁎⁎⁎ (0.0446) −0.1550⁎⁎⁎ (0.0285) 0.0439 (0.0282) 0.1170⁎⁎⁎ (0.0256) 2.2620⁎⁎⁎ (0.7360) 18,441 0.3141 0.3123

0.3200⁎⁎⁎ (0.0406) 0.6140⁎⁎⁎ (0.0257) 0.0394 (0.0369) 0.8700⁎⁎⁎ (0.0266) 0.2050⁎⁎⁎ (0.0137) 0.0001 (0.0009) −0.0002⁎⁎ (0.0001) 0.2160⁎⁎⁎ (0.0225) 0.2100⁎⁎⁎ (0.0462) 0.1940⁎⁎⁎ (0.0445) −0.1570⁎⁎⁎ (0.0285) 0.0471⁎ (0.0282) 0.1150⁎⁎⁎ (0.0256) 2.3490⁎⁎⁎ (0.7600) 18,441 0.3175 0.3156

Notes: Standard errors in parentheses. Provincial dummy variables were controlled for in all models. ⁎⁎⁎ p < 0.01. ⁎⁎ p < 0.05. ⁎ p < 0.1. 128

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

and M4/M9 (0.2610/0.2770). Based on the magnitudes of coefficient differences and the standard errors, it is difficult to be definitive regarding the significance and robustness of the indirect effect of green coverage through subjective health evaluation. Based on the overall results regarding Hypothesis 2 and 3, we can draw the conclusion that the total effect of green coverage area on LS (0.3970 from M1/M6) is inclusive of its direct effect (0.2610/0.2770 from M4/M9) and indirect effects through reduction of negative impact from AQI and health risk, especially the reduction of negative impact from AQI. The green coverage improves individual's well-being not only with its direct positive impact but also through indirect effects of improving air quality and the residents' subjective health conditions. In terms of Hypothesis 4, the results of M5/M10 show significant positive coefficients of the interaction variables between the subjective health measure and green coverage; this result is generally consistent with the previous findings (Ambrey, 2016; van den Berg et al., 2010; van den Berg et al., 2015), and indicates that poor subjective health status attenuates the effect of the positive impact of green coverage on LS. However, the coefficient of interaction variable between short-term AQI and subjective poor health is not statistically significant, and the coefficient of the interaction variable between long-term average AQI and poor subjective health evaluations (M5 and M10) is statistically significant and positive. These results provide partial support for Hypothesis 4; poor subjective status enhances the effect of green coverage and long-term air pollution on LS but does not enhance the effect of short-term air pollution. The subjective health measure evaluates an individual's general vitality and may reflect both physical and mental health, and green coverage may alleviate the negative effect of poor physical and/or mental health conditions. The significant interaction effect between green coverage areas and poor subjective health status on LS (M5 and M10) indicates that people with low subjective health would benefit more from accessible green spaces compared to the individuals with relatively high subjective health. Providing empirical evidence for the mechanisms behind the positive sign of interaction between long-term air pollution measure and poor subjective health is beyond the scope of this paper but the future studies could further discuss and provide insights to the possible mechanisms that explain the result. Also, given that the absolute value of negative coefficient for the long-term AQI is larger than the absolute value of positive coefficient for the interaction of long-term air pollution and poor health status, the net average impact of AQI on LS remains negative for the respondents with poor subjective health.

coefficient for the interaction variable, Ii represents the variable that interacts with the target variable (AQI or GC). βiIi is included when the interaction variable is statistically significant. We acknowledge the limitation in terms of the accuracy of the income measure used to calculate the MV; the related previous studies have also often mentioned this issue as problematic and difficult to resolve (see Ambrey and Fleming (2014a) for further details). 5. Results and Discussion 5.1. The Impact of the AQI and Green Coverage The results in Table 1 indicate that air pollution is statistically significantly and negatively correlated with LS, which is consistent with previous studies on poor air quality and SWB in Chinese populations (Smyth et al., 2008; Du et al., 2018). This result also shows a statistically significant and positive relationship between green areas and respondents' LS, which supports the empirical evidence from previous studies, i.e., the positive role of forests and green spaces in self-reported well-being of the residents in urban areas (Carrus et al., 2015; Tsurumi and Managi, 2015). The statistical significance and direction of the AQI and green coverage coefficients are consistent across all models provide strong support for Hypothesis 1. As for the Hypothesis 2, the evidence indicates that green coverage partially increases LS by reducing air pollution. The green coverage coefficient in M1/M6 (0.3970/0.3970) is larger than in M3/M8 (0.2950/0.3110) (controlling for both AQI and green coverage), and the AQI coefficient in M3/M8 (−0.0037/−0.0044) is less negative than in M2/M7 (−0.0044/−0.0057). The differences in the green coverage and AQI coefficients between models indicate that green coverage increases LS through reducing air pollution. This finding suggests that there is a potential overestimation of the negative impact of air pollution on LS provided by previous research that has discussed pollution effects without accounting for green coverage. The results also show that the long-term AQI has a stronger negative correlation with LS than the short-term AQI. According to M3 (without controlling for subjective health), 1-point increases in a city's short- and long-term AQI levels are associated with 0.0037 and 0.0044 points reductions in residents' average LS. Given the sizes of theses coefficients, the impact of pollution reduction on LS seems to be rather marginal. Three-month and annual average AQI values in the Chinese cities are approximately 103 and 91, air pollution results in a reduction of approximately −0.28 to −0.41 on the 11-point LS scale. Since the upper limit of the “excellent” levels of pollution as suggested in the official guidelines of the Chinese Ministry of Environmental Protection (see the “Ambient Air Quality Standard” (GB 3095-2012)) is an AQI = 50, if the average air pollution level was reduced to obtain the upper limit in the “excellent” category, the average LS for short-term air pollution would improve more (0.20 points) than long-term air pollution (0.18 points). Given that marriage and having a child positively affects LS by approximately 0.21 and 0.17 points, respectively, the negative impact of air pollution on respondent's well-being seems to be relatively substantial.

5.3. Monetary Valuation for the AQI and Green Coverage Table 2 shows the implicit monetary valuation that was calculated using the AQI and green coverage variables' coefficients in M4, M5, M9 and M10 that are estimated following the Eqs. (5) and (6). The average USD values (the proportion of the monetary value relative to annual gross individual income, expressed as a %) for a 1-unit reduction in the short- and long-term AQI corresponds to 239 USD (1.70% of the annual gross individual income) and 280 USD (2.00% of the annual gross individual income), respectively. The monetary value of a 1% increase in green coverage is much higher than that of the AQI, with 420 USD (3.00% of the annual gross individual income) and 444 USD (3.17% of the annual gross individual income), depending on the specification of the AQI variables. The impact of the short-term AQI is not affected by the subjective evaluation given that the interaction effect is not significant. Moreover, the average benefit of a 1% decrease in air pollution for people who have a poor subjective health evaluation is similar to the average monetary value, in which a 1% increase in green coverage for individuals who have a poor subjective health evaluation is approximately 2 times higher than that for the average monetary value. Given the variation in the functioning of ecosystems across the

5.2. AQI, Green Coverage and Subjective Health Evaluations The results for M4-M5, M9-M10 indicate that a poor subjective health evaluation decreases LS by approximately 1 point on a 5-point scale. Nevertheless, controlling for subjective health status does not significantly change the overall impact of AQI on LS. The AQI coefficients in M4/M9 (0.0035/0.0041) do not differ much from the coefficients in M3/M8 (0.0037/0.0044). Hence, we do not find strong evidence to support the Hypothesis 3 with respect to air pollution that the air pollution affects LS partially through reduction of people's selfhealth evaluation. On the other hand, the coefficients of green coverage differ by approximately 0.03 unit between M3/M8 (0.2970/0.3110) 129

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Table 2 Monetary valuation for the AQI and green coverage. Mean ± std. err. (95% conf. interval)

AQI

Without the interaction (M4/M9)

With the interaction (M5/M10)

Green coverage

Without the interaction (M4/M9)

With the interaction (M5/M10)

USD CNY Percentage USD CNY Percentage USD CNY Percentage USD CNY Percentage

Short-term

Long-term

−239 ± 58 (125–352) −1551 ± 378 (810–2291) 1.70 ± 0.42 (0.89–2.52) /

−280 ± 89 (149–455) −1818 ± 580 (681–2955) 2.00 ± 0.64 (0.75–3.25) −266 ± 89 (86–447) −1731 ± 599 (558–2905) 1.90 ± 0.66 (0.61–3.19) 444 ± 159 (133–755) 2884 ± 1031 (863–4905) 3.17 ± 1.13 (0.95–5.39) 903 ± 183 (545–1262) 5872 ± 1190 (3540–8203) 6.45 ± 1.31 (3.89–9.02)

420 ± 158 (111–729) 2730 ± 1026 (719–4741) 3.00 ± 1.13 (0.79–5.21) 905 ± 178 (556–1254) 5885 ± 1157 (3617–8152) 6.47 ± 1.27 (3.98–8.96)

Notes: 1. Groups are classified according to the corresponding dummy variable. 2. The number is the average amount that respondents are willing to pay for a 1-unit reduction in the average AQI or a 1% increase in green coverage. 3. Expressed as USD and CNY (percentage of the individual gross income). Exchange rate: 1 USD = 6.5 CNY.

Although the actual value of the urban green space includes the value of reducing air pollution and increasing the LS levels, green areas have a strong influence on pollution policy and SWB, especially for residents who have poor health self-evaluations. However, increasing green coverage in urban areas to maximize pollution mitigation may not necessarily be the implemented policy, as there may be other goals, such as improving air quality or health conditions, which may require different budgets or measures for improving LS through mitigating air pollution. Although air quality improvements from green roofs and walls may be relatively small, this should be expanded in future green infrastructure scenarios to improve air pollution (Jayasooriya et al., 2017).

cities, it is difficult to calculate the relative average value of air pollution mitigation by urban green spaces in all cities. Nonetheless, the case of Beijing can provide useful insights into the estimated valuation. As of 2013, the city had a registered population of 13.16 million, and 64,137 ha of urban greens covered 51% of the area (National Bureau of Statistics of China, 2014). Using the estimates for the long-term pollution levels, the total MVs for a 1-unit reduction in air pollution and a 1% increase in green coverage in Beijing are approximately 23.92 billion CNY (3.68 billion USD) and 37.95 billion CNY (5.83 billion USD), respectively. To obtain a 1-point reduction in the AQI level in Beijing, the government must spend approximately 2.29 billion CNY.5 In comparison, a 1% increase in green coverage would have cost the government approximately 3.03 billion CNY in 2013 and 2014 (Beijing Municipal Statistics Bureau, 2015). Green areas can improve LS by removing pollutants (Wu and Dong, 2014). Hence, if respondents are not fully aware of the pollution reduction effects of urban greens, then the value of urban green spaces may be underestimated. According to Yang et al. (2004), the total annual pollutant removal effect for one square meter of green space in Beijing is approximately 4.19 g.6 Applying the estimated reduction effect to the 2014 air pollution level (National Bureau of Statistics of China, 2015), a 1% increase in green coverage would have reduction impacts of 0.09% for the NO2 and 0.13% for the SO2 levels. Based on previous references, the monetary value is 600 CNY/mg for both NO2 (State Environmental Protection Agency, 2004; Chunju et al., 2004) and SO2 (Ouyang et al., 1999; Xiao et al., 2000). Based on the annual pollutant removal per square meter of tree cover that was stated above, the service value of tree cover year−1 for reducing NO2 and SO2 is 0.26 CNY/m2 and 0.38 CNY/m2, respectively. This indicates that the value of a 1% increase in the green area for reducing NO2 and SO2 emissions in Beijing could reach 0.59 million and 0.85 million CNY/year, respectively; 0.063% of the total Beijing government's budget would be needed to reduce existing air pollution by 1%.

5.4. City and Personal Characteristics The variables of city characteristics, the number of hospitals and health centers, and consumption and fixed asset investments, which include real estate developments and public infrastructures, have positive and significant impacts on LS. As noted in the previous section, the city-level data and provincial dummy variables may induce the modifiable area unit problem (MAUP). Hence, we should be cautious when interpreting the coefficients and robustness of the city-level variables. Nonetheless, according to the results that were presented in Table A5, the model without the provincial dummy variables suffers from omitted variable bias and leads to an underestimation of the negative impacts of air pollution. Therefore, it is crucial to include available data to avoid other types of biases. The results also indicate a significant impact of personal characteristics variables, which are generally consistent with previous findings (e.g., Ambrey et al., 2017; Blanchflower and Oswald, 2008; Dolan et al., 2008). The following variables positively impact LS: the importance of trust, community attachment, annual gross individual income, the educational attainment of respondents and their partners, satisfaction with gross equivalised household income and being female. These results are consistent with previous studies on happiness in urban China (Appleton and Song, 2008). We found an inverse U-shaped relationship between age and SWB, which differs from the commonly found U-shaped relationship between age and SWB in China in previous studies (Smyth et al., 2008; Du et al., 2018). Moreover, the result confirms the overall positive impact of children in a household on LS that was found in previous results (Spector et al., 2004) as well as the negative impact of having young children on LS (Manning et al., 2016; Shields and Wooden, 2003). These results may reflect the high stress environment of Chinese urban residents, especially the young couples with young children.

According to the “Measures for Evaluating the Implementation of the Air Pollution Prevention and Control Action Plan”, the Beijing-Tianjin-Hebei region is expected to spend 249.03 billion CNY to reduce the existing pollution level by 25%; in other words, a decrease of 1% would cost 9.96 billion CNY. Based on the fixed asset investment ratios for each city (National Bureau of Statistics of China, 2015), Beijing accounts for 23.56%. 6 The removal effect was 0.44 g/m2 of tree cover year−1 for NO2; 0.85 g/m2 of tree cover year−1 for O3; 0.33 g/m2 of tree cover year−1 for SO2 and 2.56 g/ m2 of tree cover year−1 for PM10. In contrast, the total pollutant removal was 5.89 g/m2 of tree cover year−1 for developed countries (Selmi et al., 2016). 5

130

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

6. Conclusion

This study has several limitations that can be addressed in future studies to improve our understanding of the relationship between environmental quality and individuals' subjective well-being. While some recent studies examined the impact of green spaces on SWB using disaggregated green types and the variability of their impacts (Krekel et al., 2016) but these studies tend to cover a limited area. Our data set has sample size that is comparable to that of other large social surveys that are conducted in China and covers broad area, and is difficult to obtain or construct disaggregated urban green data. Further disaggregation of green types and a more detailed spatial analysis would be possible when the detailed satellite images for the specific time period are made available along with the timely surveyed data. Also, collecting samples from older age groups tends to be expensive and time consuming but the efforts to improve the representativeness of the data would improve the precision of the empirical results. Moreover, in addition to resolving the above limitations, future research should explore the relationship between environmental determinants and SWB through more well-controlled analyses using more granular data. The usage of Geographic Information System (GIS) would allow for more detail spatial analysis of environmental determinants. Finally, there are promising areas for future research that stem from this study. Given that the green coverage seems to have a direct and indirect mechanisms to improve individual's well-being by mitigating both air pollution and improving subjective health conditions, further examination of green space expansion and effectiveness of various urban green types (e.g., street tree varieties; trees distribution density) would provide useful and practical implications for the policy makers. Lastly, the additional psychology studies regarding the perception of air pollution and urban green would provide the further insight into the possible mechanisms of the relationship that we have observed between environmental determinants and subjective well-beings of the residents in this study.

The impact of environmental degradation is a major issue both in terms of policy-making and research. Life satisfaction (LS) approach has increasingly adopted to evaluate the impact of environmental determinants. Yet, the previous researches using LS approach have yet to inclusively examine the relationship between green coverage, air pollution and SWB. Hence, this study examined the relationship between, air pollution, green coverage and life satisfaction of the residents using a dataset that combines original social survey data and city-level objective data of Chinese cities. The results show that air pollution has a statistically significant negative association with LS; and urban green coverage has a statistically significant positive correlation with LS. These results are consistent with that of previous studies: Smyth et al. (2008) found the negative impact of air pollution on LS; Carrus et al. (2015) provided the evidence of positive effect of greenspace. Air pollution's negative association with people's daily lives may be largely explained by the limited mobility caused by hazes and increased health risks, which are both induced by the air pollution. On the other hand, Green coverage is associated with higher LS; green coverage has positive direct effect as well as indirect effect through air pollution reduction and improvement of subjective health. Moreover, the results show that the negative relationship of long-term air pollution and LS is more statistically significant than that of short-term air pollution. According to the estimated monetary valuations, the average monetary valuation for a 1-unit reduction in the AQI is approximately 239–280 USD (1.70%–2.00% of the annual gross individual income), and the monetary valuation of a 1% increase in green coverage is approximately 420–444 USD (3.00%–3.17% of the annual gross individual income), depending on the specifications of the AQI variables. The results also indicate that people with poor subjective health evaluation have a lower monetary value for a unit of air pollution reduction and a higher monetary value of green coverage improvement compared to people with good subjective health evaluation. The average MV for a 1% improvement in green coverage for individuals with a poor subjective health evaluation is about 2 times of the average MV of all samples. We also provided the discussion regarding the cost and possibility of SWB improvements from reducing the air quality through managing urban green spaces using the estimated impacts and valuations from previous studies in the context of urban China.

Acknowledgement This research is supported by the Ministry of Education, Culture, Sports, Science and Technology in Japan (MEXT) under a Specially Promoted Research grant [26000001]. Any opinions, findings, and conclusions expressed in this material are those of the authors and do not necessarily reflect the views of the MEXT.

Appendix A

Fig. A1. The pattern of life satisfaction.

131

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Fig. A2. The distribution of average life satisfaction scores at the city level.

Fig. A3. The distribution of AQI-short term.

Fig. A4. The distribution of AQI-long term.

132

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Fig. A5. The distribution of green-coverage areas.

Table A1 Socio-demographic characteristics of the respondents. Comparative surveya

2016 survey (N = 18,441)

Rural and urbanc Gender Male Female

49.50% 50.50%

51.22% 48.78%

Ageb 20–29 30–39 40–49 50–59 60–64

20.87% 17.97% 26.12% 28.01% 4.77%

21.34% 18.52% 22.52% 16.90% 5.68%

Gross equivalised household income (CNY) Mean 133,950 0–10,000 4.20% 10,001–25,000 3.24% 25,001–50,000 6.68% 50,001–75,000 8.51% 75,001–100,000 10.44% 100,001–125,000 19.15% 125,001–150,000 12.34% 150,001–300,000 31.60% 300,000+ 3.83%

81,405 7.63% 11.93% 22.88% 20.81% 9.68% 10.03% 7.28% 7.64% 2.13%

Urband

173,170 0.06% 0.99% 4.46% 9.18% 11.62% 14.03% 12.21% 37.04% 10.43%

a Age and gender distributions were obtained from the 2016 China Statistical Yearbook, which was a 1% population sample survey in 2015. The sampling fraction is 1.55%. Gross equivalised household income distributions are derived from the Chinese General Social Survey (CGSS) 2013 and the Chinese Household Income Project (CHIP) 2013. b Given that there is almost no respondent under 20 years old in this survey (2.71%), or above 65 years old (1.13%), for comparative results, the population under 20 and above 65 years were excluded from the original dataset. c CGSS 2013 only has an annual gross equivalised household income for 2011, while our survey collects gross equivalised household income for 2015. To address this temporal difference and our focus on urban areas, we collected yearly per capita disposable income for urban households from 2011 to 2015 from the Statistical Yearbooks, and estimate the household income for 2015 by applying each year's gross equivalised household income increase rate to the gross equivalised household income for 2011. (Per capita disposable income of urban households: 2015-31,195 CNY; 2011-21,820 CNY). d CHIP 2013 only has an annual gross equivalised household income for 2007, thus, we estimate gross equivalised household income for 2015 by applying each year's income increase rate to the urban income of 2007 using the same method as CGSS (per capita disposable income of urban households: 2015-31,195 CNY; 2011-13,786 CNY).

133

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Table A2 Air quality index and the corresponding concentration limits of pollutants. AQI Concentration limits of pollutant items in μg/m3

0 50 100 150 200 300 400 500

SO2, 24h average

SO2, hourly averagea

NO2, 24h average

NO2, hourly average

PM10, 24 h average

CO, 24h average

CO, hourly average

O3, hourly average

O3, 8h' moving PM2.5, average hourly average

0 50 150 475 800 1600 2100 2620

0 150 500 650 800

0 40 80 180 280 565 750 940

0 100 200 700 1200 2340 3090 3840

0 50 150 250 350 420 500 600

0 2 4 14 24 36 48 60

0 5 10 25 60 90 120 150

0 160 200 300 400 800 1000 1200

0 100 160 215 265 800

b b b

c c

0 35 75 115 150 250 350 500

a

The hourly average concentrations of SO2, NO2 and CO are only used in the hourly reports; 24 hours' average concentration limits are required in the daily report. When the hourly average concentration of SO2 is higher than 800 μg/m3, the corresponding AQI is no longer calculated, the AQI for SO2 is reported in the 24 hours' average concentration. c When the 8-hour average concentration is higher than 800 μg/m3, the corresponding AQI is no longer calculated, the AQI for O3 is reported in the 24 hours' average concentration. b

Table A3 Air quality index (AQI) for different countries. China

US/Japan

Canada

UK

0–50 Excellent 51–100 Good 101–150 Light pollution 151–200 Moderate pollution 201–300 Heavy pollution > 300 Serious pollution

0–50 Good 51–100 Moderate 101–150 Sensitive groups 151–200 Unhealthy 201–300 Very unhealthy 301–500 Hazardous

1 2 3 Low health risk 4 5 6 Moderate health risk 7 8 9 10 High health risk 10+ Very high health risk

1 2 Low 4 5 Moderate 7 8 High 10 Very high

3 6 9

Europe

Hong Kong

Singapore/Malaysia

0–25 Very Low 25–50 Low 50–75 Medium 75–100 High > 100 Very high

0–25 Low 26–50 Medium 51–100 High 101–200 Very high 201–500 Severe

0–50 Good 51–100 Moderate 101–200 Unhealthy 201–300 Very unhealthy > 300 Hazardous

References: China (http://www.mep.gov.cn/); US and Japan (http://airnow.gov/); Canada (http://www.airhealth.ca); UK (http://uk-air.defra.gov.uk/); Europe (http://www.airqualitynow.eu/); Hong Kong (http://www.epd-asg.gov.hk/); Singapore (http://www.nea.gov.sg/psi/); and Malaysia (http://www.doe.gov.my/ apims/).

Table A4 Variable descriptions. Variable

Definition

Shortterm

Longterm

Mean (std. dev.) Environmental and subjective health (ES) Average AQI concentration for the last 3 months (November 2015 to January 2016) before 102.80 AQI-short terma respondent took the online survey (30.41) AQI-long terma Annual average AQI concentration for 2015, the year prior to the survey 90.88 (22.35) Log (green-coverage)b Log of the green-coverage area as the % of completed areas in 2013 in the city in which the 3.73 respondent lives (0.17) Poor subjective health Personal poor subjective health, a dummy variable when 1 = poor and very poor; and is 0.25 evaluation zero otherwise (0.43) Interaction between AQI and Interactions between the AQI and poor subjective health 0.09 −0.02 poor health (12.96) (9.60) (continued on next page)

134

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Table A4 (continued) Variable

Definition

Shortterm

Longterm

Mean (std. dev.) Interaction between greencoverage and poor health City characteristics (CC) Hospitals & HC per capitab Log (consumption)b Investment per capitab Log (public transportation per capita)b Personal characteristics (PC) Trust

−0.01 (0.07)

Interactions between poor subjective health and the log of the green- coverage areas

Number of hospitals and health centers per capita in 2013 in the city in which the respondent lives (unit/10,000 persons) Log of the ratio of consumption in 2013 in the city in which the respondent lives

0.45 (0.40) 4.53 (0.15) Investment in fixed assets per capita in 2013 in the city in which the respondent lives (100 5.14 million yuan) (1.94) Log of the number of public transportation vehicles per 10,000 populations in the city in 2.61 which the respondent lives (0.56)

0.91 (0.29) Community attachment 0.61 (0.49) Activities participation A dummy variable when 1 = participate in community activities more than one day per 0.10 week; zero otherwise (0.30) Household income attitude A dummy variable when 1 = satisfied with gross equivalised household income; zero 0.31 otherwise (0.46) Log (individual income) Log of the respondent's individual gross income, which is calculated as the gross equivalised 4.23 household income divided by a household size equivalence factor (a + 0.7c)0.5, a is the (0.88) number of adults and c the number of children in the household. Age Age of the respondent in years after centering 0.00 (12.29) Age2 The square form of age 151.05 (153.91) Female A dummy variable when 1 = female; zero = male 0.50 (0.50) Married A dummy variable when 1 = married; zero = single/separated/divorced/widowed 0.76 (0.43) Children A dummy variable when 1 = having one child and above; zero = having no child 0.74 (0.44) Children under 6 A dummy variable when 1 = having one child under 6 and above; zero = having no child 0.24 under 6 (0.43) Personal educational The highest educational background of the respondent, where each item ranges from 6.81 background 1 = ‘Never attended school’ to 10 ‘Completed doctoral education’ (0.39) Partner educational background The highest education level achieved by the partner, where each item ranges from 7.40 1 = ‘Never attended school’ to 10 ‘Completed doctoral education’ (0.49) a b

A dummy variable when 1 = to be able to trust people/organizations is important; zero otherwise A dummy variable when 1 = attached to the local community; zero otherwise

Collected from China's city air quality monitoring stations. Collected from China City Statistical Yearbook-2014.

Table A5 Regression models with and without province dummies.

Environmental and subjective health (ES) AQI-short term

Without province dummies

With province dummies

Coefficient

Coefficient

Coefficient

Coefficient

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

−0.0017⁎⁎⁎ (0.0004)

−0.0039⁎⁎⁎ (0.0008) −0.0015⁎⁎⁎ (0.0005)

AQI-long term

135

−0.0047⁎⁎⁎ (0.0013) (continued on next page)

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Table A5 (continued)

Log (green-coverage) Poor subjective health evaluation Interaction between AQI and poor health Interaction between green-coverage and poor health City characteristics (CC) Hospitals & HC per capita Log (consumption) Investment per capita Log (public transportation per capita) Personal characteristics (PC) Trust Community attachment Activities participation Household income attitudes Log (individual income) Age Age2 Female Married Children Children under 6 Personal educational background Partner educational background Constant Observations R-squared

Without province dummies

With province dummies

Coefficient

Coefficient

Coefficient

Coefficient

(Std. err.)

(Std. err.)

(Std. err.)

(Std. err.)

0.4570⁎⁎⁎ (0.0697) −0.9830⁎⁎⁎ (0.0288) 0.0002 (0.0009) 1.3470⁎⁎⁎ (0.1500)

0.4540⁎⁎⁎ (0.0713) −0.9840⁎⁎⁎ (0.0288) 0.0032⁎⁎⁎ (0.0012) 1.2760⁎⁎⁎ (0.1530)

0.2200⁎⁎ (0.0988) −0.9870⁎⁎⁎ (0.0288) 0.0002 (0.0009) 1.3820⁎⁎⁎ (0.1510)

0.2380⁎⁎ (0.1030) −0.9870⁎⁎⁎ (0.0288) 0.0031⁎⁎⁎ (0.0012) 1.3070⁎⁎⁎ (0.1530)

0.0433 (0.0277) 0.2890⁎⁎⁎ (0.0802) 0.0163⁎⁎ (0.0067) 0.0293 (0.0234)

0.0364 (0.0277) 0.2740⁎⁎⁎ (0.0803) 0.0186⁎⁎⁎ (0.0067) 0.0308 (0.0234)

0.0642⁎⁎ (0.0301) 0.2570⁎⁎ (0.1070) 0.0287⁎⁎⁎ (0.0111) 0.0145 (0.0329)

0.0681⁎⁎ (0.0301) 0.2770⁎⁎⁎ (0.1070) 0.0343⁎⁎⁎ (0.0112) −0.0112 (0.0325)

0.3060⁎⁎⁎ (0.0405) 0.6190⁎⁎⁎ (0.0256) 0.0358 (0.0369) 0.8740⁎⁎⁎ (0.0264) 0.2130⁎⁎⁎ (0.0133) 0.0002 (0.0009) −0.0002⁎⁎ (0.0001) 0.2180⁎⁎⁎ (0.0224) 0.2040⁎⁎⁎ (0.0461) 0.1830⁎⁎⁎ (0.0443) −0.1400⁎⁎⁎ (0.0281) 0.0473⁎ (0.0282) 0.1220⁎⁎⁎ (0.0254) 0.8910⁎ (0.4870) 18,441 0.3150

0.3130⁎⁎⁎ (0.0405) 0.6180⁎⁎⁎ (0.0256) 0.0391 (0.0369) 0.8730⁎⁎⁎ (0.0264) 0.2130⁎⁎⁎ (0.0133) 0.0003 (0.0009) −0.0002⁎⁎ (0.0001) 0.2190⁎⁎⁎ (0.0224) 0.2050⁎⁎⁎ (0.0461) 0.1860⁎⁎⁎ (0.0443) −0.1400⁎⁎⁎ (0.0281) 0.0480⁎ (0.0282) 0.1190⁎⁎⁎ (0.0254) 0.9100⁎ (0.4890) 18,441 0.3150

0.3170⁎⁎⁎ (0.0406) 0.6140⁎⁎⁎ (0.0257) 0.0391 (0.0369) 0.8710⁎⁎⁎ (0.0266) 0.2050⁎⁎⁎ (0.0137) 0.0001 (0.0009) −0.0002⁎⁎ (0.0001) 0.2170⁎⁎⁎ (0.0225) 0.2080⁎⁎⁎ (0.0462) 0.1900⁎⁎⁎ (0.0445) −0.1550⁎⁎⁎ (0.0285) 0.0476⁎ (0.0282) 0.1160⁎⁎⁎ (0.0255) 2.3680⁎⁎⁎ (0.7170) 18,441 0.3180

0.3200⁎⁎⁎ (0.0406) 0.6140⁎⁎⁎ (0.0257) 0.0394 (0.0369) 0.8700⁎⁎⁎ (0.0266) 0.2050⁎⁎⁎ (0.0137) 0.0001 (0.0009) −0.0002⁎⁎ (0.0001) 0.2160⁎⁎⁎ (0.0225) 0.2100⁎⁎⁎ (0.0462) 0.1940⁎⁎⁎ (0.0445) −0.1570⁎⁎⁎ (0.0285) 0.0471⁎ (0.0282) 0.1150⁎⁎⁎ (0.0256) 2.3490⁎⁎⁎ (0.7600) 18,441 0.3170

Notes: 1. Standard errors in parentheses. 2. Columns (1) and (2) are the baseline model and only include the survey variables and hard data. Columns (3) and (4) are the full model with the province dummies. ⁎⁎⁎ p < 0.01. ⁎⁎ p < 0.05. ⁎ p < 0.1.

136

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

References

Environ. Polit. 15, 171–189. Ferreira, S., Akay, A., Brereton, F., Cuñado, J., Martinsson, P., Moro, M., Ningal, T.F., 2013. Life satisfaction and air quality in Europe. Ecol. Econ. 88, 1–10. Fleming, C.M., Manning, M., Ambrey, C.L., 2016. Crime, greenspace and life satisfaction: an evaluation of the New Zealand experience. Landsc. Urban Plan. 149, 1–10. Frey, B.S., Luechinger, S., Stutzer, A., 2010. The life satisfaction approach to environmental valuation. Ann. Rev. Resour. Econ. 2, 139–160. Fuller, R.A., Irvine, K.N., Devine-Wright, P., Warren, P.H., Gaston, K.J., 2007. Psychological benefits of greenspace increase with biodiversity. Biol. Lett. 3, 390–394. Gobster, P.H., 1994. The urban savanna: reuniting ecological preference and function. In: Restoration and Management Notes. 12. pp. 64–71. Goetzke, F., Islam, S., 2017. Testing for spatial equilibrium using happiness data. J. Reg. Sci. 57, 199–217. Guo, J., Chen, J., 2009. Patterns of seasonal change in air pollution and suggestions on pollution control in Beijing. J. Ecol. Environ. Sci. 18 (3), 952–956 (In Chinese). Helliwell, J.F., Putnam, R.D., 2012. The social context of well-being. In: The Science of Well-being. Henwood, K., Pidgeon, N., 2001. Talk about woods and trees: threat of urbanization, stability, and biodiversity. J. Environ. Psychol. 21, 125–147. Heywood, D. Ian, Cornelius, S., Carver, S., 1998. An Introduction to Geographical Information Systems. Addison Wesley Longman, New York. Honold, J., Lakes, T., Beyer, R., van der Meer, E., 2014. Restoration in urban spaces: nature views from home, greenways, and public parks. Environ. Behav. 48, 796–825. Index, E.P., 2016. Global Metrics for Environment. Yale University Information available online at: http://epi.yale.edu/sites/default/files/2016EPI_Full_Report_opt.pdf. Israel, D., Levinson, A., 2003. Examining the Relationship Between Household Satisfaction and Pollution. Indiana State University Mimeo. Jayasooriya, V.M., Ng, A.W.M., Muthukumaran, S., Perera, B.J.C., 2017. Green infrastructure practices for improvement of urban air quality. Urban For. Urban Green. 21, 34–47. Jim, C.Y., 2004. Green-space preservation and allocation for sustainable greening of compact cities. Cities 21, 311–320. Jim, C.Y., Chen, W.Y., 2009. Ecosystem services and valuation of urban forests in China. Cities 26, 187–194. Joung, Stronks, K., van de Mheen, H., van Poppel, van der Meer, Mackenbach, 1997. The contribution of intermediary factors to marital status differences in self-reported health. J. Marriage Fam. 59, 476–490. Knight, T.M., Dunn, J.L., Smith, L.A., Davis, J., Kalisz, S., 2009. Deer facilitate invasive plant success in a Pennsylvania forest understory. Nat. Areas J. 29, 110–116. Krekel, C., Kolbe, J., Wüstemann, H., 2016. The greener, the happier? The effect of urban land use on residential well-being. Ecol. Econ. 121, 117–127. Levinson, A., 2012. Valuing public goods using happiness data: the case of air quality. J. Public Econ. 96, 869–880. Li, F., Wang, R., Liu, X., Zhang, X., 2005. Urban forest in China: development patterns, influencing factors and research prospects. Int. J. Sustain. Dev. World Ecol. 12, 197–204. Li, Z., Folmer, H., Xue, J., 2014. To what extent does air pollution affect happiness? The case of the Jinchuan mining area, China. Ecol. Econ. 99, 88–99. Liu, T., Marlier, M.E., DeFries, R.S., Westervelt, D.M., Xia, K.R., Fiore, A.M., Mickley, L.J., Cusworth, D.H., Milly, G., 2018. Seasonal impact of regional outdoor biomass burning on air pollution in three Indian cities: Delhi, Bengaluru, and Pune. Atmos. Environ. 172, 83–92. Luechinger, S., 2010. Life satisfaction and transboundary air pollution. Econ. Lett. 107, 4–6. Maas, J., Verheij, R.A., Groenewegen, P.P., De Vries, S., Spreeuwenberg, P., 2006. Green space, urbanity, and health: how strong is the relation? J. Epidemiol. Community Health 60, 587–592. Mabahwi, N.A.B., Leh, O.L.H., Omar, D., 2014. Human health and wellbeing: human health effect of air pollution. Procedia. Soc. Behav. Sci. 153, 221–229. MacKerron, G., 2012. Happiness economics from 35 000 feet. J. Econ. Surv. 26, 705–735. MacKerron, G., Mourato, S., 2009. Life satisfaction and air quality in London. Ecol. Econ. 68, 1441–1453. MacKerron, G.J., Egerton, C., Gaskell, C., Parpia, A., Mourato, S., 2009. Willingness to pay for carbon offset certification and co-benefits among (high-)flying young adults in the UK. Energy Policy 37, 1372–1381. Manes, F., Silli, V., Salvatori, E., Incerti, G., Galante, G., Fusaro, L., Perrino, C., 2014. Urban ecosystem services: tree diversity and stability of PM10 removal in the metropolitan area of Rome. Ann. Bot. 4, 19–26. Manning, M., Fleming, C.M., Ambrey, C.L., 2016. Life satisfaction and individual willingness to pay for crime reduction. Reg. Stud. 50, 2024–2039. Nassauer, J.I., 1995. Messy ecosystems, orderly frames. Landsc. J. 14, 161–170. National Bureau of Statistics of China, 2014. China City Statistical Yearbook 2014. China Statistics Press, Beijing. National Bureau of Statistics of China, 2015. China Statistical Yearbook on Environment 2015. China Statistics Press, Beijing. National Bureau of Statistics of China, 2016. China Statistical Yearbook 2016. China Statistics Press, Beijing. Nowak, D.J., Hirabayashi, S., Bodine, A., Greenfield, E., 2014. Tree and forest effects on air quality and human health in the United States. Environ. Pollut. 193, 119–129. Ouyang, Z.Y., Wang, R.S., Zhao, J.Z., 1999. Ecosystem services and their economic valuation. Chin. J. Appl. Ecol. 10, 635–640 (In Chinese). Pope, C.A., Dockery, D.W., 2006. Health effects of fine particulate air pollution: lines that connect. J. Air Waste Manage. Assoc. 56, 709–742. van Praag, B.M.S., Baarsma, B.E., 2005. Using happiness surveys to value intangibles: the case of airport noise. Econ. J. 115, 224–246.

Ambrey, C.L., 2016. Greenspace, physical activity and well-being in Australian capital cities: how does population size moderate the relationship? Public Health 133, 38–44. Ambrey, C.L., Fleming, C.M., 2012. Valuing Australia's protected areas: a life satisfaction approach. N. Z. Econ. Pap. 46, 191–209. Ambrey, C.L., Fleming, C.M., 2014a. The causal effect of income on life satisfaction and the implications for valuing non-market goods. Econ. Lett. 123, 131–134. Ambrey, C.L., Fleming, C.M., 2014b. Valuing ecosystem diversity in south East Queensland: a life satisfaction approach. Soc. Indic. Res. 115, 45–65. Ambrey, C.L., Fleming, C.M., 2014c. Public greenspace and life satisfaction in urban Australia. Urban Stud. 51, 1290–1321. Ambrey, C.L., Fleming, C.M., Chan, A.Y.C., 2014. Estimating the cost of air pollution in South East Queensland: an application of the life satisfaction non-market valuation approach. Ecol. Econ. 97, 172–181. Ambrey, C.L., Ulichny, J., Fleming, C.M., 2017. The social connectedness and life satisfaction Nexus: a panel data analysis of women in Australia. Fem. Econ. 23, 1–32. Appleton, S., Song, L., 2008. Life satisfaction in urban China: components and determinants. World Dev. 36, 2325–2340. Asadullah, M.N., Xiao, S., Yeoh, E., 2018. Subjective well-being in China, 2005–2010: the role of relative income, gender, and location. China Econ. Rev. 48, 83–101. Au, N., Johnston, D.W., 2014. Self-assessed health: what does it mean and what does it hide? Soc. Sci. Med. 121, 21–28. Bayer, P., Keohane, N., Timmins, C., 2009. Migration and hedonic valuation: the case of air quality. J. Environ. Econ. Manag. 58, 1–14. Beijing Municipal Statistics Bureau, 2015. Beijing Statistical Yearbook 2014. China Statistics Press, Beijing. Blanchflower, D.G., Oswald, A.J., 2008. Is well-being U-shaped over the life cycle? Soc. Sci. Med. 66, 1733–1749. Bonnes, M., Passafaro, P., Carrus, G., 2011. The ambivalence of attitudes toward urban green areas: between proenvironmental worldviews and daily residential experience. Environ. Behav. 43, 207–232. Bottalico, F., Chirici, G., Giannetti, F., De Marco, A., Nocentini, S., Paoletti, E., Salbitano, F., Sanesi, G., Serenelli, C., Travaglini, D., 2016a. Air pollution removal by green infrastructures and urban forests in the City of Florence. Agric. Agric. Sci. Procedia 8, 243–251. Bottalico, F., Pesola, L., Vizzarri, M., Antonello, L., Barbati, A., Chirici, G., Corona, P., Cullotta, S., Garfì, V., Giannico, V., Lafortezza, R., Lombardi, F., Marchetti, M., Nocentini, S., Riccioli, F., Travaglini, D., Sallustio, L., 2016b. Modeling the influence of alternative forest management scenarios on wood production and carbon storage: a case study in the Mediterranean region. Environ. Res. 144, 72–87. Brereton, F., Clinch, J.P., Ferreira, S., 2008. Happiness, geography and the environment. Ecol. Econ. 65, 386–396. Brunekreef, B., Holgate, S.T., 2002. Air pollution and health. Lancet 360 (9341), 1233–1242. Cai, Y., Zhu, N., Han, H., 2002. Photosynthesis and transpiration of urban tree species in Harbin. In: Advances in Urban Forest Ecology, pp. 186–194. Carrus, G., Passafaro, P., Bonnes, M., 2004. Ambientalismo, autoritarismo ed etnocentrismo: una scala di atteggiamento verso il verde urbano. Bollettino Di Psicologia Applicata 242, 3–12. Carrus, G., Scopelliti, M., Lafortezza, R., Colangelo, G., Ferrini, F., Salbitano, F., Agrimi, M., Portoghesi, L., Semenzato, P., Sanesi, G., 2015. Go greener, feel better? The positive effects of biodiversity on the well-being of individuals visiting urban and peri-urban green areas. Landsc. Urban Plan. 134, 221–228. Chen, W.Y., Jim, C.Y., 2008a. Cost-benefit analysis of the leisure value of urban greening in the new Chinese city of Zhuhai. Cities 25, 298–309. Chen, W.Y., Jim, C.Y., 2008b. Assessment and valuation of the ecosystem services provided by urban forests. In: Springer Der. Env. Man, pp. 53–83. Chen, J., Davis, D.S., Wu, K., Dai, H., 2015. Life satisfaction in urbanizing China: The effect of city size and pathways to urban residency. Cities 49, 88–97. Chunju, C., Zhenhua, P., Cheng, W., 2004. The ecological benefit of urban Forest and trees and its value. World For. Res. 3, 004. Cohen-Cline, H., Turkheimer, E., Duncan, G.E., 2015. Access to green space, physical activity and mental health: a twin study. J. Epidemiol. Community Health 69, 523–529. Cuñado, J., de Gracia, F.P., 2013. Environment and happiness: new evidence for Spain. Soc. Indic. Res. 112, 549–567. van den Berg, A.E., Maas, J., Verheij, R.A., Groenewegen, P.P., 2010. Green space as a buffer between stressful life events and health. Soc. Sci. Med. 70, 1203–1210. van den Berg, M., Wendel-Vos, W., van Poppel, M., Kemper, H., van Mechelen, W., Maas, J., 2015. Health benefits of green spaces in the living environment: a systematic review of epidemiological studies. Urban For. Urban Green. 14, 806–816. van den Berg, M., van Poppel, M., van Kamp, I., Andrusaityte, S., Balseviciene, B., Cirach, M., Danileviciute, A., Ellis, N., Hurst, G., Masterson, D., Smith, G., Triguero-Mas, M., Uzdanaviciute, I., de Wit, P., van Mechelen, W., Gidlow, C., Grazuleviciene, R., Nieuwenhuijsen, M.J., Kruize, H., Maas, J., 2016. Visiting green space is associated with mental health and vitality: a cross-sectional study in four European cities. Health Place 38, 8–15. Dolan, P., Peasgood, T., White, M., 2008. Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective wellbeing. J. Econ. Psychol. 29, 94–122. Du, G., Shin, K., Managi, S., 2018. Variability in impact of air pollution on subjective wellbeing. Atmos. Environ. https://doi.org/10.1016/j.atmosenv.2018.04.018. Economy, E., 2006. Environmental governance: the emerging economic dimension.

137

Ecological Economics 153 (2018) 124–138

L. Yuan et al.

Tsurumi, T., Managi, S., 2015. Environmental value of green spaces in Japan: an application of the life satisfaction approach. Ecol. Econ. 120, 1–12. Ulrich, R., 1984. View through a window may influence recovery from surgery. Science 224 (4647), 420–421. Welsch, H., 2006. Environment and happiness: valuation of air pollution using life satisfaction data. Ecol. Econ. 58, 801–813. Welsch, H., 2007. Environmental welfare analysis: a life satisfaction approach. Ecol. Econ. 62, 544–551. Welsch, H., 2009. Implications of happiness research for environmental economics. Ecol. Econ. 68 (11), 2735–2742. Welsch, H., Ferreira, S., 2014. Environment, well-being, and experienced preference. Int. Rev. Environ. Resour. Econ. 7, 205–239. Welsch, H., Kühling, J., 2009. Using happiness data for environmental valuation: issues and applications. J. Econ. Surv. 23, 385–406. World Bank, 2006. Water Quality Management Policy and Institutional Considerations. The World Bank, Washington, DC. Wu, W., Dong, G., 2014. Valuing the “green” amenities in a spatial context. J. Reg. Sci. 54, 569–585. Xiao, H., Ouyang, Z., Zhao, J., Wang, X., 2000. Forest ecosystem services and their ecological valuation-a case study of tropical forest in Jianfengling of Hainan Island. Chin. J. Appl. Ecol. 11, 481–484. Yang, J., McBride, J., Zhou, J., Sun, Z., 2004. The urban forest in Beijing and its role in air pollution reduction. Urban For. Urban Green. 3, 65–78. Yu, K., Padua, M.G., 2007. China's cosmetic cities: urban fever and superficiality. Landsc. Res. 32, 255–272. Zhang, X., Zhang, X., Chen, X., 2017a. Valuing air quality using happiness data: the case of China. Ecol. Econ. 137, 29–36. Zhang, X., Zhang, X., Chen, X., 2017b. Happiness in the air: how does a dirty sky affect mental health and subjective well-being? J. Environ. Econ. Manag. 85, 81–94.

Rasmussen, K.H., Taheri, M., Kabel, R.L., 1975. Global emissions and natural processes for removal of gaseous pollutants. Water Air Soil Pollut. 4, 33–64. Robinson, C., Schumacker, R., 2009. Interaction effects: centering, variance inflation factor, and interpretation issues. In: Mult. Linear Regres. Viewpoints. 35. pp. 6–11. Roy, S., Byrne, J., Pickering, C., 2012. A systematic quantitative review of urban tree benefits, costs, and assessment methods across cities in different climatic zones. Urban For. Urban Green. 11, 351–363. Selmi, W., Weber, C., Rivière, E., Blond, N., Mehdi, L., Nowak, D., 2016. Air pollution removal by trees in public green spaces in Strasbourg City, France. Urban For. Urban Green. 17, 192–201. Shields, M., Wooden, M., 2003. Marriage, Children and Subjective Well-being. Paper Presented at the Australian Institute of Family Studies Conference, Melbourne. Smyth, R., Mishra, V., Qian, X., 2008. The environment and well-being in urban China. Ecol. Econ. 68, 547–555. Smyth, R., Nielsen, I., Zhai, Q., Liu, T., Liu, Y., Tang, C., Wang, Z., Wang, Z., Zhang, J., 2011. A study of the impact of environmental surroundings on personal well-being in urban China using a multi-item well-being indicator. Popul. Environ. 32, 353–375. Spector, P.E., Cooper, C.L., Poelmans, S., Allen, T.D., O'Driscoll, M., Sanchez, J.I., Siu, Oi Ling, Dewe, P., Hart, P., Lu, Luo, De Moraes, L.F.R., Ostrognay, G.M., Sparks, K., Wong, P., Yu, S., 2004. A cross-national comparative study of work-family stressors, working hours, and well-being: China and Latin America versus the Anglo world. Pers. Psychol. 57, 119–142. State Environmental Protection Agency, 2004. State of the Environment in China: 2003. Sun, C., Lin, T., Zhao, Q., Li, X., Ye, H., Zhang, G., Liu, X., Zhao, Y., 2017. Spatial pattern of urban green spaces in a long-term compact urbanization process-a case study in China. Ecol. Indic (Available online: 14 October 2017, In Press). Thayer, R.L., 1989. The experience of sustainable landscapes. Landsc. J. 8, 101–110. Tian, Y., Jim, C.Y., Tao, Y., 2012. Challenges and strategies for greening the compact city of Hong Kong. J. Urban Plann. Dev. 138, 101–109.

138