Air Pollution, Residents’ Happiness, and Environmental Regulation: Evidence from China

: Using data from the China General Social Survey and data on air pollution, this study explores the impact and the critical 7 path of air pollution on residents ’ happiness in China and evaluates whether environmental regulations can alleviate these effects. A 8 probit model is used to analyze the impact of air pollution on residents ’ happiness, and wind speed is taken as the instrumental 9 variable of air pollution to overcome endogeneity. A stepwise regression is used to test the critical path of air pollution on residents ’ 10 happiness. Finally, the effects of environmental regulations are considered by adding an interaction term between environmental 11 regulation and air pollution. The following conclusions are drawn. First, air pollution can significantly impair residents ’ happiness, 12 especially those who have children, are younger, are in poor health, have a lower education level, have lower income, and live in a 13 rural area. Second, there are two critical paths through which air pollution impairs residents ’ happiness: mental health level and the 14 frequency of leisure activities. Finally, command-and-control, market-based, and voluntary environmental regulations can all ef- 15 fectively alleviate the impact of air pollution on residents ’ happiness.


Introduction
An increase in happiness is conducive to the improvement of residents' health (Diener and Chan 2011), social 20 productivity (Oswald et al. 2015), and social harmony (Clark 2018). However, since the reform and opening-up in 1978, 21 China's rapid economic development has greatly satisfied people's material needs, but it has also created serious air  (Cohen et al. 2017). Air pollution adversely affects people's health and lives, which 26 may reduce their happiness. In this context, it is of great practical significance to clarify the relationship between air 27 pollution and happiness to improve residents' quality of life. 28 Previous studies have shown that air pollution can damage residents' happiness (Ferreira et al. 2013;Zhang et al. 29 2017; Li et al. 2014). However, what is the mechanism of its influence? Do the measures proposed to deal with envi- 30 ronmental pollution, such as environmental regulation, influence the relationship between air pollution and happi-31 ness? Existing studies do not provide answers to these questions, so this paper attempts to fill this gap, taking China as 32 the research object to investigate the impact of air pollution on residents' happiness.
Second, this paper enriches our understanding of the influential mechanism of air pollution on happiness. Studies 47 on the direct impact of air pollution on happiness are relatively dated. In recent years, research has explored the 48 mechanism of air pollution on happiness, such as through reducing the level of physical health and self-rated health 49 (Yuan et al. 2018; Li and Zhou 2020), increasing residents' pollution perceptions (Chen et al. 2020), and reducing qual- 50 ity of life (Ma et al. 2019) thereby reducing residents' happiness. However, relevant studies are limited. Therefore, this 51 paper explores two new ways in which air pollution reduces residents' happiness: by reducing the frequency of leisure 52 activities and the level of mental health. 53 Third, we confirm that environmental regulation has an alleviating effect on the relationship between air pollution 54 and happiness. There are two relevant articles. One shows that an air pollution tax has an alleviating effect on the 55 relationship between air pollution and residents' happiness (Liu et al. 2019), and the other holds that the implementa-56 tion of "China's South-to-North Water Diversion Eastern Route Pollution Control Project" will not improve happiness 57 (Jin et al. 2020). However, the environmental regulations used in these two articles are the same policies implemented 58 in many provinces across the country, which cannot reflect well the differences in environmental regulations. There-59 fore, this paper takes the three kinds of environmental regulations (The first is command-and-control environmental 60 regulation; the second category is market-based environmental regulation, and the third refers a type of voluntary 61 environmental regulation) as the moderating variable and finds that environmental regulations could alleviate the 62 negative impact of air pollution on residents' happiness. 63 The rest of the paper is organized as follows. Section 2 is a literature review. Section 3 analyzes the theory and 64 makes corresponding assumptions. Section 4 introduces the data source, variable selection, and model setting. Sections 65 5 and 6 discuss the empirical results and conclusions and offer some policy suggestions, respectively. 66

67
There is abundant literature on residents' happiness, which can be divided into two categories according to 68 whether or not the influencing factors are economic factors. One approach has been to explore the impact of economic 69 factors, such as absolute income (Sacks et al. 2010), relative income (Easterlin 1974), and income inequality (Oishi et al. Family Panel Studies (CFPS) and found that air pollution reduces happiness and increases the incidence of depressive 82 symptoms. By constructing a hierarchical spatial autoregressive probit model, Ma et al. (2020) showed that the welfare 83 effect of air quality could not be replaced by other factors of people's livelihood. 84 Based on the direct impact of air pollution on residents' happiness, previous studies began to pay attention to the 85 mechanism of the impact of air pollution on happiness. Chen  pollution abatement project may not enhance happiness in its duration. However, the latter literature related to this 105 paper only involves a relatively small part of the environmental regulation system, which may not be representative. 106 Therefore, it remains to be verified whether other environmental regulations have a moderating effect on the rela-107 tionship between air pollution and happiness. Hierarchy of Needs Theory divides human needs into five levels from low to high, namely, physiological needs, safety 112 needs, love and belonging, esteem needs, and self-actualization (Maslow 1943). With the most basic physiological 113 needs, such as food, clothing, shelter, and transportation, people begin to pursue higher-level needs, such as health, 114 safety, leisure, and spirituality. Low-level needs are material needs that are affected by income, where income growth 115 can increase happiness. However, high-level needs are non-material needs that cannot be satisfied by an increase in 116 income, and where non-economic factors may be at work. As one of the non-economic factors, air quality suffers from 117 pollution, which will hinder the satisfaction of human higher-level needs. When humans' needs are not satisfied, their 118 happiness will also decrease. Therefore, we propose the following hypothesis: Macao, and Taiwan. After deleting missing and abnormal value cases, 30,745 observations remained. 157 The data on SO2 emission, regional economy, industrial structure, medical conditions, and education levels were  The mediator variables were mental health and leisure activities. The CGSS includes the following question about 187 mental health: "How often did you feel depressed in the past four weeks?" Answer choices (coded 1-5) were: always, 188 often, sometimes, rarely, and never, with higher score indicating a higher level of mental health (Wang and Lu 2020). 189 Leisure activities explored in this paper include shopping, watching movies, getting together with friends, and exer-190 cising. They were measured using the following CGSS question: "In the past year, did you often engage in the fol-191 lowing activities in your spare time?" Answer choices (coded 1-5) were: never, several times a year or less, several 192 times a month, several times a week, and every day, with higher scores indicating a higher frequency of leisure activi-193 ties. 194 The moderator variable was environmental regulations. Based on current research, this paper divides environ-195 mental regulations into three types. The first is command-and-control environmental regulations, which are govern- where i represents the individual, lnSO2 represents air pollution, X is the vector of individual characteristics and pro- (2) and (3).The first stage isolates the exogenous part of the endogenous variable: In the second stage, the estimated value of lnSO2 in (2) is used to make a regression model of happiness: where IV represents the instrumental variable, and To test whether air pollution affects residents' happiness by reducing their mental health level and frequency of 236 leisure activities, this paper refers to Baron


Step 2: Test the mediation effects. If the regression coefficients a and b in Equations (5) and (6) where Mod represents environmental regulation, and  The dependent variable was replaced by tri-classification 9 and five-classification 10 variables representing happi-299 ness, and an ordered probit model was used to reevaluate Equations (2)-(3). The results are show in Columns (1) and 300 (2) of Table 3, and the regression coefficients for air pollution (lnSO2) are −0.2173 and −0.3838 with p<0.01. These indi-301 cates that air pollution has a significant negative impact on happiness, which is consistent with the baseline regression 302 results obtained in this paper. 303 Columns (3)-(5) of Table 3 Table 4 report the estimated results ob-318 tained when the independent variable was replaced by AQI, PM2.5, PM10, CO, and NO2 concentrations, respectively. 319 The results show that the regression coefficients are all negative and significant at the 5% level, which is consistent with 320 the baseline results in this paper. This suggests that the results are robust when the independent variable is replaced 321 with different air pollution indicators.  west, and Northeast. Then, the sample from one of the regions was removed from the analysis and Equations (2) and 338 (3) were reevaluated. Columns (1)-(7) of Table 5 report the results after excluding one region per model. Regardless of 339 which region was excluded, the regression coefficients for air pollution are negative and significant at the 5% level, 340 indicating that the influence of air pollution on happiness is not affected by regional influence. 341 Table 5. Robustness checks: Eliminate regional samples.  Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; standard errors in parentheses. groups. Therefore, this paper conducts heterogeneity analysis using different family roles, age structure, health status, 352 education level, income status, and residence type (urban vs. rural). 353 According to whether or not the interviewees have children, we conducted a sub-sample regression, and the re-354 sults are presented in Row (1) of Table 6. The absolute value of regression coefficient for air pollution (  Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; standard errors in parentheses.

366
Another sub-sample regression was conducted according to respondents' self-rated health status (good, average, 367 or poor), and the results are provided in Row (3) of Table 6. Air pollution is found to has the greatest impact on the 368 happiness of residents with poor self-rated health. Such people may have poor immunity, so air pollution will do more 369 harm to their bodies, thereby increasing their sensitivity to air pollution. Next, all the samples were divided into three 370 groups according to education level: low, medium, and high. 11 The regression results of these three sub-samples are 371 shown in Row (4) of Table 6. Here, air pollution has the greatest impact on the happiness of less educated people, 372 perhaps because people with higher education may have a better working environment, making them less affected by 373 air pollution. 374 Ranking income levels from low to high, we defined the groups as low income (bottom 25%), middle income 375 (25%-50%), and high income (over 50%). Sub-sample regressions were conducted, and the results are displayed in Row 376 (5) of Table 6. Air pollution had the greatest impact on the lowest income group, but it had no significant impact on 377 people in the top of the income distribution. People with higher income can choose a better living environment with 378 lower air pollution and have a better ability to deal with air pollution by buying better equipment, such as air purifier. 379 Finally, another sub-sample regression was conducted according to residential region (urban vs. rural), and the 380 results are shown in Row (6) of Table 6. Here, air pollution has a greater impact on rural residents. This may be because 381 rural residents work outdoors and are exposed to air pollution for a longer period of time each day. Moreover, they are 382 less aware and less able to take protective measures against air pollution, which leads to more loss of happiness. 383

384
According to the steps of the stepwise regression, the first step tests the total effect. Column (6) of Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively; standard errors in parentheses. 402 Command-and-control environmental regulations are mandatory, mainly rely on the government formulating a 404 series of emission standards, which are mandatory for enterprises to meet (Blackman 2010 There are three reasons why environmental regulations can alleviate the impact of air pollution on happiness. 424 First, command-and-control and market-based policies represent the government's attention to environmental pollu-425 tion, which reduces residents' dissatisfaction with the government to some extent, thereby improving their happiness. 426 Second, the three types of environmental regulations can all improve air quality. When people believe that environ-

434
In this study, we combine provincial characteristics data with individual characteristics data from the CGSS (2013, 435 2015, and 2017), and use wind speed as the instrumental variable of air pollution. A 2SLS regression was used to em-436 pirically study the effects of air pollution on residents' happiness. In addition, mediating and moderating variables 437 were introduced to explore the mechanism and the alleviating effect of air pollution's impact on residents' happiness. 438 The main conclusions are as follows. 439 First, air pollution has a significant negative impact on residents' happiness, and the effect is different among 440 groups with different family roles, ages, residence types, self-rated health, education levels, and income levels. Second, 441 there is a significant mediating effect on the level of mental health and the frequency of leisure activities-that is, air