High-speed Railways and Environmental Pollution: The Mediating Effect of Environmental Regulations and Moderating Effect of Ocials' Political Promotion Incentives

: This study examines the relationship between high-speed railways (HSRs) and environmental 10 pollution by focusing on the mediating role of environmental regulations and the moderating role of officials ’ 11 political promotion incentives. Based on a sample of 113 prefecture-level cities, with balanced panel data in 12 China from 2009 to 2017, using the difference-in-differences (DID) model, the results show that HSRs can 13 reduce environmental pollution via the mediating effect of environmental regulations. Additionally, high 14 officials ’ political promotion incentives can strengthen this mediating effect. A propensity score matching with 15 difference-in-differences (PSM-DID) model is used to solve endogenous problems, and a placebo test and a 16 parallel trend test indicate that these results are robust. This study encourages the government to rationally 17 promote the construction of high-speed railways and expand the social advantages of high-speed railways to 18 improve environmental regulations and reduce environmental pollution.

4 and environmental pollution controlled through environmental regulations was tested (DeVaro, 2006).

85
The contribution of this study to the literature is threefold. First, previous research has emphasized the 86 direct effect of HSRs only. We propose a new conceptual framework for the indirect influence of HSRs on 87 environmental pollution with respect to environmental regulations, which enriches the literature on path  HSRs, but also attach importance to their social advantages. Promoting the influence of HSRs on environmental 102 regulations is an effective way to control environmental pollution. Second, our paper exposes officials' strong 103 political incentives that can prompt environmental regulations and further positively control pollution.

104
Governments could develop a suitable officials' political promotion mechanism by considering environmental 105 pollution control performance assessment.

106
The remainder of this paper is organized as follows. In Section 2, we review the relevant literature and 107 introduce five hypotheses. Then, we introduce the data, variables, and research methods in Section 3. In Section 108 4, we discuss the empirical results. Finally, in Section 5, the principal findings are summarized, and theoretical  The effect of HSRs on environmental pollution can be analyzed from the perspective of direct and indirect 114 impacts. Regarding the direct impact, HSRs are encouraged because of their low carbon and high efficiency 115 advantages. Givoni (2007) found that, comparing the emissions, impact, and damage costs of air travel and 116 high-speed rail travel, it is beneficial to replace airplane seats with high-speed rail seats. Janic (2011) stated that 117 this substitution effect involves reducing the amount and associated costs of social and environmental impacts, 118 such as airport airside delays, noise, and local and global emissions of greenhouse gases. Considering the 119 indirect impacts, the high-speed rail can reduce air pollution through an innovative effect, allocating effect, and 120 substituting effect, which is discussed by Yang et al. (2019). HSRs can not only transport people and goods but 121 also accelerate innovation, improve resource utilization efficiency, and substitute industrial structures, which 122 can reduce pollution (Vickerman, 2018). Based on these results, we can develop hypothesis 1:

125
There is asymmetric information between firms and regulators, which increases the difficulties and costs of 126 environmental regulations. According to information asymmetry theory, a party with more information is in an 127 advantageous position in terms of economic activity (Quanqi Liu and Li . Meanwhile, firms are in 128 an advantageous position in terms of pollution information disclosure. To pursue maximum economic benefits, 129 firms tend to sacrifice environmental protection. Nevertheless, according to reputation theory, they need to build 130 a good public reputation for environmental protection to further economic development (Gioia et al., 2000). As 131 a result, firms are motivated to deliberately withhold information on pollution. This asymmetric information 132 will increase the difficulty for regulators to discover the truth and obtain comprehensive information.

133
Furthermore, regulators may make unsound environmental regulations decisions that adversely affect pollution 134 control. Therefore, reducing asymmetric information is the key point for environmental regulations.

135
Nowadays, the internet and videoconferencing have been greatly developed. However, for business, 136 scientific, and creative activities, face-to-face contact is still necessary. During face-to-face communication,

137
individuals' awareness and participation in discussions will increase, more problems will be identified, and 138 more information will be shared. As a result, negative information spreads more rapidly and frequently 139 according to the negativity bias theory (Yan and Jiang, 2018). Some existing literature has indicated that the 140 introduction of HSRs can reduce the time and geographical distance among different cities, promote face-to-141 face communication, and enhance knowledge spreading between regions (Lin, 2017). HSRs not only carry 6 people and goods but also "carry" information and act as information nodes. The introduction of HSRs provides 143 people more opportunities for discussion and communication. More information on firms' environmental 144 pollution will be discovered during face-to-face communication by the government and investors, which can 145 reduce asymmetric information (DiMicco et al., 2007). In this situation, environmental regulations will be 146 strengthened, and firms will be forced to reduce emissions (Gioia et al., 2000). Therefore, this discussion leads

154
According to reputation theory, reputation is a precious resource that can gain a competitive advantage for firms 155 in comparison to similar distributors. To prevent negative judgment, investors and shareholders use 156 environmental information in unsophisticated ways to pressure firms to voluntarily reduce their pollution.

218
The original resource of environmental pollution data used in this study is from the Chinese Cities 219 Statistical Yearbook. We supplement the missing data by using the average growth rate of each city. The       We set HSR as the independent variable in our empirical analysis and operationalized it as a dummy variable.

245
Because we use the difference-in-differences (DID) method for our empirical analysis, we construct a group 246 dummy variable, dz, and a time dummy variable, dt. The interaction term dz*dt is our focus, which is described 247 as the "policy treatment effect" of an HSR, and we define it as HSR ≡ dz*dt. Regarding dz, we set cities with 248 HSR services from 2009 to 2017 as the treated group and let dz = 1. We set cities that do not have HSR services

286
According to the Interim Provisions for Party and Government Leading Cadre Tenure, mayors below the age 287 of 57 can be promoted to higher-level positions (Kou and Tsai, 2014). Zhou and Zeng (2018) also found that 288 mayors who were below the age of 57 had stronger promotion incentives. If they cannot get promoted before 289 this age, they are less likely to be promoted afterward and retire with relatively lower pension packages. Based 290 on this analysis, we choose a mayor's age as a proxy variable to represent his or her political promotion 291 incentives and define it as a dummy variable. If a mayor's age is less than 57, we regard this mayor as having 292 strong political promotion incentives, and Pro_incent equals 1; otherwise, Pro_incent equals 0.

293
The variables, brief descriptions of the variables, and basic descriptive statistics are shown in Table 1

297
The correlation matrix for all variables is described in Table 2. These variables were used in logarithmic 298 form, except for the dummy variable. The correlation coefficient is lower than 0.60. Additionally, the maximum 299 variance inflation factor (VIF) in the regression is 3.12, which is lower than the critical threshold value of 10 300 (Giorgio and Bedogni, 2010). This indicates that there was no multicollinearity problem.

303
The DID model is adopted in this study because the introduction of an HSR is a part of a national-level strategic where i is the city index, and t is the time index. Further, Pollutionit is a dependent variable; i and t are city 317 and year fixed effects, respectively; ∑ is control variables; and it is an error term. Because the specification 318 includes city and year fixed effects, it is not necessary to include a noninteraction treatment or a period dummy 319 variable. The estimate of the effect of HSR on pollution is 1.

320
The hierarchical regression method was applied to test the mediating effects ( Baron and Kenny, 1986

327
However, if the coefficient of PITI is significant, and the coefficient of HSR remains significant, the partially 328 mediating effect is significant. The three-step hierarchical regression method (Model 2) is expressed as follows:

329
Step 1. Examining the impact of the introduction of HSR on environmental pollution: (2) Step 2. Testing the impact of the introduction of HSR on environmental regulations: Step 3. Putting HSR and PITI into the regression equation: To examine the moderated mediation effects of officials' political promotion incentives on the relationship 333 between environmental regulations and environmental pollution, hierarchical regression was applied (James

334
and Brett, 1984). In the first step, the relationship between the introduction of HSRs and officials' political 335 promotion incentives was tested. If the coefficient of HSR is significant, we proceed to the next step. In the 336 second step, the relationship between environmental regulations, the introduction of HSRs, and officials' 337 political promotion incentives is examined. If the coefficient of HSR is significant, we proceed to the third step.

338
In the third step, the relationship among pollution, the introduction of HSR, environmental regulations, and

345
Step 1. Examining the impact of the introduction of HSRs and political promotion incentives on environmental 346 pollution: Step 2. Testing the impact of the introduction of HSR and officials' political promotion incentives on 348 environmental regulations: Step 3. Putting HSR, Pro_incent, and PITI into the regression equation: Step 4. Putting HSR, Pro_incent, PITI, and the interaction term PITI×Pro_incent into the regression equation:  Table 3. Column (1) shows the   represents the k ( k = 1, 2, 3, 4) years before the HSR is first connected. If the 371 observation unit includes the data from years k before the policy impact, the unit is set to 1; otherwise, it is 372 set to 0. Here, _ , + represents the z ( z = 1, 2, 3, 4) years after the HSR is first connected. If the 373 observation unit includes the data from z years after the policy impact, the unit is set to 1; otherwise, it is set 17 to 0.

375
The parallel trend test results are presented in Table 4. We found that the regression coefficient of HSR is 376 not significant in the 4 years before the introduction of HSR. After the HSR is introduced, the regression 377 coefficient of the HSR is significantly negative, indicating that the introduction of an HSR influences pollution 378 and the DID parallel trend assumption is satisfied. Based on the benchmark regression result and parallel trend 379 test, we can verify that the introduction of an HSR can reduce pollution emissions, and hypothesis 1 is supported.    were excluded from the regression due to missing lag variables; VIF was lower than 10; t-statistics in parentheses;

412
In column (3), we place the independent variable HSR and the mediating variable PITI into a regression. The

465
In column (2), we find that the coefficient of HSR is still negatively significant (β = -0.2296, p < 0.01),  were excluded from the regression due to missing lag variables; VIF was lower than 10; t-statistics in parentheses; * p < 478 0.10, ** p < 0.05, *** p < 0.01. We changed the proxy of the moderator variable, political turnover, to check the moderator mediation effect.

481
Political turnover indicates that mayors are placed in either more important (still at the mayoral level) or higher-482 ranked (vice-provincial level) positions, retiring, being transferred to another position at the same rank, or 483 termination. If a mayor is promoted, we regard his or her political promotion incentive as strong, and Pro_incent 484 is set to 1; otherwise, we regard his or her political promotion incentives as poor, and Pro_incent is set at 0. Table 9 shows robustness check result. In column (1), we examine the impact of HSRs on environmental 486 pollution, and the coefficient of HSR is significantly negative (β = -0.2181, p < 0.01). In column (2), we examine  observations were excluded from the regression due to missing lag variables; VIF was lower than 10; t-statistics in 499 parentheses; * p < 0.10, ** p < 0.05, *** p < 0.01.