The Empirical Research on the Impact of Pro-Environmental Factors on the Financing Cost of Green Bond

: Green bonds are an important part of green finance and a significant financing method 15 for enterprises to make socially responsible investments. This thesis analyzes the impact of pro- 16 environmental factors on the financing cost of green bonds by the data of green bonds issued from 17 2016 to 2020. The results show that the better the environmental performance of the issuer's region, 18 the lower the financing cost of green bonds, and the third-party certification reduces the financing 19 cost of green bonds. Further research shows that high pollution areas and high pollution industries 20 enhance the punitive role of environmental pollution financing. Regional environmental 21 performance mainly affects the financing cost of green bonds through tax suppression mechanism 22 and credit penalty mechanism, while third-party certification affects the financing cost of green 23 bonds through tax relief mechanism and financing channel mechanism. This paper provides 24 empirical evidence and policy inspiration for reducing the financing cost of issuing green bonds and 25 promoting the perfection of the green bond system.


Introduction
In December 2015, the People's Bank of China issued Announcement on Issues 45 Related to the Issuance of Green Financial Bonds, and Chinese National Development 46 and Reform Commission issued the Guidelines on the Issuance of Green Bonds, which 47 basically formed Chinese green bond issuance system and marked the beginning of the 48 development of the green bond market. By the end of 2020, the size of Chinese green 49 bond market had reached 278.662 billion yuan, and the number of issuance is more than 50 2019. The total size of labeled green bonds exceeding 1.4 trillion yuan, however, the 51 issuance of non-labeled green bonds totaled 1.67 trillion yuan, up nearly threefold year 52 on year ① 53 Green bonds are corporate bonds funding for green, circular and low-carbon 54 sustainable development projects, such as the energy conservation and emissions 55 reduction technology reform, green urbanization, clean and efficient use of energy, new 56 energy development and utilization, recycling economy development, water resources 57 saving and unconventional water resources development and utilization, pollution 58 control, ecological forestry, energy conservation, environmental protection, low carbon 59 industry, ecological civilization demonstration experiment in advance, pilot 60 demonstration and so on. ② There are so many factors affecting the financing cost of 61 green bonds. In addition to the general factors such as the issuer's credit rating, bond 62 maturity, market interest rate and so on, there are also specific factors unique to green 63 bonds, such as the environmental conditions of the issuer's area and third-party 64 certification, which are called pro-environmental factors in this thesis. 65 As a special kind of bonds, green bonds also have the general properties of bonds. Second, Suggestions on the development and system of green bond market, scholars 78 studying this part mainly conduct some quantitative analysis by improving the 79 information disclosure system, "labeling" issuance system and third-party certification 80 system of green bonds (Clapp 2014). Third, the research on the relationship between 81 green bonds and other assets, they mainly analyse the prices between other financial 82 products and green bonds, such as the price of conventional corporate bonds (Juan and 83 Andrea 2018), "black bonds" (David 2019) and stocks (Wang et al. 2020 By now, the research on the financing cost of green bonds mainly starts from the 89 market factors, and analyzes the characteristics of the green bond market, the 90 relationship between the relevant market factors and the financing cost of green bonds. 91 Mathews and Kiney (2010) studied the characteristics, scale, expected risk and other 92 general factors of the issuing enterprises. Febi et al. (2018) and Olivier (2018) 93 researched the relationship between the liquidity of green bonds and the financing cost, 94 and concluded that the stronger the liquidity of green bonds, the lower the financing 95 cost. Gao and Ji (2018) found in their research that the financial status of issuers has no 96 influence on the financing cost of green bonds, while the higher the rating of the issuer, 97 the lower the financing cost of green bonds. Diaz and Escribano (2021) analyzed the 98 differences between green bonds issued by green energy companies and non-green 99 energy companies, whose results showed that the ESG (environment, social 100 responsibility and corporate governance) standard and credit rating of the company had 101 an inverse relationship with the financing cost of green bonds. With the development 102 of green bond market, pro-environmental factors have gradually attracted the attention 103 of bond issuers, bond investors, government departments and regulatory authorities. So 104 more and more scholars have combined the environmental information contained in 105 green bond with its financing cost to conduct research. Eichholtz et al. (2019) found 106 that the bonds issued by the real estate trust with environmental certification in the 107 market have lower credit spreads, lower environmental risks and lower financing costs. 108 Wang et al. (2020) found that the stronger the social responsibility and the higher the 109 social reputation of the issuer, the lower the financing cost of the green bond issued. 110 Meanwhile, the more fully the green bond information is disclosed, the corresponding 111 financing cost will be reduced. The research of Pham and Huynh (2020) shows that 112 investor attention of environmental information of issuers is closely related to the 113 financing cost of green bonds. 114 We selected 543 green non-financial bonds issued from 2016 to 2020 that were 115 suitable for the definition of Guidelines on the Issuance of Green Bonds to analyze the 116 impact of pro-environmental factors on the financing cost of green bonds. It is found 117 that the lower the level of environmental pollution in the region where the issuer is 118 located, the lower the financing cost of green bonds; Third-party certification reduces 119 the financing cost of green bonds. The contribution of this paper is adopt an pro-120 environmental perspective to explore for the impact of pro-environmental factors on 121 the financing cost of green bonds, and to analyze the asymmetry of this influence under 122 different moderating effects and the specific influencing mechanism. 123 The rest of this thesis is arranged as follows. Section 2 puts forward research 124 hypothesis by the theoretical analysis. Section 3 describes research design including 125 data source, variable definition and model design. In section 4，we present the 126 empirical results and analysis, and the robustness test is put in section 5. Then, we 127 make further analysis, that is the moderating effect and the mechanism analysis which 128 are in section 5 and 6 separately. The last section shows research conclusions. 129 2.Theoretical analysis and research hypothesis 130 2.1 Regional environmental performance 131 For an area, the higher the level of pollution, the higher the cost of treatment. High 132 pollution will also attract the attention of the central government, leading to the central 133 government to reduce the fiscal bias or related investment, forcing the local government 134 to pay more attention to environmental protection and increase the punishment for 135 pollution, which makes the development of local enterprises subject to more 136 administrative restrictions. The increase of environmental risk and political risk will 137 lead to the increase of the issuance cost, that is, increase the financing cost of the local  138  issuance of green bonds. On the contrary, the stronger the local environmental  139  awareness, the more suitable it is for the goal of pro-environmental, the more investors  140 will be attracted, and the higher the environmental protection degree will be more in 141 line with the national policy, which will be subject to more fiscal bias, so the financing 142 cost of local green bonds will be reduced. At the same time, according to the theory of 143 ESG, when regional environment degrades, enterprises will face greater environmental 144 risks, the benefit of the parties involved will be damaged, the enterprise competitiveness 145 has also dropped, the enterprise value to fall, which will increase difficulty that 146 enterprises access to capital, the credit risk and liquidity risk will increase, credit 147 spreads will be bigger, so it to issue bonds financing costs will rise. At present, there 148 are few literatures on the relationship between regional environmental performance and 149 bond financing cost, but the existing scholars have given a consistent view on this 150 relationship: Painter (2020) 's study shows that regions affected by climate change need 151 to pay more costs to issue long-term municipal bonds, which will increase their issuing 152 interest rate; Cui et al. (2019) found in their research that the more investment in local 153 pollution control, the better the local pollution control effect and environmental 154 performance, the lower the environmental cost faced by enterprises, and the lower the 155 financing cost of environmental protection. Conversely, the less region invests in 156 pollution control, the higher the level of local pollution, the worse its environmental 157 performance, the higher the environmental costs faced by enterprises, and the higher its 158 financing costs. Zhang et al. (2019) found that the greater the pressure of regional 159 environmental governance, the easier it is to force local officials to urge enterprises to 160 invest in environmental protection. The above studies show that regional environmental 161 differences will affect the behavior of economic entities, and the degree of 162 environmental pollution will affect the cost of issuing bonds for enterprises. In other 163 words, the better the local environmental performance, the lower the financing cost of 164 issuing green bonds. Based on the above analysis, we propose the following hypotheses: 165 H1: The better the region's environmental performance, that is to say, the lower the 166 level of environmental pollution, the lower the financing cost of issuing green bonds. 167 168 Third-party certification, also known as the second party opinion, is the approval and 169 certification of green bond information disclosure by independent third-party 170 institutions as well as the assessment and disclosure of the environmental effects faced 171 by the issuer. Therefore, third-party certification can reduce information asymmetry, 172 more objectively reflect the pro-environmental characteristics of green bonds and 173 effectively show whether green bonds are "green-wash" or "true green". According to 174

Third-party certification
Notice on the Pilot Program of Green Bonds issued by Shanghai Stock Exchange and 175 Shenzhen Stock Exchange in 2016, "labeled" green bonds are green bonds are found 176 for the industries in which the funds raised by the bonds conform to Catalogue of Green 177 Bond Support Projects, as well as, are accepted, reviewed and uniformly labeled by 178 Chinese National Development and Reform Commission, Shanghai Stock Exchange 179 and Shenzhen Stock Exchange. The national policy does not enforce "labeling" of green 180 bonds, but does encourage. The introduction of the third-party certification make 181 "label" more authentic, can reflect more reality of the investment projects by green 182 bonds to raise funds, reduces the subjectivity of "label", enhance the credibility of the 183 green securities information disclosure, enhance the green bond credit and accord with 184 the requirement of ESG rating, which reduces the issue facing the enterprises credit risk 185 and liquidity risk. Hence the credit spreads of these green bonds will drop, which will 186 reduce the issuing cost of green bonds. However, the "labeled" green bonds without 187 third-party certification, without more objective information certification, cannot gain 188 the trust of investors, do not reflect the principles of ESG, and have no substantial 189 reduction in credit risk and liquidity risk, which will increase the financing costs of 190 green bonds. He et al. (2016) found in their research that financial certification of bonds 191 can alleviate information asymmetry and reduce bond financing costs. The financing 192 projects of green bonds are closely related to environmental risks. Third-party 193 certification can more effectively reflect information related to the environment, 194 improve the information transparency of green bonds, and thus gain recognition from 195 more investors, which will help reduce their financing costs. Jiang and Fan (2020), Yang 196 and Shi (2020) found that the "label" issue does not reduce the financing cost of green 197 bonds. Furthermore "label" green bonds without the third-party certification have 198 greater "green-wash" risk that increase the financing cost of green bonds. But the green 199 subsidy and third-party certification can reduce the financing cost of green bonds. 200 Based on the above analysis, we state second hypotheses: 201 H2: The financing cost of green bonds with third-party certification is lower than that 202 without third-party certification. 203 bonds that meet the definition of "green bonds" from 2016 to 2020. We have carried out 208 the following three aspects of processing based on the actual situation: (1) In the 209 duration of the bond, the issuer may issue additional bonds. We eliminate the additional 210 bonds and take the original bonds as the research object.

3.Research design
(2) There is a phenomenon 211 that corporate bonds are issued in one place and managed in two places, but two bonds 212 are actually one bond. Therefore, we exclude the green corporate bonds listed and 213 traded in the interbank market in cross-market trading, and only retain the green bonds 214 listed and traded in the exchange market.
(3) We exclude some private bonds with 215 undisclosed interest rate, financial bonds and incomplete data. 216 After the adjustment and processing of the above three aspects, we finally obtained 217 the observed sample values of 543 non-financial green bonds. In order to eliminate the 218 influence of extreme values, we conducted a bilateral 1% Winsorize for the main 219 variables in the paper. 220 We collect the characteristic information of sample bonds and the financial 221 characteristics information of issuers from the Wind database, the data of national debt 222 issuance and stock market yield from the CSMAR database, the data of regional 223 environment, economy and financial status from the statistical yearbooks of all 224 provinces and the macro-economic status from the statistical database of China 225 Economic Network. 226 and Fan (2020), namely, the credit spread between green bonds and the yield of 231 government bonds of the same maturity: 232

Variable definition
where rgp represents the credit spread of green bonds, rb indicates the interest rate of 234 green bonds issued, rg dontes the average interest rate of national debt of the same year 235 with green bonds (If not, the average interest rate of national debt of the adjacent years 236 is taken.). This proxy variable not only reflects the financing cost of green bonds after 237 deducting the risk-free interest rate of the same issuing period, but also reflects the 238 interest rate risk and market risk faced by investors as well as the risk preference of 239 investors. 240 3.2.2 Regional environmental pollution index 241 We refer to the regional environmental pollution index PI used by Li and Tao (2012), 242 Su and Lian (2018) as the proxy variable of regional environmental performance, and 243 the specific construction steps are as follows: 244 First, we chose the emissions of four types of pollutants to measure the pollution 245 situation in the region that are industrial sulfur dioxide, industrial smoke (powder) dust, 246 industrial wastewater and industrial solid waste. 247 Second, to construct a regional environmental pollution index, we calculated the 248 pollution situation two years before the issuance of green bonds. The specific 249 calculation steps are as follows: 250 First of all, we calculate regional pollutant emissions per unit of industrial output: 251 where Pollutionij denotes the discharge amount of the i pollutant in the j region 253 (province, autonomous region or municipality), GIPj represents the total industrial 254 Second, the emissions of per unit of industrial output of the i pollutant in the same 256 area are linearized: 257 PIj stands for the standardized value,and the regional pollution index. 278 Thirdly, according to the regional environmental pollution index, the paper divides 279 the area into high-pollution area and low-pollution area. The standard and result are 280 shown in Table 1. 281 Table1  In this paper, the Environmental Pollution Index of 31 provinces (municipalities and 283 autonomous regions) is calculated as average value, and the areas above the average 284 value are recorded as high pollution areas, otherwise as low pollution areas. bonds, the longer the maturity of bonds, the greater the risk and the higher the cost of 296 financing. Thirdly, the credit rating of bonds (rank) , which is considered to reflect the 297 specific credit of bonds, generally speaking, the higher the bond rating, the bond 298 repurchase right is the right granted by the bond contract to the bond issuer. When the 299 interest rate falls, the bond issuer can buy back the bond from the bond investor, 300 therefore, there is a positive correlation with the financing cost. financing. Thirdly, the credit rating of bonds (rank), which is considered to reflect the 311 specific credit of bonds, generally speaking, the higher the bond rating, the bond 312 repurchase right is the right granted by the bond contract to the bond issuer. When the 313 interest rate falls, the bond issuer can buy back the bond from the bond investor, 314 therefore, there is a positive correlation with the financing cost. 315 8 2. Financial characteristic variables of the bond issuers 316 The financial characteristic variable of the bond issuer is mainly used to measure the 317 ability to repay the debt, that is, the credit risk of the enterprise, but it can also reflect 318 the capital demand degree of the enterprise from another angle, thus the liquidity risk 319 that the enterprise faces. The financial characteristic variables of bond issuers in this 320 paper mainly include Yield valve(roe), EBIT(ebit), leverage (leverage) and net cash 321 ratio(fund). Generally speaking, the increase of the first two profitability indicators 322 indicates that the increase of the profitability of the enterprise and the decrease of the 323 default probability will strengthen the enterprise's credit and reduce the enterprise's 324 financing cost, while the higher the asset-liability ratio, the greater the credit risk and 325 the higher the financing cost, the larger the proportion of cash flow generated by fund 326 raising, the higher the cost of fund raising, the higher the cost of fund raising. (Hong 327 and Use one-phase lag(t-2),the rate of growth of tax revenue as a proportion of regional GDP loan regional credit Use one-phase lag(t-1),The logarithmic ratio of total regional credit to regional gross domestic product tax tax expenditure VAT/business profits payable bankloan bank loan The natural logarithm of the bank loan that the enterprise obtains years, the median is 5 years, and the maximum is 15 years, which indicates that most 368 of the green bonds issued in China are medium-and long-term bonds, it indicates that 369 the green bond issued by our country has a high credit rating. 370

372
In order to control the year effect and the industry effect, two kinds of fixed effects, 373 two-way fixed effect and combined fixed effect, are used in this paper. Table 4 shows 374 the baseline regression results: the first four are the regression results of the two-way 375 fixed effects model, and the last four are the regression results of the combined fixed 376 effects. Column (1) , column (2) and column (5) , column (6) are the unit regression 377 results of two core explanatory variables for the green bond credit spread (rgp) 378 respectively, the Regional Environmental Pollution Index (env) is positive at the level 379 of 10% significance in the two-way fixed-effect model, indicating that the lower the 380 level of environmental pollution in the region, the lower the financing cost of issuing 381 green bonds in the region, the baseline regression results support H1. Column (4) and 382 column (8) are the regression results after controlling variables are added. According to 383 the regression results of two kinds of fixed effect models, the regional Environmental 384 Pollution Index has a positive effect on the green bond credit spread under the level of 385 5% significance, that the lower the pollution level in a region, the lower the financing 386 cost of issuing green bonds, which supports H1; that third party certification is 387 significantly negative at a 10% confidence level, which means that third party 388 certification reduces the green bond credit spread, h2 is supported by the fact that third 389 party certification reduces the cost of financing green bonds. 390 and fourth, the difference between the interest rate of green bond issue and the 406 annualized yield of CNB00013( rgmp). 407 As can be seen from table 5, the results of the regression after the replacement of  408  the explanatory variables are similar to the original, and the control variables are  409 stable, so the benchmark regression is robust. Furthermore, in the two-way fixed-410 effect model, the third-party authentication significance increased to 5% , and the 411 significance of the joint fixed-effect model was consistent with the benchmark 412 regression. At the same time, the goodness of fit of the model of the last three 413 explained variables is gradually increasing, which indicates that the explaining 414 strength of environment-friendly preference factors and other controlled variables is 415 also increasing. 416   Table 6 shows the regression results for the replacement of the core explanatory 421 variable, certification by third parties (certificate). In this paper, we put the explanatory 422 variable of bigcertificate into the regression. From the results, we can get the following 423 conclusions: first, we compare the results of the unit regression, we can see that the 424 results of bigcertificate are more significant, the results of two-way fixed-effect 425 regression are increased to 10% , and the results of joint fixed-effect regression are 426 increased to 5% , comparing the results of baseline regression with the control variables 427 before and after the substitution variables, the significance of the core explanatory 428 variables in the two regression results is basically the same, and the control variables 429 are stable, which shows the robustness of the regression results, a comparison of the 430 regression coefficients between the pre-replacement core explanatory variable, 431 certification by a third party (certificate), and bigcertificate, after the replacement, 432 reveals that the regression coefficients for bigcertificate are significantly higher than 433 for third-party certification (certificate), green bonds certified by the top three 434 certification bodies have lower credit spreads and lower financing costs than those 435 certified by other non-top three certification bodies, this is with Jiang and Fan (2020) 436 the research conclusion is consistent. 437

454
The previous empirical study shows that regional environmental performance and 455 environmental-friendly preferences such as third-party certification will affect the 456 financing cost of green bonds, as the lower the regional environmental pollution index, 457 the lower the cost of financing for green bonds; the higher the cost of financing for 458 "Labelled"green bonds; and the lower the cost of financing for third-party certified 459 green bonds. The following two questions will be discussed in this paper: first, will the 460 influence of environment-friendly preference on green bonds change under different 461 moderating effects? Second, what is the specific mechanism by which the 462 characteristics of environment-friendly preference affect the financing cost of green 463 bonds? To answer the first question, this paper will examine the asymmetry of the 464 impact of environment-friendly preference on the financing cost of green bonds from 465 two aspects: the heterogeneity of regional pollution and the heterogeneity of industry 466 pollution. For the second question, this paper divides the impact mechanism of regional 467 environmental performance and third-party certification into two groups, considering 468 the impact of fiscal punishment mechanism, credit restraint mechanism, third-party 469 certification tax relief mechanism and financing channel mechanism on the financing 470 cost of green bonds. 471

Regulatory effect 472
5.1.1 Regional pollution heterogeneity 473 Research by Su and Lian (2018) found that under the impact of the Green Credit 474 policy, enterprises in highly polluted areas will not only find it more difficult to obtain 475 bank loans, but also their commercial credit will be weakened, the cost of its debt will 476 rise; meanwhile, businesses in less polluted parts of the country will have more access 477 to bank credit, and their business credit will increase, this shows that the higher the 478 degree of regional pollution will be proportional to the difficulty and cost of corporate 479 financing. 480 Table 8 presents the regression results for regional pollution heterogeneity.
(2) rank 481 (3) is the regression result of single environment-friendly variable under the regional 482 pollution heterogeneity, the regional environmental pollution index (PI)is significantly 483 negative under the 5% confidence level, but the third party certification (certificate) is 484 not significant; Column (5) is the regression result of two environment-friendly 485 variables under the heterogeneity of regional pollution when the control variables are 486 added. The regional pollution index (PI) is significantly negative at the level of 1% 487 significance, and the absolute value of the coefficient is greater than the absolute value 488 of the coefficient of the base regression, which shows that the highly polluted area has 489 the function of amplifying the regional environmental pollution and raising the 490 financing cost of the green bond, while the certification of the third party (certificate) 491 has become insignificant, this indicates that the regional environmental pollution in the 492 high pollution area has strengthened the function of punishing the financing cost of the 493 green bond. 494  difficult to obtain financing either through commercial credit or bank credit, financing 502 costs are higher than in the green and green sectors. 503 Table 9 presents the regression results of heterogeneous regulation of industry pollution. 504 Column (2) to column (3) is the regression result of the regulatory effect of regional 505 pollution heterogeneity on a single environment-friendly variable. The regional 506 environmental pollution index (PI) is significantly positive at 1% confidence level, third-507 party certification (certificate) was also positive at 1% confidence level; column (4) is a 508 regression of the regulatory effects of regional pollution heterogeneity on two 509 environmentally friendly variables after the control variables were added, the area pollution 510 index (PI) was significantly positive at the level of 10% significance, and the absolute value 511 of its coefficient was much larger than that of the standard regression, this suggests that 512 highly polluting industries have the effect of amplifying regional environmental pollution 513 and raising the cost of financing green bonds, while third party certification certification 514 (certificate) has become less significant, this indicates that the regional environmental 515 pollution of high pollution industries has strengthened the function of punishing the 516 financing cost of green bonds. 517

522
The model for the mechanism analysis used in this paper is as follows: 523 the variables of regional finance, regional credit, tax expenditure and bank loan 528 respectively, and the control variables are consistent with the benchmark regression 529 model. a1 represents the total effect of the tax suppression mechanism and the credit 530 punishment mechanism of the regional environmental performance, and a2 represents 531 the total effect of the tax relief mechanism and the financing channel mechanism of the 532 third party certification. c ' 1 is the direct effect of the tax suppression mechanism and 533 the credit punishment mechanism of the regional environmental performance, c ' 2 is the 534 direct effect of the tax relief mechanism and the financing channel mechanism of the 535 third party certification, a1*b is the intermediary effect of the tax suppression 536 mechanism and the credit punishment mechanism of the regional environmental 537 performance, and a2*b is the intermediary effect of the tax relief mechanism and the 538 financing channel mechanism of the third party certification. 539 5.2.1 Analysis on the mechanism of regional environmental performance 540 1. Regional tax restraint mechanism 541 The uncertainty of economic policy in a region to some extent will increase the 542 political risk of enterprises, especially for some enterprises with high emission, the 543 easier it is for them to become the object of government's monitoring, thus increasing 544 the tax revenue of the enterprises, similarly, the uncertainty of tax policy will aggravate 545 the tax burden of enterprises and increase the financing cost of enterprises, while the 546 uncertainty of fiscal tax policy and the decrease of fiscal revenue, increased pressure 547 on local governments to increase revenue will also increase the cost of corporate 548 financing. Liang et al. (2006) took taxation one step further by showing that 549 implementing environmental taxes can not only improve the local environment and 550 achieve sustainable development, but also increase income tax and labor-related tax 551 revenue under the reduction, to improve the efficiency of economic development and 552 promote economic growth. Local governments, then, are more willing to raise taxes 553 under such incentives to achieve "Win-win". 554 Table 10 shows the regression results of the regional tax penalty mechanism. The 555 coefficient of the Environmental Pollution Index (PIt-2) in column (2) is significantly 556 positive at the confidence level of 1% , which indicates that the indirect effect of the 557 intermediary effect exists and is positive. In column (3) the tax revenue growth (Taxt-2) 558 coefficient is significantly positive under the 1% confidence level, which indicates that 559 the direct effect of the intermediary effect exists and is positive. Therefore, the 560 regression result shows that the intermediary effect is positive, which shows that the 561 regional environmental pollution increases the financing cost of the local enterprises 562 issuing green bonds by reducing the local fiscal revenue. 563 Green Credit Guidelines, which formally put forward the concept and role of "Green 573 Credit"and explicitly proposed the use of credit to promote energy conservation, 574 emission reduction and environmental governance, we will guide the rational allocation 575 of credit resources and accelerate the development of green industries and the 576 restructuring of the economy. After the promulgation of the Green Credit policy, the 577 regions with better environmental performance will get more credit resources, while the 578 regions with worse environmental performance will have less credit resources. June 579 Chan (2019) found that the "Two high one leftover"enterprise loans by the Green 580 Credit Guidelines of the greater the impact of the repression, the higher the cost of 581 financing. Green credit constraints are even more severe for heavily polluting 582 enterprises and heavily polluted areas, and illiquid debt financing has declined 583 significantly, which inhibits credit financing for heavily polluting enterprises and 584 heavily polluted areas, resulting in an increase in credit financing costs. The cost of 585 bond financing is also affected by other financing methods. Credit financing is an 586 important financing method. When a regional environment performs well, when 587 companies in a region have better access to bank credit resources, that is, when the cost 588 of credit financing is lower, the cost of financing the issue of green bonds is 589 correspondingly lower; and when a regional environment is performing poorly, 590 companies in the region will find it more difficult to obtain credit from banks, which 591 means that the cost of financing will be higher, as will the cost of other means, including 592 green bonds. 593 Table 11 shows the regression results of regional tax penalty mechanisms. In column 594 (2) the regression coefficient of the environmental pollution index (PIt-2) is significantly 595 negative at 1% confidence level, which indicates that the indirect effect of the 596 intermediate effect exists and is negative; The regression coefficient of the regional 597 credit (loant-1) in column (3), loant-12, is significantly negative at 1% confidence level, 598 indicating that the direct effect of the intermediary effect exists and is negative. 599 Therefore, the regression results show that the intermediary effect is positive, and that 600 regional environmental pollution leads to the increase of the cost of various financing 601 methods by reducing the total amount of regional credit through regional environmental 602 pollution, this has raised the cost of financing for local companies to issue green bonds. 603 In the traditional microeconomics hypothesis, the firm pursues profit maximization, 610 that is to say, when the price is fixed, the firm pursues cost minimization under the 611 condition of given output, and the firm pursues output maximization under the 612 condition of given cost. Assuming that the economy is perfectly competitive and the 613 firm produces only one product, the profit to be gained by the firm is: 614 Profit = (unit price-unit cost) * product sales (11) 615 According to the theoretical analysis above, in real economic life, in order to prevent 616 and control pollution, save energy and reduce emissions, and promote environmental 617 protection, the government adopts administrative measures to achieve environmental 618 achievements, in particular, the government uses administrative punishment, that is, 619 fines or heavier taxes to control pollution, and uses administrative subsidies, such as 620 government subsidies and tax breaks, to promote environmental protection. In this 621 paper, for the sake of simplification, administrative fines are regarded as increasing tax 622 revenue, while government subsidies are regarded as reducing tax revenue and taxing 623 the price of individual products: 624 Profit = (unit product price-unit cost-unit product tax) * product sales (12) 625 The formula (13) shows corporate profits with taxes. When companies meet 626 environmental requirements, they are more likely to receive tax breaks from the 627 government, increase their profits, improve their financial position, have less difficulty 628 in obtaining financing, and have lower financing costs. 629 Table 12 shows the regression results of the tax relief mechanism. In the column 630 (2) the regression coefficient for third-party certification (certificate) was positive at 631 the 5% significance level, that is, the issuance of a third certified green bond can 632 increase the financing constraint index (FC), indicating that third-party certification can 633 alleviate the financing constraint of the issuer, it shows that the indirect effect of 634 intermediary effect exists and is positive, and the regression coefficient of the financing 635 constraint index (FC) is negative at the level of 1% significance, that is, the looser the 636 financing constraint, the lower the financing cost of green bonds, it shows that the direct 637 effect of mediating effect exists and is negative. Therefore, the result of regression 638 shows that the intermediary effect is negative, which shows that the listed third party 639 certification reduces the financing cost of the green bond issuers by easing the financing 640 constraints of the issuers. 641 Enterprise's credit financing belongs to the enterprise debt financing channel, also is 646 one of enterprise bond financing alternative ways. When Enterprises obtain more bank 647 loans, which shows that their credit is rising, the availability of other debt financing is 648 enhanced, the financing channels available to enterprises are gradually broadened, and 649 their financing costs are correspondingly reduced. In particular, following the 650 introduction of the Green Credit policy, the other debts of the enterprises that had access 651 to green credit increased significantly while the cost of financing decreased. According 652 to the above analysis, the third-party certification of green bonds meets the 653 requirements of ESG, embodies the concept of green finance, and is one of the 654 manifestations of corporate social responsibility in green credit policies, so the third-655 party certification can increase the bank loan, especially the green credit, thus widen 656 the financing channel of the enterprise, and then reduce the financing cost of issuing 657 the green bond. Where shortloan and longloan represent banks'short-term and long-term loans to 662 businesses, respectively, with all variables measured in billion dollars. 663 Table 13 shows the regression results of the financing channel mechanism. In column 664 (2) the regression coefficient for third-party certification (certificate) was positive at 665 the 10 per cent significance level, that is, the issuance of a third-party certification green 666 bond increased bank credit by bankloan, indicating that third-party certification 667 increased bank credit for businesses; In column (3) the regression coefficient of the 668 bankloan is negative at the 1% significance level, that is, the more bank credit the 669 enterprise gets, the lower the financing cost of the green bond is, it shows that the direct 670 effect of mediating effect exists and is negative. Therefore, the regression results show 671 that the intermediary effect is negative, which shows that the issuance of the third-party 672 certified green bonds enhances the credit of the issuer and increases the bank credit of 673 the issuer, this reduces the financing cost of the green bond issuers. 674

Conclusions and policy recommendations
678 This paper shows that environment-friendly preference factors will affect the 679 financing cost of green bonds, and the specific conclusions are as follows: 680 First, the degree of pollution in a region is positively correlated with the financing 681 cost of green bonds in that region, and the high degree of pollution in that region has a 682 disciplinary effect on the financing of green bonds issuing enterprises in that region, the 683 environment-friendly areas, that is, the low-pollution areas, have an encouraging effect 684 on the financing of the green bond issuers in these areas. 685 Second, the third-party certification contains more environmental information, 686 investors prefer it, and the certified green bond financing costs are reduced, and third-687 party certified green bonds from the top three are cheaper to finance than those from 688 other institutions. 689 Third, green bonds issued in highly polluted areas also magnify the impact of 690 regional pollution on raising financing costs and weaken the role of third-party 691 certification in reducing the financing costs of green bonds, enterprises in high-692 pollution areas and high-pollution industries do not meet the requirements of social 693 responsibility, their financing costs are higher. 694 Fourth, regional environmental pollution affects the financing cost of green bonds 695 through fiscal penalty mechanism and credit restraint mechanism, that is, regional 696 pollution index affects the financing cost of green bonds through increasing regional 697 tax revenue and reducing the total amount of local credit, thus raising the financing cost 698 for local enterprises to issue green bonds; at the same time, the third-party certification 699 factors affect the financing cost of green bonds through the mechanism of tax relief and 700 financing channels, that is, enterprises issue green bonds with third-party certification 701 by reducing their tax expenses and widening their financing channels, thus reducing the 702 financing cost of the main issue of green bonds. 703 According to the influence of environment-friendly preference factors on the 704 financing cost of green bonds, this paper puts forward the following policy suggestions 705 to form reasonable financing cost of green bonds and perfect China's green bond 706 market. 707 First of all, improve the green bond market third-party certification system and 708 information disclosure system. Improve the third party certification standards, so that 709 both with the international track, and combined with national conditions. We will 710 accelerate the formulation and verification of relevant standards for the disclosure of 711 information on green bonds, truly reflect the funds invested in green bonds, that is, the 712 operation of the projects, and allow more financial resources to be invested in projects 713 that are in line with the catalogue of Green Bond support projects. 714 Second, foster institutional investors with environmentally friendly preferences. To 715 guide investors to pay more attention to the environmental risks of the issuers of green 716 bonds and capital operation projects, so that more funds will be invested in the 717 environmental protection and low-carbon industries and less in the polluting industries, 718 environmentally friendly preferences for realizing capital flows. Industries that meet 719 the needs of long-term economic and social development, such as new energy industry, 720 low-carbon transportation industry, environmental protection construction industry and 721 other low-pollution and environmental protection industries, will be sustainable 722 development; and not in line with economic development, causing greater damage to 723 the environment of high pollution, high energy-consuming enterprises, will be 724 eliminated. 725 Third, to give full play to the role of local governments, local governments should 726 strengthen the supervision of polluting enterprises and subsidies to environmental 727 protection enterprises, reduce the tax burden of environmental protection-oriented 728 enterprises, and at the same time improve the local market environment and improve 729 the local credit system, thereby reducing the financing cost of enterprises issuing green 730 bonds. 731 Fourth, enterprises should practice the concept of social responsibility investment 732 and implement the requirements of ESG. Enterprises need to implement the spirit of 733 green financial policy and constantly strengthen their awareness of social responsibility, 734 so as to conform to the trend of social sustainable development, tax expenditure can be 735 reduced or exempted, financing channels can be broadened and more government 736 subsidies can be obtained, thus reduces each kind of financing way including the Green 737 Bond, the financing channel financing cost. 738 739 740 741 742