How does green finance affect cleaner industrial production and end-of-pipe treatment performance? Evidence from China

Effectively identifying the role and mechanism of green finance in environmental governance provides an important guarantee that green finance serves the ecological environment. Based on the panel data of 30 provinces in China from 2001 to 2015, this paper explores the impact of green finance on cleaner industrial production and end-of-pipe treatment and further reveals the mediating effect of industrial structure optimization and the moderating effect of environmental regulation. The results show that (1) China’s cleaner industrial production performance, end-of-pipe treatment performance and systematic governance performance show a clear upwards trend, distributed in stages in the eastern, central and western regions from high to low. (2) At the national level, green finance promotes cleaner production performance but inhibits end-of-pipe treatment performance. However, it can be seen from the results of the sub-sample, the eastern and central regions are consistent with the overall effect, with heterogeneous effects in the western region. (3) The optimization of industrial structure plays a partial intermediary role in the impact of green finance on cleaner production and end-of-pipe treatment. (4) Both “market-incentive” and “command-and-control” environmental regulations weaken the positive impact of green finance on cleaner production; “market-incentive” environmental regulation alleviates the negative impact of green finance on end-of-pipe treatment, while the moderating effect of “command-and-control” environmental regulation on end-of-pipe treatment is not significant.


Introduction
Environmental pollution is not only related to the happiness index of human beings but also directly represents the high-quality development of a country's economy. Over the past 40 years of reform and opening, China's economy has created a miracle of rapid growth over a long period of time in which industrialization has played a key role (Huang et al. 2020 advancement of industrialization, resource extraction and environmental problems caused by a growth pattern of high inputs and high energy consumption have become the "shackles" of social and economic development (Shao et al. 2019). Especially over the past 10 years, the negative impact of environmental degradation has been prominent, ranging from ecological destruction and health loss to agricultural productivity decline and global warming. Meanwhile, the Chinese government has gradually realized the importance of environmental governance and has promulgated a series of environmental policies that have improved environmental quality to a certain extent . Even so, China's economic system remains fragile, and its environmental carrying capacity is still facing enormous challenges (Ronald et al. 2017). In the face of this severe environmental situation, in October 2017, the Central Committee of the Communist Party of China strategically deployed the promotion of green development and acceleration of the reform of ecological civilization in the report of the 19th National Congress of the Communist Party of China. The party further clarified the importance of improving environmental quality at the fifth plenary session of the 19th Central Committee of the Communist Party of China, and systematically elaborated on the key tasks and main measures to reach this goal. At present, China's environmental governance is in a critical stage of "pressure multiplication", and there is an urgent need to develop appropriate environmental governance tools that can not only promote economic growth but also contribute to environmental protection (Zha et al. 2020;Mikhno et al. 2021).
As an important starting point to promote the sustainable development of the regional economy and environment, green finance can improve the operational efficiency of the market through resource integration and promote the green transformation of industry with the help of financial integration functions, effectively ensuring the advancement of regional environmental governance (Yu et al. 2021). Developed economies such as the USA and the European Union are focusing on how to build green financial systems to be the players in climate change mitigation and adaptation (Amighini et al. 2022). Scholars from developing countries such as China, India and Iran have also published a large number of articles and built extensive international cooperative relations. Due to the differences between China and developed economies in terms of financial market operation mechanisms and economic policy implementation goals, China pays more attention to the role of green finance in economic growth, pollution control, energy conservation and emission reduction (Wang et al. 2022). Domestic and foreign research on green finance has also spawned research areas such as green innovation, green productivity, corporate environmental responsibility, green investment, green credit and green credit policy (Fahim and Mahadi 2022). In general, the overall work on green finance is still limited, which makes the mechanism between green finance and environmental governance a cutting-edge research direction. At present, China's green finance business scale ranks among the highest in the world, becoming an essential part of the global green finance system, and its influence has gradually expanded from China to countries along the "Belt and Road" . Exploring a viable green finance development model in China will not only solve China's serious environmental problems, but also provide other countries with China's experience and solutions for green transformation. However, there is a lack of literature that empirically tests the environmental governance performance in tandem with China's green finance and a lack of literature that conducts an in-depth discussion of how green finance works. These deficits make it difficult to provide a scientific decision-making basis for promoting regional environmental governance with the help of green finance.
To better solve the increasingly urgent issue of promoting environmental governance through green finance, this paper summarizes and analyzes the research of domestic and foreign scholars on the relationship between green finance and environmental governance and specifically carries out the following work: (1) it constructs an indicator system for green finance from the four dimensions of green credit, green securities, green insurance and government support and uses the entropy method to calculate the green finance development index of each province in China from 2001 to 2015; (2) industrial environmental governance is divided into two stages: cleaner production and end-ofpipe treatment, and the data envelopment model of the two-stage network DEA is used to measure the systematic environmental governance performance, cleaner production performance and end-of-pipe treatment performance for industrial systems in each province in China from 2001 to 2015; (3) the dynamic relationship between green finance and environmental governance performance is empirically analyzed by using a fixed-effects model, and the mediating role played by industrial structure optimization is further explored; (4) in view of the fact that it is difficult for a single environmental policy to effectively guarantee the implementation of industrial environmental governance, different types of environmental regulations are introduced to examine the moderating effect between green finance and environmental governance.
The contribution of this paper is mainly reflected in the following: (1) compared with the existing singledimensional green finance evaluation indicators, we rely on green credit, securities, insurance and government funds to build a multidimensional indicator system for green finance, which ensures that the long-term trends of indicators are presented and makes the evaluation results more systematic and complete; (2) different from the existing research, which mainly measures the performance of environmental governance from a single stage, we decompose China's environmental governance into two stages of cleaner production and end-of-pipe treatment, and based on the two-stage network DEA method, the performance of the decomposition process of China's environmental governance from 2001 to 2015 is measured. This analysis makes it easy for us to characterize the dynamic changes in the effectiveness of environmental governance and reveal the possibility of the win-win outcome of economic growth and environmental protection at different stages of development; (3) it is the first time that green finance and the decomposition performance of industrial environmental governance have been incorporated into the same research framework. Based on exploring the connection between green finance and the decomposition performance of environmental governance, we further test the intermediary role of industrial structure optimization and the moderating effect of environmental regulation, thus providing important theoretical value and practical significance for opening the "black box" of the connections between regional environmental governance and green finance.
The structure of the rest of this paper is as follows. "Literature review" section reviews previous research. "Empirical model and methodology" section introduces the empirical model and methodology. "Data description" section describes the data for this study. "Results and discussion" section discusses the empirical findings. "Research conclusions and policy implications" section summarizes the findings and puts forward corresponding policy recommendations.

Literature review
(a) Scholars mainly use qualitative or quantitative analytical methods to evaluate the operation of green finance. In qualitative analysis, some international organizations, such as the International Finance Corporation (Kyte 2008), the World Wildlife Fund (China Banking Regulatory Commission 2014) and the Organization for Economic Co-operation and Development (OECD 2007), aim to guide financial institutions to provide better green financial services by judging the effectiveness of financial institutions in implementing green responsibilities. In quantitative analysis, some scholars use the perspective of financial institutions. Penny and Monaghan (2001) evaluated the energy conservation and environmental protection performance of financial institutions in operations and management. Marcel (2001) selected several financial institutions around the world as research samples and examined the stage of green finance development of different financial institutions. Hafner et al. (2020) and Dogan et al. (2022) chose the progress on low-carbon financial flows and the S&P Green Bond Index as alternative indicators for green finance, respectively. Chinese scholars mainly choose a single indicator or build a comprehensive indicator system to evaluate the development level of green finance. Wen et al. (2022) regarded the ratio of financial resources in the environmental protection industry to all industrial financial resources as an alternative indicator of green finance. Xie (2021) took the coupling and coordination degree of regional financial development and green development as a proxy for green finance. Wang and Wang (2021a) believe that green credit occupies the largest weight in the green financial system, so the proportion of green credit to GDP is used to characterize the development level of green finance. However, some scholars believe that the use of a single indicator for multiple regions will obscure the relevant evaluation of green finance development; it is difficult to reflect the essence of green finance development, and the construction of a comprehensive green finance indicator system can better reflect the true level of green finance (Lee and Lee 2022). Guo (2022) selected green credit, green investment, green venture capital and government support to conduct a comprehensive assessment of China's green finance development level from 2001 to 2018. Lee and Lee (2022) believed that green credit, green securities, green insurance, green investment, and carbon finance more comprehensively cover the connotations of green finance, and on this basis, they calculated China's green finance development index for 2006-2018.
(b) Data envelopment analysis (DEA) and stochastic frontier analysis (SFA) are the main methods for scholars to assess environmental governance performance (Ghosh and Kathuria 2016). Since DEA has advantages over SFA in terms of function form setting and evaluation of homogeneous decision units, DEA or its extended model is usually used for efficiency analysis (Shao et al. 2021). The existing environmental governance performance measurement mainly adopts single-stage and multistage DEA models. The single-stage DEA model mainly includes the traditional DEA model considering undesirable output (Sueyoshi and Yuan 2015;Avilés-Sacoto et al. 2021), the slack-based SBM-DEA model (Xie et al. 2017), the DEAbased Malmquist productivity index model that considers cross-period environmental governance performance (Oh and Heshmati 2010;Miao et al. 2019), and various extended DEA models (Chen and Delmas 2011;Dimaria 2014;Wang et al. 2018). In the single-stage DEA model, the entire process from production to pollution treatment is still a "black box", which makes it difficult to reflect the real situation of environmental governance performance. The proposal and development of the network DEA model provides a possibility to open the "black box" of environmental governance from a multistage perspective. Färe and Grosskopf (2006) proposed from a theoretical perspective that the network DEA model can provide a unified framework for evaluating environmental governance performance. Hampf (2014) uses the network DEA model to decompose environmental governance efficiency into production efficiency and emission reduction efficiency from the application level, realizing a leap from theory to practice. Song et al. (2018) took the lead in putting a two-stage look at environmental governance performance into the actual production process in China and chose the expanded network SBM model as a new tool to analyze the internal environmental governance performance of the system. Wang and Feng (2020) and Shao et al. (2021) further calculated overall efficiency and separate efficiencies of three substages of wastewater, waste gas and solid waste treatment based on measuring the total factor productivity and environmental governance efficiency of the industrial system in the two connected stages.
(c) Green finance and environmental governance have formed a symbiotic relationship of mutual influence and interdependence. Ionescu (2021) believes that green finance is an important tool for both developed and developing countries to cope with global climate change and regional environmental governance. Lee (2020) believes that the development of green finance is an inevitable requirement for realizing the coordinated progress of the economy, society and the environment and achieving sustainable development. Ng (2018) pointed out that green finance is an economic activity that can support the improvement of the ecological environment and improve the efficiency of resource use to effectively combat climate change. Mohd and Kaushal (2018) believe that green finance focuses on environmental protection and pollution control and balances the contradictory relationship between economic development and environmental governance through the rational allocation of financial resources. Zhou et al. (2020) examine the impact of green finance on economic growth and environmental quality by constructing empirical models, and the research results show that the development of green finance promotes economic growth and environmental governance to achieve a "win-win situation".
In the process of green finance promoting environmental governance, industrial structure optimization plays a key role as an intermediary (Wang and Wang 2021b). Relevant studies show that green finance is a booster for the improvement of industrial structure, and there is a dynamic coordination relationship between industrial structure and green finance (Mahat et al. 2019). First, the environmental information disclosure and capital supervision system upheld by green finance has aggravated the debt financing costs of traditional industrial enterprises, and while promoting the transformation of traditional industrial enterprises to green enterprises where possible, green financing conditions have forced traditional industrial enterprises that cannot go green to stop production and withdraw from the market (Shi et al. 2022;Yu et al. 2021). Second, green finance guides the flow of funds to industrial enterprises that focus on green development, creates the scale of green funds by lowering the investment threshold and increasing financial support, and encourages industrial enterprises to increase investment in clean technologies, thereby promoting the development of green industries (Jiang et al. 2022).
In addition, environmental policy coordination will integrate development among institutions related to the environmental system and effectively integrate external resources, policy objectives and policy measures (Herrera et al. 2019). At present, the "command-and-control" environmental regulation represented by laws and regulations and the "marketincentive" environmental regulation represented by environmental taxes and sewage charges play a key role in the practice of environmental governance, and it is uncertain whether they can promote environmental governance in synergy with green finance. On the one hand, there are complementary effects between environmental regulation and green finance (Bolton and Foxon 2015). Environmental regulation directly restricts traditional industrial enterprises by setting access thresholds and governance standards, and green finance focuses on cultivating green industries, with a view to improving the cleaner production efficiency of industrial enterprises through controlling capital and promoting a virtuous cycle of capital ). On the other hand, green finance and environmental regulation have similar mechanisms of action (Zhu et al. 2021). Among them, "command-and-control" environmental regulation controls traditional industrial enterprises by setting regulatory thresholds, and green finance limits the threshold for industrial enterprises to obtain green funds by setting environmental protection standards. "Marketincentive" environmental regulations encourage industrial enterprises to reduce pollutant emissions through cleaner production technologies, while compliant industrial enterprises can shift the cost of paying sewage charges to the research and development of cleaner production technologies and promote the all-round governance of industrial production. Green finance opens up "green channels" for industrial enterprises that focus on the use of clean technologies, alleviating their financing pressure by increasing loan amounts, reducing loan interest rates and enhancing their willingness to carry out clean technology innovation (Wang and Xu 2015).
Through the review of the above literature, we find that there are still the following problems: (1) scholars mostly use a single indicator such as green credit, green investment, green securities, and green insurance to indicate the level of green finance development in different regions, and few studies have conducted a comprehensive evaluation by building a systematic indicator system.
(2) Previous studies have mainly used a single-stage DEA model to measure environmental governance performance, and the evaluation of environmental governance performance must be multistage, and treatment of the entire process from production to pollution treatment as a "black box" will produce estimation bias. (3) At present, domestic and foreign scholars have conducted in-depth research on the measurement of green finance and environmental governance performance. We have reached a consensus on the symbiotic relationship between green finance and environmental governance. However, what impact will green finance have on environmental governance performance? Can green finance work with existing environmental regulations to promote environmental governance? Scholars at home and abroad lack relevant research on such issues.
In view of the shortcomings of the existing literature, this paper carries out the following work: (1) based on the mainstream research views of scholars at home and abroad and considering the availability of data, this paper constructs an indicator system of green finance based on the four dimensions of green credit, green securities, green insurance and government support, which reflect long-term changes to green finance while maintaining relative comprehensiveness.
(2) We divide environmental governance into two stages of cleaner production and end-of-pipe treatment and uses the two-stage network DEA model to establish an indicator system for environmental governance performance, trying to open the "black box" of regional environmental governance. (3) Based on examining the relationship between green finance and environmental governance performance, we further examine the intermediary role of industrial structure optimization and the moderating effect of environmental regulation.

Two-stage network DEA model
As shown in Fig. 1, this paper decomposes industrial environmental governance performance into cleaner production performance and end-of-pipe treatment performance. By constructing a two-stage network DEA model, we evaluate the state of interprovincial environmental governance in China from 2001 to 2015. The inputs of the cleaner production stage are the industrial capital stock, the number of industrial employees and industrial energy consumption. The outputs of the cleaner production stage are the gross industrial output value and the volume of three kinds of industrial wastes, which is also used as the input of the end-of-pipe treatment stage. Other inputs in the end-of-pipe treatment stage are the number of full-time environmental protection personnel and the capital stock of the three industrial waste treatments. The output of the end-of-pipe treatment stage is the volume of three industrial waste treatments. It should be noted that the volume of industrial waste water is equal to the volume of treated waste water and the amount of discharged waste water, and the volume of industrial waste gas is equal to the volume of gas emissions and gas removal. Since industrial NOx emissions were not counted before 2011, the industrial exhaust gases included in this article only include industrial sulfur dioxide, soot, and dust. In addition, the industrial capital stock is expressed as the annual average of the net industrial fixed asset value. The capital stock of the three industrial waste treatments refers to the study of Caselli (2005), which sets the depreciation rate to 6.0% and is measured by the perpetual inventory method.
In this paper, the performance of environmental governance is calculated based on the two-stage network DEA model proposed by Wu et al. (2016). Suppose there are n decision-making units, each of which is labelled DMU j (j = 1, 2, · · · , n). The cleaner production stage has m 1 inputs The two-stage system performance (SP ) calculation for decision-making unit d is as follows: Based on Eq. 1, the first stage of performance measures is as follows: Based on Eq. 1, the second stage of performance measures is as follows:

Benchmark model
The benchmark regression model is as follows: where the subscripts i and t represent the province and year, GF represents the level of green finance development, CP P represents cleaner production performance, EP P represents the end-of-pipe treatment performance, X represents the control variable of each province, σ i represents the fixed effect of the province, λ t represents the fixed effect of time, ε it represents the random error term, and α 1 , α 2 and α are the coefficients of GF , I SR and X, respectively. Note that although I SR is a mediating variable in Eqs. 4 and 5, it is still treated as a control variable, and it is marked separately to facilitate the interpretation of its coefficients in the mediating effect model section.

Mediating effect model
Effectively identifying the mechanism of green finance on environmental governance performance is of great significance for accelerating green transformation. This study uses stepwise regression to verify the mediating variable of industrial structure optimization. Equations 4, 6, 7 and 5, 6 and 8 list the test steps of industrial structure optimization in green finance for cleaner production performance and end-of-pipe treatment performance, respectively.
where I SR represents industrial structure optimization, β 1 and γ 1 represent the coefficients of GF , and β and γ represent the coefficients of X. It should be noted that Eqs. 7 and 8 are different from Eqs. 4 and 5; the difference is that the control variables of Eqs. 4 and 5 contain I SR, while the control variables of Eqs. 7 and 8 do not contain I SR, and the rest of the variables remain consistent. The specific mediating effect judgement method refers to the research of Wen and Ye (2014).

Moderating effect model
Unlike the mediating effect, the moderating effect is a test of whether green finance is affected by other relevant variables when it has an impact on environmental governance performance. This paper examines the moderating effect of different types of environmental regulations in the mechanism of green finance influencing environmental governance performance and then provides an important value reference for the coordinated formulation of regional environmental regulation policies and green finance policies. We refer to the practice of Balli and Sørensen (2013) and decentralize the interaction terms. The specific moderating effect model is set as follows: where GF i,t , MER i,t and CER i,t represent the average of green finance development levels, "market-incentive" environmental regulations, and "command-and-control" environmental regulations, respectively, and α 4 represents the regression coefficient of the interaction term between green finance and environmental regulation after subtracting the sample mean.

Explained variable
Cleaner production performance (CP P ) and end-ofpipe treatment performance (EP P ) are the measurement results of the cleaner production stage and the end-ofpipe treatment stage of the above two-stage DEA model, respectively. In the benchmark regression results, CP P and EP P are obtained based on Eqs. 2 and 3, and the detailed calculation results are shown in "Results and discussion" section.

Core explanatory variable
To present the long-term dynamic trend of green finance development, this paper sets green credit, green investment, green insurance, and government support as first-level indicators and constructs the green finance development index of each provincial-level region. Among them, green credit is represented by the ratio to GDP of the interest expenses of the six high-energy-consuming industries, green investment is represented by the ratio to GDP of investment in pollution control, green insurance is represented by the ratio of agricultural insurance revenue to total agricultural output value, and government support is represented by the ratio of fiscal environmental protection expenditure to fiscal general budget expenditure. Based on determining each indicator layer, this paper uses the entropy method to measure the development level of green finance in each provincial-level region, and the specific index calculation is carried out in three steps.
First, each indicator is standardized through mathematical transformations to eliminate the problem of incommensurability between indicators. To this end, indicators are divided into positive and negative groups, which are processed using the formula of the range value between groups. Therefore, the standardized treatment is as follows: Negative indicator: where x θij is the original value of the j th indicator of the ith province in the θth year and x max and x min are the maximum and minimum values of the j th indicator of each provincial-level region in China, respectively. Second, after the standardization of each indicator, a weight matrix w j is established, and the indicators are weighted according to the relative importance of each indicator in the evaluation process. Third, using the standardized values and the weight of each indicator, the comprehensive index of green finance in each province is obtained: where GF θi is the development level of green finance of the ith province in the θ th year.

Mediation variable
Industrial structure optimization (I SO). Industrial structure optimization is an important indicator to measure the transformation and upgrading of regional industrial structure, and previous studies have only used a structural deviation index that compares industrial structure and employment structure to measure the degree of industrial structure optimization. However, this structural deviation index gives different industries the same weight, ignoring the heterogeneity of the economy overall. As suggested by Cheng et al. (2018), we use the reciprocal of the Thiel index to indicate industrial structure optimization. The calculation is as follows: where Y represents the output value, L represents the number of workers, and n represents the number of industries.

Moderating variable
"Command-and-control" environmental regulation (CER) and "market-incentive" environmental regulation (MER) are used as moderating variables. We use the number of environmental impact assessment programs and "Three simultaneous" programs as the proxy for the "commandand-control" environmental regulations and use the ratio of the sewage charges to GDP as the proxy for the "marketincentive" environmental regulation.

Control variable
Income level (P GDP ). We use real GDP per capita to characterize regional income level, and the specific indicator is calculated based on the use of a GDP deflator to convert nominal GDP into real GDP and then divide by the average local population at the end of the year. The environmental Kuznets curve (EKC) hypothesis suggests an inverted U-shaped curve between income level and environmental quality. To further validate the existence of the EKC hypothesis, we add GDP per capita and its square term to the model. If the EKC hypothesis holds, the coefficient for PGDP should be positive, and the coefficient for its square term should be negative, indicating a decoupling relationship between China's income level and environmental quality .
Urbanization level (U R). The rapid aggregation of labour and resources has led to a surge in energy consumption demand, which has hindered regional environmental governance (Chatti and Majeed 2022). In this paper, we use the ratio of the urban population to the total population to characterize the urbanization level, and its impact on environmental governance performance is negative.
Foreign Trade level (T RAD). The hypothesis of "the Environmental Gains from Trade" holds that foreign trade brings advanced clean technology and management experience to host countries and improves the efficiency of capital and resource utilization, thus having a positive impact on environmental governance performance (Caldwell and Vogel 1996). The hypothesis of "race to the bottom" posits that attracting foreign investment by lowering environmental standards will increase the pressure on regional environmental governance, thereby having a negative impact on environmental governance performance (Frankel and Rose 2005). Therefore, we measure the foreign trade level by the ratio of total import and export trade to GDP, and the direction of its impact on environmental governance performance is uncertain.
Economic Growth Target (EGG). Regional economic growth targets are set with the direct involvement and responsibility of the officials, which is both a centralto-local assessment criterion and a local commitment to the central government's performance. When the actual economic growth rate deviates from the set growth target, local officials may reduce their control over environmental governance to meet the set economic growth targets (Yu and Pan 2019). Therefore, we believe that the impact of regional economic growth targets on environmental governance performance is negative.
Informatization level (I NT ). The popularization of internet technology has created a transparent and efficient network platform for the public, broadened the channels for the public to participate in politics and deliberations, and enabled the public to freely express their needs for environmental quality, thus having a positive impact on the formulation of central or local environmental policies (Xie et al. 2022). Therefore, we use internet penetration to measure the degree of regional informatization, and its impact on environmental governance performance is positive.

Sample selection and data sources
For the following reasons, we selected panel data from 30 provinces in China from 2001 to 2015 for empirical investigation. The first criterion for the selection of research samples was the exclusion of Tibet, Macao, Hong Kong, and Taiwan because of the unavailability of relevant data. The second criterion is the selection of the research interval. 2001 is the first year of China's policy on green finance, and the data in 2001 are also the earliest green finance data that we can obtain, so we chose 2001 as the starting period of this study. Considering that since 2016, the statistical style of the "China Environment Yearbook" has changed, resulting in the inability to obtain the required environmental governance data, 2015 is selected as the termination period of this study.
The raw data of this study comes from the China Energy Statistical Yearbook (NBSC 2016a), China Environment Yearbook (NBSC 2016b), China Industry Statistical Yearbook (NBSC 2016c), China Statistical Yearbook (NBSC 2016d), and the statistical yearbooks for related provinciallevel regions. All data at current prices are deflated to be constant at 2000 prices. Table 1 shows the descriptive statistical results for the variables. to 2015, China's cleaner production performance (CP P ), end-of-pipe treatment performance (EP P ) and system governance performance (SP ) showed an upwards trend, especially from 2001 to 2011. It is worth noting that the change direction of system governance performance is similar to that of cleaner production performance, and cleaner production performance is much lower than endof-pipe treatment performance, which indicates that the lag of cleaner production performance has become a bottleneck restricting the improvement of China's environmental governance performance.

Trends in environmental governance performance
Due to the different trends in environmental governance performance at each stage, we divided the sample period into three phases: 2001-2006, 2007-2011 and 2012-2015 based on the key nodes of green finance and environmental governance policies.
Phase I: 2001-2006. In 2001, the China Securities Regulatory Commission issued the "Guidelines for the Content and Format of Information Disclosure of Companies Publicly Offering Securities No. 9 -Application Documents for Initial Public Offering of Shares", which directly linked corporate financing to environmental responsibilities, marking the beginning of green finance in China. In 2002, the promulgation of the "Cleaner Production Promotion Law of the People's Republic of China" prompted the transformation of China's pollution control model from end-of-pipe treatment to cleaner production. Although cleaner production performance began to decline in 2005 under the impact of the oil crisis, cleaner production performance generally increased to a certain extent at this stage, while end-ofpipe treatment performance declined during this period, Risks", marking the official construction of China's green credit policy system and the transformation of green finance from the embryonic stage to the development stage. At this stage, China's systematic environmental governance performance, cleaner production performance and end-of-pipe treatment performance showed a clear upwards trend. However, the occurrence of the US subprime mortgage crisis in 2008 forced China to shift its focus to restoring its domestic economy, temporarily reducing the intensity of environmental governance and resulting in a certain degree of decline in environmental governance performance.
Phase III: 2012-2015. In 2012, the China Banking and Insurance Regulatory Commission issued the "Green Credit Guidelines", marking a period of large-scale development of green finance. At the same time, the regional environment entered a stage of comprehensive governance, and environmental regulations such as the Environmental Protection Law of the People's Republic of China were promulgated in succession to form a synergy with green finance policies. However, China's environmental governance performance had not risen and had even shown a slight downwards trend. We believe that the reason for this phenomenon is that there was a clear alternative relationship between green finance policies and environmental governance policies influencing environmental governance performance, an idea that we have confirmed in "Moderating effects test" section.

Interprovincial comparison of environmental governance performance
To gain more insight into the regional differences of environmental governance performance, Fig. 3 presents a direct visualization of the results in the form of topographic maps, where the darker red colors represent the highest SP , CP P and EP P level in 30 provinces of China. It can be seen that the average SP values range between 0.539 and 0.947, with an overall mean value of 0.792, indicating that most provinces have low industrial systematic environmental governance performance. The average CP P values range between 0.402 and 0.943, with an overall mean value of 0.696, which is a large distance from the efficiency frontier of cleaner production performance. Cleaner production performance is the highest in the eastern region, followed by the central and western regions. Furthermore, provinces located in the eastern region such as Fujian, Guangdong, and Zhejiang are the top three with greatest cleaner production performance, while those located in the western region such as Shanxi, Xinjiang, and Gansu are the bottom three with the least cleaner production performance. The average EP P values range between 0.702 and 0.992, with an overall mean value of 0.910, indicating that most provinces are close to the efficiency frontier of end-of-pipe treatment performance. Similar to cleaner production performance, end-of-pipe treatment performance is distributed in stages in the eastern, central and western regions. In addition, provinces with low end-of-pipe treatment performance are mainly concentrated in the northeast and west, such as Heilongjiang, Liaoning, Ningxia, Gansu, Sichuan, Shaanxi and Xinjiang.
In general, compared with end-of-pipe treatment performance, cleaner production performance is generally lower. Because end-of-pipe treatment has developed relatively well Fig. 2 Trends of cleaner production performance, end-of-pipe treatment performance and systematic environmental governance performance during [2001][2002][2003][2004][2005][2006][2007][2008][2009][2010][2011][2012][2013][2014][2015] in China and the concept of cleaner production was not formally proposed until 2002, its development stage is shorter than that of end-of-pipe treatment, and a systematic and complete governance system for cleaner production has not been formed. In addition, CP P and EP P show a positive correlation, indicating that most provinces that focus on industrial end governance are gradually beginning to pay attention to cleaner production.

Effect of green finance on environmental governance performance
Regression analysis of green finance on cleaner production performance and end-of-pipe treatment performance Table 2 shows the regression results of the effect of green finance on cleaner production performance and end-of-pipe treatment performance. The estimation results in Model (1) of Table 2 show that the coefficient of green finance for cleaner industrial production is 0.649 at the 1% level, revealing that the development of green finance has positively contributed to cleaner industrial production. This is because cleaner production starts from the perspective of the whole life cycle, emphasizes source reduction, realizes emission reduction through green product design, production process optimization and comprehensive utilization of byproducts, and ultimately encourages enterprises to achieve economically sustainable and environmentally friendly development. This coincides with the development concept of green finance, which mainly supports green industries, guides the flow of financial funds to green projects, and optimizes the economic structure by promoting the green transformation and upgrading of the industry. Therefore, green finance can provide sufficient financial support for industrial enterprises that focus on cleaner production, alleviate the financing pressure of such enterprises, and then improve the performance of cleaner production. The estimation results in Model (2) of Table 2 show that the coefficient of green finance for end-of-pipe treatment is −0.398 at the 1% level, indicating that the development of green finance has an inhibitory effect on end-of-pipe treatment performance. Compared with cleaner production, the traditional end-of-pipe treatment model focuses on the treatment of the pollutants that have been generated, and as such, industrial enterprises are more concerned about how to meet current pollution emission standards; that is, they tend to develop short-term environmental performance improvement plans. Moreover, we believe that the negative impact of green finance on industrial end-of-pipe treatment performance is staged. Existing research has also confirmed that there is a U-shaped relationship between financial development and environmental governance performance, and when the development of green finance crosses the threshold, it will achieve a transition from negative to positive (Khan et al. 2022).
Regarding the control variables, the coefficients of ln(P GDP ) and [ln(P GDP )] ∧ 2 in Model (1) of Table 2 are significantly positive and negative, respectively, thus verifying the existence of an EKC curve (Zhao et al. 2016). The coefficient of UR is significantly negative in both Model (1) and Model (2) of Table 2, indicating that the rapid advancement of urbanization has caused a surge in energy consumption demand, thus hindering industrial environmental governance. The coefficient of EGG is significantly positive in Model (1) of Table 2, indicating that regional economic growth targets promote cleaner industrial production performance, and the pattern of economic growth at the expense of the environment gradually disappears. The coefficient of I NT is significantly positive in both Model (1) and Model (2) of Table 2, indicating that the improvement of the informatization level has broadened the channels for the public to participate in politics and deliberations. Currently, the public can make suggestions to Table 2 Baseline regression results of the effect of green finance on cleaner production performance and end-of-pipe treatment performance

V ariable
(1)  Standard errors of coefficients are reported in parentheses; *, **, and *** indicate that the coefficients are statistically significant at the 10%, 5%, and 1% levels, respectively government departments through formal network platforms, and government departments have also given positive feedback to the public. As the effect of green finance on cleaner production performance and end-of-pipe treatment performance may vary in different regions, we further divided the overall sample into eastern, central and western regions. The estimation results of models (1), model (3) and model (5) in Table 3 reveal that green finance in the eastern and central regions is more likely to have a large positive effect on cleaner production performance, while green finance in the western region has an inhibitory effect on cleaner production performance. Possible reasons are as follows. First, there are big differences in the structure of green financial institutions in the eastern and central regions and in the western region. The eastern and central regions have a multi-level green financial system, and the securities market and fund market have effectively supplemented the green credit of commercial banks. Enterprises in the eastern and central region have diversified financing channels, and green finance can effectively provide financial support for industrial enterprises that focus on cleaner production. Compared with those in the eastern and central regions, the financial industry in the western region is underdeveloped, and the green financial institutions are mainly banks and only have a single structure, resulting in less support for industrial enterprises that focus on cleaner production. Second, the eastern and central regions and the western region are currently in different stages of economic green transformation. In the process of cleaner production, the eastern and central regions focus more on the green transformation of production technology, and tend to consider the future green transformation in terms of capital, personnel and technology, while the west region pays more attention to the rapid development of the economy and prefers to adopt applicable technologies to meet current needs. The estimation results of models (2), model (4) and model (6) in Table 3 show that green finance in the eastern and central regions exert a significant inhibitory effect on end-of-pipe treatment performance, but the effect in the western region is not significant. One possible explanation is that the western region takes economic growth as the core indicator and considers cleaner production and end-ofpipe treatment as a whole. As a result, green finance has no obvious effect on end-of-pipe treatment.

Mediation effect test
Green finance has a significant positive impact on cleaner industrial production performance and a significant negative impact on industrial end-of-pipe treatment performance. However, the mechanism of impact of green finance on cleaner industrial production and end-of-pipe treatment is still unclear. As the core tool for coordinating the economy and the environment, industrial structure optimization is directly related to how the economic system uses resources and discharges waste, so we analyze the transmission path from the perspective of industrial structure optimization.
Model (3) of Table 4 shows the effects of green finance on industrial structure optimization, indicating that green finance can effectively promote industrial structure optimization. Model (1) of Table 2 shows that industrial structure optimization has promoted cleaner production performance. Combined with the comprehensive comparison of Model (1) in Table 4, it can be found that green finance can improve cleaner industrial production by guiding industrial structure optimization, thus verifying the transmission path of "green finance-industrial structure optimizationimproved cleaner production performance". For industrial end-of-pipe treatment performance, Model (2) of Table 2  Standard errors of coefficients are reported in parentheses; *, **, and *** indicate that the coefficients are statistically significant at the 10%, 5%, and 1% levels, respectively shows that industrial structure optimization has a significant inhibitory effect on end-of-pipe treatment performance. Combined with the comprehensive comparison of Model (2) in Table 4, industrial structure optimization has a suppressive effect on industrial end-of-pipe treatment performance, even though green finance encourages optimization of the industry, thus verifying the existence of the transmission path of "green finance-industrial structure optimizationdecline in industrial terminal governance performance".

Robustness test
(1) In our first test, we attempt to alleviate the omission bias. Although we added control variables such as ln(P GDP ), [ln(P GDP )] ∧ 2, T RAD, UR, EGG, and I NT to Table 2 and controlled the fixed effects of both the year and the region, it is impossible to eliminate the influence of other factors on the benchmark results, and there may still be missing variables. Therefore, we further set government intervention (GI ), regional population density (P D), and years of schooling (EDU ) as control variables to alleviate endogenous problems caused by missing variables. As shown in Model (1) and Model (2) of Table 5, the coefficient of green finance on cleaner production performance is significantly positive, and the coefficient of influence on end-of-pipe treatment performance is significantly negative, indicating that the research results of this paper are reliable.
(2) Considering the mutual causal relationship between green finance and heterogeneous environmental governance performance, we used the current green finance development level and the environmental governance performance of the lagged period to conduct an empirical analysis. As shown in Model (3) and Model (4) of Table 5, there is a positive correlation between the green finance development level and the cleaner production performance of the lagged phase, and there is a negative correlation relationship with the end-of-pipe treatment performance of the lagged phase; both are significant at the level of at least 10%, which also shows that the implementation effect of green finance policies has a certain degree of sustainability.
(3) We then attempted to alleviate self-selection bias. Due to the advantages of the eastern region in terms of development strategy, economic development level and institutional environment, financial institutions are more inclined to carry out green financial activities in the eastern region. As of 2019, more than 40% of the branches of urban commercial banks have gathered in the eastern coastal areas, indicating that there is an obvious geographical "selfselection" problem in the expansion of financial institutions. To alleviate the distortion caused by self-selection bias, this paper refers to the research of Zhang and Chen (2021) and excludes samples from Jiangsu, Zhejiang, and Shanghai in a re-estimation of the model. As shown in Model (5) and Model (6) of Table 5, we found that even after excluding the sample data of eastern economic centers, the impact of green finance on cleaner production performance and endof-pipe treatment performance is consistent with the results of the baseline regression.

Moderating effects test
The regression results of Model (1) in Table 2 show that the influence coefficients of "market-incentive" environmental regulation and "command-and-control" environmental regulation on cleaner production are significantly positive at the 1% level, indicating that both "market-incentive" and "command-and-control" environmental regulation have a significant role in promoting cleaner production when implemented separately. According to the regression results of Model (1) and Model (2) in Table 6, it can be found that the interaction term coefficients between green finance and "market-incentive" regulation and between green finance and "command-control" regulation are significantly negative, indicating that the implementation of heterogeneous  (2) correspond to Eqs. 7 and 8; Model (3) corresponds to Eq. 6. Standard errors of coefficients are reported in parentheses; *, **, and *** indicate that the coefficients are statistically significant at the 10%, 5%, and 1% levels, respectively environmental regulation weakens the positive impact of green finance on cleaner production and that there is a clear substitution between China's green finance and heterogeneous environmental regulation in the process of influencing cleaner production. On the one hand, there is an overlapping effect between green finance and "commandand-control" environmental regulation in the establishment of environmental standards. "Command-and-control" policies regulate high-pollution industrial enterprises by setting relevant regulatory requirements, and green finance regulates high-pollution industrial enterprises by setting relevant environmental protection standards; only industrial enterprises that meet these standards can be supported by green funds. On the other hand, green finance and "marketincentive" regulations intersect in terms of economic benefits. "Market-incentive" regulations create economic incentives for industrial enterprises to reduce pollutant emissions through cleaner production technology. Industrial enterprises that meet regulatory standards can shift the amount of their sewage charges to invest in the research and development of cleaner production technologies, thereby promoting the source governance and all-round governance of indus-trial production. Green finance opens a green channel for industrial enterprises that focus on adopting clean technology, alleviates the financing pressure of such industrial enterprises by increasing loan amounts and reducing loan interest rates, and enhances industrial willingness to carry out clean technology innovation, thereby promoting the green and sustainable development. From the regression results of Model (2) in Table 2, it can be obtained that the "market-incentive" environmental regulation and the "command-and-control" environmental regulation have a significant role in promoting the end-of-pipe treatment performance of Chinese industries. According to the regression results of Model (3) and Model (4) in Table 6, the interaction term coefficients between green finance and "market-incentive" environmental regulation are significantly positive, indicating that "marketincentive" environmental regulation mitigates the negative impact of green finance on industrial end-of-pipe treatment performance. This is because financial institutions are more willing to invest green funds in industrial enterprises that tend to implement cleaner production technology, thus crowding out the funds that might go to industrial Standard errors of coefficients are reported in parentheses; *, **, and *** indicate that the coefficients are statistically significant at the 10%, 5%, and 1% levels, respectively enterprises that focus on end-of-pipe treatment. Industrial enterprises that meet environmental protection standards through end-of-pipe treatment, however, can also internalize the costs originally used to pay sewage charges so that some funds go to the improvement of end-of-pipe treatment technology. The interaction term coefficient between green finance and "command-and-control" environmental regulation is not significant, indicating that "command-andcontrol" environmental regulation has not played a good moderating effect in the impact of green finance on the performance of industrial end-of-pipe treatment.

Conclusions
The promotion and implementation of China's regional environmental governance requires the help of green finance capital; indeed, serving environmental governance through green finance has become the proper means of promoting regional green transformation and sustainable development.
Based on panel data from 30 provincial-level regions in China from 2001 to 2015, this paper constructs an empirical model to discuss the relationship and impact mechanism between green finance and environmental governance performance. The main conclusions are as follows: (1) China's cleaner industrial production performance, end-of-pipe treatment performance and systematic environmental governance performance showed an upwards trend, distributed in stages in the eastern, central and western regions from high to low. In addition, there is a big gap between cleaner production performance and its efficiency frontier. Fujian, Guangdong, and Zhejiang are the top three provinces with greatest cleaner production performance, while Shanxi, Xinjiang, and Gansu have the worst cleaner production performance. Compared with cleaner production performance, end-of-pipe treatment performance of most provinces is close to the efficiency frontier, Heilongjiang, Liaoning, Ningxia, Gansu, Sichuan, Shaanxi and Xinjiang are provinces with low end-of-pipe treatment performance.
(2) Overall, green finance promotes cleaner industrial production performance while inhibiting end-of-pipe treatment performance. Regionally, green finance in the eastern and central regions has a positive effect on cleaner industrial production performance, while green finance in the western region has an inhibitory effect on cleaner industrial production performance. In addition, green finance in the eastern and central regions inhibit end-of-pipe treatment performance, but the effect in the western region was not significant. Standard errors of coefficients are reported in parentheses; *, **, and *** indicate that the coefficients are statistically significant at the 10%, 5%, and 1% levels, respectively (3) Industrial structure optimization plays a partial intermediary role in the impact of green finance on cleaner production and end-of-pipe treatment, which verifies two transmission paths: "green finance-industrial structure optimization-improved cleaner production performance" and "green finance-industrial structure optimization-decline in industrial terminal governance performance".
(4) For cleaner industrial production, the implementation of "market-incentive" environmental regulation and "command-and-control" environmental regulation weakens the positive impact of green finance on cleaner industrial production performance. For industrial end-of-pipe treatment, "market-incentive" environmental regulation mitigates the negative impact of green finance on industrial end-of-pipe treatment, while the moderating effect of "command-and-control" environmental regulation is not significant.

Policy implications
The policy recommendations in this paper are as follows: (1) Enrich green finance products and enhance the scale of green finance. The results show that green finance mainly contributes to cleaner industrial production, crowding out funds for industrial end-of-pipe treatment, so increasing the scale of green finance will help realize the transformation from a negative impact to a positive impact. Therefore, for industrial enterprises that focus on end-of-pipe treatment, the government and financial institutions need to formulate corresponding green financial support policies. Within the scope of green financial funds, industrial enterprises should focus on end-of-pipe treatment to raise funds and encourage them to introduce new equipment and new processes so that green finance can not only help cleaner industrial production but also help industrial end-of-pipe treatment.
(2) Relying on green finance promotes industrial structure optimization. Green finance can help enterprises introduce and develop production technologies by paying the sunk costs of technological innovation, thereby promoting the optimization and upgrading of the industrial structure. In the process of industrial restructuring, enterprises adhere to the concept of synergy between development and governance, alleviate the contradiction between development and pollution by promoting technological change, and achieve the greatest economic and environmental benefit ratio. Therefore, on the premise of increasing the scale of green finance, it is necessary to achieve the standards of cleaner industrial production and end-of-pipe treatment through the path of industrial structure optimization.
(3) Existing environmental regulation policies should be aligned and the establishment of a sound green financial system should be accelerated. This study shows that it is difficult for green finance to coordinate with the current environmental regulation policies to promote industrial production and end-of-pipe treatment. To this end, each province should implement the development concept of "green and coordinated" and actively develop green finance to make it compatible with the current environmental regulation policies. On the one hand, by reducing the interest rate of green credit and increasing the scale of green investment, the profitability of green finance will be further enhanced. On the other hand, the common goal of green finance and environmental regulation should be set in synergy, and the behaviour of enterprises that do not meet the requirements of environmental regulation should be restricted and punished financially, thereby weakening the overlapping effect of environmental regulation and green finance.