Green Credit, Digital Economy and Enterprise Pollution Reduction

Abstract

Green credit is an important practical exploration to reduce the environmental pollution of enterprises through the allocation of financial resources. Based on the panel data for the Chinese enterprises, this research constructs a difference-in-differences (DID) model to explore the influence of green credit policy (GCP) on the enterprise’s pollution emission and discusses the moderation effect of digital economy. The results show that the GCP can reduce the enterprise pollution emission. Specifically, the implementation of the GCP has reduced the sulfur dioxide emission intensity of enterprises by 1.53 kg/thousand yuan. The pollution reduction effect of GCP on different enterprises shows significant asymmetry: the implementation effect is better on enterprises with stronger financing constraints and state-owned enterprises, while enterprises in capital-intensive industries inhibit the effect of the policy. Specifically, the implementation of the GCP has reduced the sulfur dioxide emission intensity of state-owned and non-state-owned enterprises by 1.72 and 0.977 kg/thousand yuan, respectively, and reduced that for enterprises in capital-intensive and non-capital-intensive industries by 0.719 and 1.437 kg/thousand yuan, respectively. The development of digital economy will promote the pollution reduction effect of GCP, and the two will work together to reduce pollution emissions. Finally, some policy suggestions are put forward to optimize the current green credit policies.

1. Introduction

China’s economic development over the past four decades has made remarkable achievements, with sustained growth in total gross domestic product (GDP), rapid reduction in poverty and accelerated urbanization. However, over the years, the extensive high-speed growth model and the traditional thinking of the GDP competition have also caused a multiplied increase in pollutants and the rapid consumption of resources and energy, inducing huge environmental and climate risks. Recent report on the ecology and environment shows that, among the 339 cities at and above prefecture-level across China in 2023, the ambient air quality of 136 cities exceeded the standard, taking up 40.1%. The above data reveal a reality that cannot be ignored: the current environmental pollution problem in China is extremely serious, and it poses a major threat to the sound operation of the domestic economy and the health of the people. Environmental pollution has become a key obstacle to China’s economic development and social stability. In this context, taking effective measures to mitigate environmental pollution is not only a guarantee for sustainable economic development but also a necessary move to ensure long-term social stability.

Finance, as a key lever of market regulation, plays a pivotal role in the country’s environmental governance system. In order to effectively improve the ecological environment, it is insufficient to rely only on the end management measures, and financial means must be supplemented to reshape the incentive mechanism of resource allocation. Therefore, financial policy is particularly crucial in achieving the goal of ecological and environmental governance. The practice of green finance needs the active participation and support of micro enterprises in order to promote the healthy development of sustainable economy. Starting from the actual situation of Chinese industrial enterprises, this paper will deeply discuss the specific impact GCP, a financial instrument, on the environmental pollution of enterprises. At the same time, this study will further analyze how the GCP affects the pollution emission of enterprises and how enterprises can adjust and transform their strategies under the guidance of this policy to achieve green development.

The implementation of China’s GCP coincides with the digital transformation of financial institutions and industrial enterprises. In the context of the digital economy (DE), the development of GCP has been greatly promoted. The application of digital technology has reduced the operating costs of financial services and improved the efficiency of services. This means that green finance can be provided to a wider range of customers at a lower cost, that is, the cooperation of the two types of policies may play a “1 + 1 > 2” synergistic emission reduction effect. However, this synergistic emission reduction effect has not been verified in the current academic studies. Will promoting the two-wheel drive of green credit and digital economy release more efficient boosting force for green development? Under the background of digital greening, how can the two types of policies produce synergistic emission reduction effect? This study will focus on these issues, which is of great practical significance for the sustainable development of China’s economy.

This study deeply discusses the effect of GCP on corporate environmental pollution control at the micro level firstly. To this end, the data of China’s industrial enterprise and the Opinions on Implementing Environmental Protection Policies and Regulations to Prevent Credit Risks (hereinafter referred to as the “Opinions”) issued in 2007 are used to build a DID model. The actual effect of GCP in promoting enterprise emission reduction is evaluated empirically. This study adjusted the research sample in the robustness test to avoid the impact of China’s Green Credit Guidelines (hereinafter referred to as the “Guidelines”) implemented in 2012. Secondly, through the implementation of multi-dimensional heterogeneity analysis, this study verified the asymmetric characteristics of the impact of GCP on various types of high-polluting enterprises and revealed the difference in the impact of policies on different types of enterprises. Thirdly, combined with the patent database, this study analyzes the mechanism of GCP and examines the specific impact of this policy on industrial enterprises in the front-end production and the end-of-pipe pollution control. Finally, this study analyzes the moderation effect of DE on the enterprise emission reduction effect of GCP. In this research, the comprehensive development level of DE is measured by the principal component analysis method, from the two aspects of internet development and digital financial inclusion. The specific calculation process is illustrated in Section 6.

The empirical analysis yields three main insights. Firstly, the GCP significantly reduced the pollution emission of enterprises. The results of heterogeneity analysis indicate that the policy has more significant environmental performance improvement effect of enterprises with strong financing constraints and state-owned enterprises. In the capital-intensive industries, there are often a large number of industrial enterprises with high pollution, high emissions and high energy consumption. Because of the intensity and scale of these enterprises’ production activities, their emissions exert significant pressure on the surrounding environment, thereby undermining to some extent the expected emission reduction effects of GCP.

Secondly, the GCP has significantly enhanced the front-end control and end-management capabilities of enterprises in the production process. Specifically, the policy incentivizes enterprises to increase investment in green technology innovation through credit mechanisms, thereby improving energy efficiency and effectively controlling pollutants that may be produced at the source of production. Furthermore, the policy can increase the investment scale of pollution treatment equipment of enterprises to enhance the pollution treatment capacity and process the pollutants at the end of production.

Finally, the development of regional DE can positively moderate the pollution reduction effect of the GCP. The DE can help the green upgrading of industries by promoting green technology innovation of enterprises.

Compared with the existing research, the main contribution of this paper are as follows:

Firstly, this paper integrates the matching samples of several large micro-databases and moves the research time point forward to Opinions issued in 2007. The DID model was constructed for empirical analysis. This design not only extends the time span of the study but also captures the early effects of GCP and their dynamic changes more accurately, providing more abundant and reliable evidence for policy evaluation. In addition, compared with the data of listed companies, the micro enterprise data used in this paper cover a wider range and can more comprehensively reflect the behavioral changes in enterprises of different sizes and industries. This data advantage makes the research conclusions of this paper more general and representative and provides a deeper insight into the overall implementation effect of GCP.

Secondly, this paper is significantly different from the macro perspective of existing literature, focusing on the influence of GCP on the environmental pollution of micro enterprises. By analyzing the influence of GCP on enterprise environmental performance, this paper clarifies its internal mechanism and provides more abundant empirical evidence for understanding the function and transmission path of GCP.

Thirdly, this paper attempts to analyze the regulatory effect of DE on the emission reduction effect of GCP and provides a theoretical basis for promoting the deep integration of green finance and DE.

The rest of this paper is structured as follows. Section 2 is the literature review. Section 3 briefly reviews the evolution of green credit policy in China. Section 4 is the theoretical analysis and puts forward the research hypothesis. Section 5 focuses on the basic empirical research on the impact of GCP on corporate pollution emissions. Section 6 analyzes how DE moderates the emission reduction effect of GCP. Section 7 concludes and puts forward some policy suggestions.

2. Literature Review

This research is related to several strands of existing work. Firstly, this research relates to an extensive literature quantifying the effects of green finance on emission reduction effectiveness and environmental investment decisions. According to the externality theory, the pollution emission behavior of enterprises has a strong negative externality. The pollution emission reduces the ambient air quality, the production and living environment of workers deteriorates, and labor productivity decreases, which ultimately lead to the loss of economic efficiency and the reduction in welfare. However, the polluting enterprises do not compensate for this [1,2,3]. This means that the environmental cost caused by enterprises’ pollution emissions will be borne by the society, the private marginal cost will be less than the social marginal cost and the resource allocation will deviate from the Pareto optimal state [4]. As for high-polluting enterprises, the existing research has reached a consensus that GCP will have a negative impact on the financing costs and term structure and will have a significant financing penalty effect [5] and influence enterprises’ investment decisions through financing channels, thus affecting their environmental performance.

In terms of environmental performance, Bartram et al. [6] reveal that financing constraints may cause restricted firms to shift their pollution-intensive production activities to other regions without substantial improvement in the firm’s overall environmental performance. Similarly, the study of Xu and Kim [4] points out that under the pressure of financing constraints, polluting enterprises tend to weigh the cost of pollution control against the environmental fines they may face, and they tend to choose to continue to emit harmful gases. Using the data of Chinese enterprises, Fan et al. [5] find that GCP significantly reduces emissions of environmental defaulting enterprises, but the specific emission reduction methods varies according to the size of enterprises. Ibrahim and Vo [7] find that a higher degree of financial development promotes a wider range of greenhouse gas emissions, arguing that financial development leads to the expansion of production, which in turn promotes pollution emissions.

In addition, some scholars hold different views on the role of financial factors in the development of ecological environment. In particular, financial development can affect economic growth, and there is a significant Kuznets curve relationship between economic growth and environment. Therefore, there may be a nonlinear relationship between finance and pollution emission [8]. Zhang et al. [9] find that China’s other GCP, green finance reform and innovation pilot zones implemented in 2017, significantly boosts green investment by companies. External financing, enterprise environmental awareness and government concern for the environment are three potential mechanisms.

Secondly, this paper is related to research about the microeconomic effects of green credit and green bond. The former involves green credit’s impact on investment and financing decision-making and clean technology innovation [10,11,12,13,14]; the latter involves green bond pricing and its impact on corporate value [15,16,17,18,19,20,21,22]. The incentive and constraint effect of green credit is mainly realized through two channels to control environmental pollution at the beginning and the whole cycle. The first channel is the redistribution of capital factors [23,24]. This financing constraint is mainly due to the cautious attitude of financial institutions and investors to the environmental risks of high-polluting enterprises, resulting in restrictions on the access of such enterprises to external funds. The second channel is the green transformation of polluting enterprises. Goetz [25] and He et al. [26] confirm that GCP significantly promote corporate green technology research and development. The empirical research based on micro-subject is more embodied in the enterprise-level research and mainly support the view of promoting innovation [27,28].

Finally, this paper is also linked with the study on how GCP influences bank credit decisions. Earlier studies have pointed out that green credit policies face many challenges in the implementation process. Due to the quasi-public property of green finance, in the absence of effective incentives, banks often lack sufficient motivation to implement GCP [29]. However, with the gradual improvement of green finance policies and supporting measures, recent studies have shown that under certain conditions, the implementation of GCP is not only in line with environmental objectives but can also bring economic benefits to banks. He et al. [26] pointed out that GCP helps to improve banks’ risk management and enhance their reputation, thus promoting the improvement of bank efficiency. According to the studies of Fan et al. [5], Liu et al. [30] and Dong et al. [31], GCP has a significant restricting effect on the credit financing of high-polluting firms, thus promoting these enterprises to reduce pollution emissions. Hu et al. [32] further pointed out that GCP will encourage enterprises to switch to more environmentally friendly production methods. The above research indicates that green credit policies are well implemented at the bank level, and banks can guide credit resources away from heavily polluting enterprises.

From the above literature review, it can be found that existing studies have at least the following problems: (1) Most of the existing studies take China’s Guidelines implemented in 2012 as the starting point of GCP and use this as a natural experiment to build regression analysis using the data of listed companies. In fact, China’s GCP started from the Opinions issued in 2007, and the research based on Guidelines implemented in 2012 may have endogenous problems caused by policy expectations, which would lead to an incorrect evaluation of the policy effect. (2) Most of the existing studies have discussed the economic effects of GCP from the macro level but have paid relatively little attention to the micro enterprise behavior, which will not reveal the micro-mechanisms of the policy. (3) The existing studies have not analyzed the regulatory effect of digital economy on the emission reduction effect of GCP.

3. The Evolution of China’s Green Credit Policy

The development of China’s GCP can be divided into three stages. The period from 1995 to 2006 was the initial budding stage, during which relevant departments began to use credit means to regulate environmental pollution. From 2007 to 2011 was the rapid development stage, when China officially regarded GCP as a financial means to regulate the development of the green economy. The last stage is the improvement and perfection stage. Since 2012, China’s GCP has made remarkable progress in continuous improvement and deepening.

Early GCP lacked information communication channels and national mandatory constraints, and financial institutions out of their own interests did not actively respond to and implement it, resulting in insignificant implementation of GCP. In 2007, the “Opinion” document jointly issued by multiple departments made the first special regulation on the credit work banks at the regulatory level, clearly holding accountable for the illegal act of providing loans to environmentally illegal projects. A series of regulatory measures introduced by the policy effectively guided the flow of credit funds to environmentally friendly projects through clear environmental standards and credit restraint mechanisms, and at the same time formed a substantial binding force on the “two high” industries in which funds are invested. This dual mechanism not only promoted the rapid development of the green industry but also inhibited the expansion of the “two high” industries. The release of the “Opinions” marked the official launch and implementation of the GCP. This policy positions green credit as a market-oriented means to implement environmental protection and guides enterprises to transform to the green and low-carbon direction.

4. Theoretical Analysis of Green Credit Policy Affecting Enterprise Pollution Emissions

4.1. The Influence of the GCP on Enterprise Pollution Emissions

GCP is an important policy tool to optimize the allocation of resources by guiding the market mechanism. Its theoretical basis mainly includes externality theory, sustainable development theory and signal transmission theory. According to these theories, green credit policies make polluting enterprises bear the social cost of their behavior by internalizing the environmental cost, thus changing the behavior choice of enterprises.

There are game relations between market players at different stages, and the diversified participation methods determine the impact of GCP on firms throughout the whole process. In the whole life cycle of firms, GCP plays a crucial role in different stages of financing, production and completion. Specifically, at the financing stage, GCP significantly reduces the financing costs of green firms by reducing their loan interest rates, thus encouraging enterprises to increase investment spending. Enterprises in the production stage can realize low carbon environmental protection through the guidance of green credit.

The financial institutions can be in accordance with the GCP to carry out post-loan management, urging enterprises to realize green transformation. At the stage of production completion, the government can also promote the development of green industry by formulating corresponding industrial policies. Because the traditional environmental regulation is likely to lead to adverse selection between the government and banks, the effect of pollution reduction is not significant. Compared with the traditional environmental regulation, the structural effect of GCP is more significant.

Green credit policies have had a profound impact on the financing environment for companies by reshaping the lending mechanisms of financial institutions. By raising the financing cost of polluting enterprises, the policy has effectively adjusted the flow of credit resources and prompted financial institutions to allocate more funds to green environmental protection. At the same time, the GCP also changes the enterprises’ awareness of environmental risks and strengthens their awareness of environmental responsibility, thus encouraging enterprises to take more active measures to reduce pollution. On the one hand, for low-carbon environmental protection enterprises, financial institutions will give lower lending rates and improve the ease of corporate finance. On the other hand, for highly polluting industries, commercial banks have significantly increased the financing costs of these enterprises by limiting the scale of loans, raising loan thresholds and loan interest rates. This financing constraint mechanism forces enterprises to actively seek green and low-carbon transformation to meet their loan needs. In this process, enterprises’ green development practices can effectively reduce the pollutants, thus significantly improving environmental pollution.

Commercial banks, based on the environmental social responsibility of enterprises comprehensively consider the development status and prospects of the industry, strictly limit the approval of loans and mostly adopt the “environmental protection one-vote veto system” to control the credit investment in “two high” industries and inhibit the credit scale of “two high” enterprises. The efficiency of resource allocation in the industry will be significantly optimized by suppressing the scale of credit financing, new investment and market share of highly polluting companies. The policy stipulates that banks can only issue loans to companies that meet the green credit approval standards and reject or strictly limit credit applications from heavily polluting enterprises, which will stimulate the enthusiasm of enterprises to enhance their investment efficiency.

Government regulatory departments play an important role in environmental regulation, through the dual mechanism of public opinion supervision and legal supervision, to conduct comprehensive supervision of enterprises’ environmental protection. Through the extensive participation of the public and the media, public opinion supervision forms social pressure and encourages enterprises to take the initiative to improve environmental behavior. Legal supervision through strict implementation of environmental regulation and the implementation of penalties for non-compliant enterprise form a strong legal constraint. This dual supervision mechanism effectively inhibits the pollutant discharge behavior of enterprises and significantly improves the environmental pollution situation in the region.

Therefore, this paper proposes research hypothesis 1: green credit policy can reduce enterprise pollution emissions.

4.2. The Mechanism of GCP Affecting the Pollution Emissions of Enterprises

With the continuous development of productive forces and the rapid increase in population, the fundamental cause of environmental problems is the uncontrolled exploitation of environmental resources by enterprises and the direct discharge of industrial “three wastes” into the environment, leading to the deterioration of the surrounding environment and harming the health of the residents, and the cost of environmental pollution does not have to be borne by the enterprises that are responsible for the discharge of pollutants. By raising the loan cost of heavily polluting enterprises, GCP can effectively realize the internalization of external costs. Specifically, this mechanism forces heavy polluting enterprises to bear the social costs caused by their environmental pollution behavior, thus incentivizing enterprises to reduce pollution emissions or adopt more environmentally friendly production methods and significantly reducing the pollution level of heavy polluting enterprises.

When the government imposes credit constraints, it will take into account economic efficiency, social efficiency, environmental efficiency, and whether it is in line with the principle of social equity. Pollution emissions from enterprises generate a more serious negative externality problem, and the government, by imposing credit constraints on polluters, reflects the loss of social welfare with the increase in production costs, thus realizing the transformation of the external pollution costs of polluting enterprises to the internal production costs. In order to reduce or transfer this part of the cost, enterprises will use green products, equipment and green technology innovation or reduce pollution emissions and other methods, and ultimately realize the purpose of reducing pollution emissions [33]. According to Porter’s hypothesis, GCP can maximize the environmental protection through government guidance and support for enterprise green technology innovation [34]. Some scholars study that GCP may promote enterprise emission reduction from two channels, namely front-end pollution management and end-of-pipe management of the production process [35]. This research mainly analyzes the front-end pollution management and end-of-pipe management mechanism. The specific mechanisms are shown below.

4.2.1. Front-End Pollution Management

Front-end management, known as cleaner production technologies or pollution prevention technologies [36], helps firms to reduce pollutant emissions at the source by improving the efficiency of energy utilization. Wang [37] defines front-end governance as the development or adoption of innovative products, equipment or technologies that are beneficial to the environment. Through its incentive effect and green innovation effect, GCP has played a significant role in promoting the front-end management level of heavy polluting enterprises.

As for the incentive effect, financial institutions can adopt innovative financial products or tilt funds to reward environmental protection enterprises. Xu et al. [38] found that commercial banks have incentivized eligible heavily polluting enterprises to choose to carry out green projects by adopting preferential policies in loan approval procedures, collaterals, capital scale, preferential interest rates and term costs. The government has carried out a lot of substantial incentives aimed at expanding the supply of green credit funds, such as launching green bonds, prioritizing the acceptance of green loans as collateral for standing lending facilities such as medium-term lending facilities (MLF) and short-term lending facilities (SLF) and taking green credit as an important reference indicator in macroprudential assessment.

Regarding the green innovation effect, Porter’s hypothesis suggests that the main way for environmental protection policies to have an impact on the economy is to promote technological innovation by enterprises, and the differentiated approach of commercial banks to environmentally friendly enterprises and highly polluting enterprises will affect the innovation of enterprises. The inflow of large amounts of capital in the environmental protection industry can not only improve the business status quo of enterprises but also increase the enthusiasm of enterprises in green technology innovation. Highly polluting enterprises facing strict government control and increased financing costs will also prompt them to emphasize technical aspects such as research and development (R&D) innovation. These will push enterprises to eliminate outdated production equipment and apply for more green patents.

Therefore, this paper proposes hypothesis 2: GCP reduces pollution emissions by promoting enterprise green technology innovation (such as the number of green patents) and energy efficiency improvement (such as energy consumption per unit of output value).

4.2.2. End-of-Pipe Pollution Management

End-of-pipe management refers to the implementation of measures to reduce the enterprise’s pollution at the end of production, mainly in terms of the enterprise’s ability to treat pollutants and technical efficiency. The goal of front-end management is to reduce the pollutants in the production process, but the pollutants produced cannot be completely eliminated, so the enterprise also needs to carry out end-of-pipe management of the pollution that has already been produced, to control the pollutants to meet the emission standards, and to reduce the harm of the pollutants to the environment.

According to Zhao et al. [39], end-of-pipe management strategy refers to a series of measures in the production process to reduce the enterprise’s pollution. These measures include, but are not limited to, reducing emissions of waste gas, wastewater and solid waste, as well as reducing greenhouse gas emissions. In the last century, China was deeply influenced by the “pollute first, treat later” model of developed western industrial countries, and the scale of resources and technology level was limited, so the environmental protection strategy at that time was based on “end-of-pipe management”, the harm of industrial waste gas, wastewater and solid waste (hereinafter referred to as the “three wastes”) were emphasized, and special sewage standards and charging systems were formulated for the “three wastes”, with the ultimate aim of reducing the amount of pollutants discharged at the end of the production.

A common method of end-of-pipe management for enterprises is investment in pollution treatment equipment, which allows enterprises to utilize waste gas, wastewater and other pollution treatment equipment to process and treat pollutants generated during the production process and is a “symptomatic” approach to pollution prevention and control. Although end-of-pipe management strategies are widely admired for their ease of imitation and operation, such measures often require significant capital investment. In this context, the implementation of GCP provides an important way for enterprises to obtain funds for green projects and helps to expand the scale of enterprises’ investment in pollution treatment equipment.

Therefore, this paper proposes hypothesis 3: GCP reduces pollution emissions by increasing investment in pollution treatment equipment (such as the number of waste gas treatment facilities) and improving treatment capacity (such as sulfur dioxide removal intensity).

As China’s environmental pollution management ideas gradually change, the center from around the end-of-pipe management gradually transferred to the front-end management; this is due to the large difference in the role of front-end management and end-of-pipe management: first, in terms of processing efficiency, enterprises taking into account the cost factors often do not accurately calculate the proportion of pollutants to the actual degradation of emissions, but usually the proportion of the composition of emissions are different, and the use of a unified end-of-pipe management will affect their treatment efficiency.

Secondly, in terms of resource utilization, front-end management can greatly improve the efficiency by controlling pollution through the production process, while end-end management only deals with the pollutants as the derivatives of production, and the process will have a negative impact on the resource utilization; then, in terms of economic benefits, both mechanisms will invest a large amount of manpower and financial resources in the early stage, but the front-end management can develop more advanced cleaner production technologies and green innovative products, which will bring sustainable economic benefits to enterprises, while end-of-pipe management can only generate losses and costs.

Finally, in terms of result control, the existing end-of-pipe management technology has limitations. There is still a certain risk to the environment in the process of treatment, while front-end management can control the pollution in the source and does not have the risk of bringing secondary pollution to the environment. Therefore, the front-end source pollution control compared to the end-of-pipe management has obvious advantages. Front-end management has gradually become the main mode of China’s environmental protection and clean policy. However, there are certain problems with the policy, such as the relevant policies and measures are not perfect enough, and enterprises do not have enough technological research and development and lack the motivation to carry out front-end management.

In addition, the GCP effectively limits the scale of commercial banks’ loans to highly polluting enterprises by raising the credit conditions and loan interest rates for enterprises in polluting industries, affecting the supply of funds to enterprises. Enterprises that carry out front-end management will bring greater cost pressure. Considering the nature of the enterprise’s profit-oriented business, some enterprises are more inclined to control the pollution emissions through the purchase of a number of relatively low-cost equipment.

In general, both mechanisms can reduce enterprise pollution emissions, front-end management to realize the pollution control of the production process and end-of-pipe management to reduce pollution in the end of the production process, becoming an effective complement to the front-end management. Therefore, for the improvement of environmental pollution, the mechanism of action of the front end is mutually complementary and mutually reinforcing.

4.3. The Moderation Effect of DE on the Emission Reduction Effect of GCP

Along with the development of DE, the development of green credit as a financial activity supporting environmental improvement and sustainable development has been greatly promoted. The application of digital technology has reduced the cost of financial services and improved the efficiency of services. This means that green financial products and services can be provided to a wider group of customers at a lower cost, further promoting the popularity and development of GCP. DE improves the speed and transparency of information flow, makes the risk assessment of green projects more accurate and helps investors identify and invest in those projects with real green benefits, thus promoting the development and upgrading of green industries. Digital technology provides more advanced risk management tools and methods. Financial institutions can use big data analysis to predict and assess environmental risks, develop more accurate risk control strategies and reduce the potential risks of green projects. The development of the DE has also promoted the government’s policy support and supervision of green finance, and the government can use digital means to strengthen the supervision of the financial market to ensure that funds flow to projects that meet environmental standards.

In short, the DE effectively promotes the development of GCP by improving information transparency, reducing costs, strengthening risk management and supporting policy supervision and, thus, has a positive regulatory effect on enterprises’ pollution reduction.

Therefore, this paper puts forward research hypothesis 4: digital economy can promote the emission reduction effect of GCP on environmental pollution.

It should be noted that the four hypotheses presented above are similar or not similar to results derived from subjective experience to some extent. However, conclusions based on subjective experience are not necessarily correct. In order to obtain a scientific and credible conclusion, this study will test the above hypotheses one by one based on econometric methods.

5. Empirical Study on the Impact of Green Credit Policy on Enterprise Pollution Emissions

5.1. Benchmark Regression Design

5.1.1. Data Sources and Processing

The dataset of this study covers three large Chinese micro enterprise databases: the China Industrial Enterprise Database, the China Enterprise Pollution Emission Database and the Patent Database for Innovation of Chinese Enterprises, which are sourced from the National Bureau of Statistics of the People’s Republic of China (NBSP), the Ministry of Ecology and the Environment (former Ministry of Environmental Protection) and the State Intellectual Property Office (SIPO), respectively. The sample period is 1998–2014.

Among them, the China Industrial Enterprises Database records the basic information of industrial state-owned enterprises (SOEs) and non-state-owned enterprises, such as enterprise code, name and address; financial information, such as financial expenses and profits; and the production and sales information, such as total output value, sales volume and fixed assets. The China Enterprise Pollution Emission Database reports basic information, such as enterprise name and address; emission and governance data, such as pollutant generation and emission of sulfur dioxide, chemical oxygen demand, waste gas, wastewater, etc.; and energy use data such as energy consumption. The Patent Database for Innovation of Chinese Enterprises records the patent applications and authorizations of enterprises and distinguishes three categories of patents: invention, utility model and design, and this study matches them with the Green List of International Patent Classification.

The study refers to the processing method in Fan et al. [5] to match the database of industrial enterprises, using the order of “organization code” and “enterprise name” to match and merge the samples in the database across years. The final panel dataset contains information on basic characteristics, pollution emission and green patents of enterprises, with the time dimension of 1998–2014. On this basis, data cleaning is carried out on the combined database, and the specific steps include the following: (1) in view of the many changes in the national economic industry classification standards during the sample period, this paper unifies the four-digit industry classification codes to the National Economic Industry Classification; (2) strict criteria were adopted in the sample screening process to make sure the accuracy of data. This paper excludes the 2010 data sample and excludes the following logical discrepancies: the sum of current assets and net fixed assets exceeds total assets, the depreciation of the current year exceeds the accumulated depreciation, the asset–liability ratio is less than 0, the year of operation is later than the statistical year, and the emission of pollutants is less than 0. In addition, this study also eliminated samples with missing key indicator data.

5.1.2. Econometric Model

Referring to Fan et al. [5] and Zhang et al. [9], this study regards the “Opinion” issued in 2007 as an exogenous shock and sets up a benchmark model based on DID estimation framework as shown in Equation (1):

$${Polind}_{ftci}={\partial }_0+{\partial }_1{Treat}_i\times {Post}_t+\gamma X_{ft}+{\lambda }_{ftci}+{\epsilon }_{ftci}$$

(1)

The subscripts f, t, c and i represent firm, year, city and industry, respectively. The explanatory variable ${Polind}_{ftci}$ represents the firm pollution emission intensity indicator. $X_{ft}$ represents a set of firm-level control variables. ${\lambda }_{ftci}$ represents firm, year, city and industry fixed effects (FE) that are used to absorb homogeneous shocks to firms in the same group range from unobservable typical characteristics at the individual, time, region and industry levels, respectively. ${\epsilon }_{ftci}$ denotes the randomized disturbance term.

The core explanatory variable in this paper, ${Treat}_i\times {Post}_t$, where ${Treat}_i$ is a treatment group dummy variable, taking 1 for high polluting firms and 0 for non-high polluting firms; ${Post}_t$ is a dummy variable, taking 0 for the years prior to 2007, and 1 for 2007 and beyond. ${\partial }_1$ is the estimated coefficient of the key explanatory variable to the study, which quantifies the specific impact of GCP on corporate environmental performance. When constructing the difference-in-differences term, the key is to accurately define the treatment and the control group. In this paper, they are determined according to the six “two high” industries in the “Notice of The State Council on the Issuance of the Comprehensive Work Plan for Energy Conservation and Emission Reduction” (hereinafter referred to as the “Notice”) issued in May 2007. The publication of the Notice provides a clear reference point for this paper to capture the net effect of the policy by using the DID method. Based on the definition of the Notice, the sample enterprises are grouped as follows: if enterprise belongs to the six “two-high” industries, it will be identified as a high-polluting enterprise and regarded as a treatment group; if the enterprise belongs to an industry whose two-digit code does not belong to these industries, the enterprise will be identified as a non-high-polluting enterprise and acts as a control group.

5.1.3. Variable Definition and Measurement

Considering that ${SO}_2$ is the main pollutant emission of industrial enterprises, this paper chooses the proportion of ${SO}_2$ emission to the total industrial output value (Polind) as a proxy variable for the pollutant emission intensity of enterprises.

For control variables, referring to Fan et al. [5] and Sun et al. [40], this study chooses firm age (Age), firm size-asset (Asset), firm size-staff (Staff), capital structure (Capstr), financial leverage (Finlev), return on assets (Roa) and net profit margin on sales (Growth). The firm age is defined as the difference between the year of establishment of the firm and the year of statistics plus 1. The firm size is expressed in two ways: one is the total assets of the firm, and the other is the total employees of the firm. The capital structure reflects the ratio of debt and equity in the assets of an enterprise, which is expressed by the ratio of fixed assets to total assets. Financial leverage is an indicator to measure the level of corporate debt, which is expressed by the asset–liability ratio. The ratio reflects the extent to which companies rely on debt financing. Return on assets is expressed as the ratio of net profit to total assets. This ratio can reflect the profit efficiency of an enterprise’s assets. The net profit margin on sales is expressed as the ratio of net profit to total sales. Because of the lack of net profit data in the database of industrial enterprises, the net profit in this paper is indirectly measured by the difference between total profit and income tax payable.

5.1.4. Descriptive Statistics

To mitigate the absolute difference in different variables in order of magnitude and to solve the potential heteroscedasticity problem, the natural logarithm transformation is performed for some non-ratio indicators in this study.

Table 1. Descriptive statistics for key variables.

Variable	Variable Declaration	Unit	Sample Size	Mean	Standard Deviation	Minimum	Maximum
Polind	$\mathrm{proportion}\mathrm{of}{\mathrm{S}\mathrm{O}}_2$ emission to the total industrial output value	kg/thousand yuan	325,900	1.669	5.01	0	51.21
Age	difference between the year of establishment of the firm and the year of statistics plus 1	-	325,900	2.988	0.6031	1.099	7.562
Asset	natural logarithm of the total assets of the firm	thousand yuan	325,900	11.08	1.603	7.783	16.34
Staff	natural logarithm of the total employees of the firm	person	325,900	5.701	1.124	2.996	9
Capstr	ratio of fixed assets to total assets	-	325,900	0.3981	0.2117	0	0.9674
Finlev	asset–liability ratio	-	325,900	0.6119	0.2834	0.02299	1.677
Roa	ratio of net profit to total assets	-	325,900	0.05692	0.1408	−0.2707	0.9735
Growth	ratio of net profit to total sales	-	325,900	0.0147	0.1121	−0.9998	0.3602

provides descriptive statistical results of the variables in the empirical analysis.

5.2. Empirical Results and Analysis

5.2.1. Baseline Regression Results

Table 2. Benchmark regression results.

Variable	(1)	(2)	(3)	(4)	(5)
	Polind	Polind	Polind	Polind	Polind
Treat*Post	−0.171 ***	−1.610 ***	−1.423 ***	−1.575 ***	−1.530 ***
	(0.018)	(0.025)	(0.031)	(0.032)	(0.034)
Age		0.000	−0.000	−0.000	−0.000 *
		(0.000)	(0.000)	(0.000)	(0.000)
Asset		−0.476 ***	−0.212 ***	−0.211 ***	−0.212 ***
		(0.013)	(0.015)	(0.015)	(0.016)
Staff		0.017	−0.067 ***	−0.070 ***	−0.069 ***
		(0.015)	(0.015)	(0.015)	(0.016)
Capstr		0.308 ***	0.023	0.019	0.015
		(0.048)	(0.048)	(0.048)	(0.050)
Finlev		0.029	−0.001	−0.003	0.001
		(0.039)	(0.039)	(0.039)	(0.040)
Roa		−0.690 ***	−0.355 ***	−0.330 ***	−0.339 ***
		(0.073)	(0.074)	(0.074)	(0.078)
Growth		−1.301 ***	−1.504 ***	−1.517 ***	−1.512 ***
		(0.095)	(0.095)	(0.095)	(0.099)
Constant	1.728 ***	7.136 ***	4.779 ***	4.816 ***	4.847 ***
	(0.008)	(0.147)	(0.170)	(0.170)	(0.178)
N	325,900	325,900	325,900	325,900	325,900
r2	0.000	0.712	0.714	0.714	0.717
r2_a	0.000	0.619	0.621	0.622	0.621
Enterprise FE	NO	Yes	Yes	Yes	Yes
Year FE	NO	NO	Yes	Yes	Yes
Industry FE	NO	NO	NO	Yes	Yes
City FE	NO	NO	NO	NO	Yes

Note: Values in parentheses are standard errors. *** and * mean significance at 1% and 10% levels in this table.

shows the baseline regression results of the impact of GCP on enterprise pollution emission intensity. Column (1) is the result without including control variables, where the coefficient of DID variable is negative at 1% level, indicating that the GCP has significantly reduced the emission intensity of firms.

The regression results in column (5) of Table 2 are used as the main analysis basis in this paper. The results show that the coefficient of DID variable is −1.530 and significant at 1% level, indicating that the implementation of the GCP has reduced the sulfur dioxide emission intensity of enterprises by 1.53 kg/thousand yuan. This means that the emissions of the “two high” industries are significantly reduced, and green credit imposes substantive constraints on the credit channels of high-pollution enterprises through financial institutions to raise the credit threshold of enterprises and urges polluting enterprises to, on the basis of re-examining their own environmental and social responsibilities, actively take emission reduction measures to improve environmental pollution. Research hypothesis 1 proposed in the previous section is verified.

5.2.2. Parallel Trend Test

For this study, the hypothesis requires that the difference in environmental pollution intensity between enterprises in the “two high” industries and non-high polluting firms does not change systematically over time before GCP. Before the policy intervention, the pollution intensity of the two types of enterprises should maintain a relatively stable differential trend to ensure that the effect assessment after the implementation of GCP can accurately reflect the actual impact of the GCP. Referring to the method of Jacobson et al. [41], the following test model is built:

$${Polind}_{ftci}={\partial }_0+{\sum }_{t=1998,t\ne \mathrm{2006,2010}}^{2014}{\partial }_t{D}_t\times {Treat}_i+\gamma X_{ft}+{\lambda }_{ftci}+{\epsilon }_{ftci}$$

(2)

where $D_t$ is a year dummy variable, taking 1 when the sample comes from year t, and 0 otherwise. ${\partial }_t$ denotes the estimation value from 1998 to 2014. Figure 1 shows the results and supports the hypothesis of parallel trends. Therefore, the specification of model is correct.

5.2.3. Robustness Tests

(1) Excluding other policy effects

During the sample period of this study, there are several environmental policies that may have affected the accuracy of the estimated results, the most important of which are the SO₂ emission trading pilot launched in 2002 and the cleaner production standard implemented in 2003. To eliminate potential interference from these policies with the results estimated in this study, this research introduces other interaction terms to control the policy shocks in Equation (1) as follows: the interaction terms of whether the province is a pilot province for ${SO}_2$ emissions trading and whether it is post 2002 (${SO}_2$_trade*post_2002), and whether it is a cleaner production regulation industry and whether it is post 2003 (clean_standard*post_2003) are added to the benchmark regression equations, respectively, to control for the policy shocks. The results are in columns (1) and (2) of

Table 3. Robustness test—excluding other policy effects and adjusting the policy implementation window.

Variable	Excluding Other Policy Effects		Adjusting the Timing of the Policy Implementation Window
	(1)	(2)	(3)
	Polind	Polind	Polind
Treat*Post	−1.530 ***	−1.528 ***	−0.892 ***
	(0.034)	(0.034)	(0.031)
${SO}_2$_trade*post_2002	0.100 **
	(0.039)
clean_standard*post_2003		−0.652 ***
		(0.126)
Age	−0.000 *	−0.000 *	−0.003
	(0.000)	(0.000)	(0.002)
Asset	−0.213 ***	−0.210 ***	−0.221 ***
	(0.016)	(0.016)	(0.024)
Staff	−0.068 ***	−0.067 ***	−0.279 ***
	(0.016)	(0.016)	(0.028)
Capstr	0.014	0.016	−0.033
	(0.050)	(0.050)	(0.064)
Finlev	0.001	0.001	0.053
	(0.040)	(0.040)	(0.053)
Roa	−0.345 ***	−0.339 ***	−0.287 ***
	(0.078)	(0.078)	(0.102)
Growth	−1.506 ***	−1.514 ***	−1.805 ***
	(0.099)	(0.099)	(0.164)
Constant	4.835 ***	4.828 ***	5.778 ***
	(0.178)	(0.178)	(0.272)
N	325,900	325,900	132,677
r2	0.717	0.717	0.766
r2_a	0.621	0.621	0.657
Enterprise, Year, Industry and City FE	Yes	Yes	Yes

Note: ***, ** and * mean significance at 1%, 5% and 10% levels in this table.

. The coefficients for Treat*Post are roughly the same as in the benchmark regression in Table 2, implying that the formulation and implementation of other important policies during the same period did not significantly bias the estimates.

(2) Adjusting the timing of the policy implementation window

Since policy effects change over time and lags occur during implementation, the empirical regression results of the DID model may change depending on the set interval of the window time before and after the policy change. In order to better assess the policy effects and avoid the confused impact of China’s Green Credit Guidelines implemented in 2012, the timing of the policy implementation window is adjusted to further conduct the robustness test. Specifically, this study only retains the data of each of the two years before and after the introduction of the “Opinions”, i.e., the data of the period of 2005–2009 to re-estimate the empirical model. Regression results are listed in column (3) in Table 3. The coefficient of Treat*Post continues to show significant negative value. This statistical result further confirms the robustness of the aforementioned empirical analysis.

5.2.4. Heterogeneity Analysis

(1) Intensity of financing constraints

Under the GCP, financial institutions limited the scale of credit granted to high-polluting enterprises and increased the price of credit granted, which led to the obstruction of the credit channels of enterprises in the “two high” industries, further aggravating their financing constraints. Due to the differences in the intensity of financing constraints of enterprises, the emission reduction effects of GCP show heterogeneity. To explore this heterogeneity, this paper introduces the financing constraint index (SA), which is calculated based on the total asset and age of the firm. To avoid multicollinearity problems, Age and Asset are excluded from the control variables in column (1) of

Table 4. Heterogeneity test.

Variable	Level of Financing Constraints	Ownership of the Enterprise	Industry Factor Intensity
	(1)	(2)	(3)
	Polind	Polind	Polind
Treat*Post	−1.514 ***	−0.977 ***	−1.437 ***
	(0.034)	(0.017)	(0.022)
Treat*Post*SA	−0.144 ***
	(0.039)
Treat*Post*State		−0.743 ***
		(0.052)
Treat*Post*Capital			0.718 ***
			(0.026)
Age		0.001 *	0.001 **
		(0.001)	(0.001)
Asset		−0.146 ***	−0.154 ***
		(0.008)	(0.008)
Staff	−0.133 ***	−0.029 ***	−0.030 ***
	(0.015)	(0.009)	(0.009)
Capstr	0.037	0.035	0.028
	(0.050)	(0.026)	(0.026)
Finlev	0.033	0.006	−0.011
	(0.040)	(0.022)	(0.022)
Roa	−0.242 ***	−0.393 ***	−0.372 ***
	(0.077)	(0.053)	(0.052)
Growth	−1.626 ***	−1.022 ***	−1.006 ***
	(0.098)	(0.075)	(0.075)
Constant	2.840 ***	3.272 ***	3.369 ***
	(0.094)	(0.092)	(0.092)
N	325,900	325,900	325,900
r2	0.716	0.757	0.758
r2_a	0.621	0.675	0.676
Enterprise, Year, Industry and City FE	Yes	Yes	Yes

Note: ***, ** and * mean significance at 1%, 5% and 10% levels in this table.

. The coefficient of the triple interaction term is significantly negative, indicating that the greater the financing constraint pressure faced by high-polluting enterprises, the more significant the inhibition effect of GCP on their pollution emission. Credit channel is the main source of external financing of enterprises. The most direct “punishment” effect of GCP on high-polluting enterprises is to restrict their access to credit. If enterprises are already facing high financing constraint pressure before, the policy will further encourage them to alleviate the financing pressure caused by the reduction in the supply of credit funds through pollution reduction.

(2) Enterprise ownership

The result in column (2) of Table 4 means that the GCP has a more significant promoting effect on the pollution reduction in SOEs. Specifically, the implementation of the GCP has reduced the sulfur dioxide emission intensity of state-owned and non-state-owned enterprises by 1.72 (0.977 + 0.743) and 0.977 kg/thousand yuan, respectively. This result is mainly attributed to the following reasons: first, the quality of environmental information disclosure of SOEs is high, and the environmental protection investment and environmental protection ability are strong; second, SOEs’ decisions and behaviors are more directly affected by national policies, so they are more rapid and thorough in implementing emission reduction measures. Finally, in addition to pursuing economic benefits, SOEs also need to shoulder social responsibilities, actively respond to national environmental protection policies and take proactive emission reduction measures to achieve sustainable development goals.

(3) Industry factor intensity

In this paper, according to the two-digit code of national economic industries, the industries in which enterprises are located are classified into capital-intensive and non-capital-intensive industries, and the dummy variable (Capital) is introduced, with capital-intensive industries taking the value of 1, and otherwise taking the value of 0. Column (3) of Table 4 shows the abatement results between enterprises with different factor intensities after the implementation of GCP, and the coefficient of triple interaction term is significantly positive, and it can be found that the abatement effect of capital-intensive industries is worse when comparing labor-intensive industries. Specifically, the implementation of the GCP has reduced the sulfur dioxide emission intensity of enterprises in capital-intensive and non-capital-intensive industries by 0.719 (1.437–0.718) and 1.437 kg/thousand yuan, respectively. Since capital-intensive industries usually have more pollutant emissions, which imply that they have higher initial emission levels, it may be difficult to see significant abatement effects in the short term even if abatement measures are taken. In addition, capital-intensive industries usually need to invest a lot and have a slow cycle for equipment upgrading and renovation and upgrading of green innovations, so the emission reduction effect of their policies is weaker.

5.3. Mechanism Analysis

Through the previous theoretical analysis and research hypotheses, the pollution abatement effect of GCP is mainly realized through the front-end management and end-of-pipe management pathways of the enterprise production process. Referring to Zhao et al. [39] and Cao and Gao [42], the following mechanism test model is constructed:

$$M_{ft}={\beta }_0+{\beta }_1{Treat}_i\times {Post}_t+\gamma X_{ft}+{\lambda }_{ftci}+{\epsilon }_{ftci}$$

(3)

In the above model, $M_{ft}$ is the mediating variable, representing front-end and end-of-pipe management. ${Treat}_i\times {Post}_t$ is the core explanatory variable, and the coefficient ${\beta }_1$ is the core indicator of mechanism test. If the coefficient is significant, there is a correlation between the GCP and the front-end pollution management, and only then can the mechanism be further explained. This paper adopts the direct regression of mechanism variable to verify the research hypotheses 2 and 3.

In terms of front-end pollution control, this research focuses on the influence of GCP on enterprises’ green technology innovation and energy consumption structure and efficiency. The former one uses the total number of green innovations (LnTotal), the number of green invention patent applications (LnInva) and the number of green utility model patent applications (LnUma). For energy consumption structure and efficiency, this paper selects three variables: total energy consumption (Energyamo), energy consumption structure (Energystr) and energy utilization efficiency (Energyeff), and takes coal consumption, the ratio of gas consumption to coal consumption and the ratio of total output value to coal consumption as measurement indicators, respectively.

In terms of end-of-pipe management, the mechanism variables include sulfur dioxide removal intensity (Sudremoval), number of exhaust gas treatment facilities (Facility) and exhaust gas treatment facility treatment capacity (Capacity). Sudremoval is expressed as the ratio of the sulfur dioxide removal amount to the total output value. Facility directly uses the number of enterprise waste gas treatment facilities. Capacity is measured by the actual treatment capacity of waste gas treatment facilities. These variables jointly reflect the investment and effect of enterprises in the end pollution control and provide a multi-dimensional empirical basis for analyzing the environmental governance mechanism of GCP.

To circumvent the problem of heteroskedasticity, this paper takes the natural logarithm of all mechanism variables.

Table 5. Front-end pollution management mechanism testing: green technology innovation, energy consumption structure and efficiency.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	LnTotal	LnInva	LnUma	Energyamo	Energystr	Energyeff
Treat*Post	0.003 **	0.003 ***	−0.000	0.006	0.171	0.066 ***
	(0.001)	(0.001)	(0.001)	(0.008)	(0.161)	(0.009)
Age	−0.000 ***	−0.000 ***	−0.000 ***	0.002 ***	−0.006	−0.003 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.006)	(0.000)
Asset	0.008 ***	0.004 ***	0.005 ***	0.142 ***	0.045	0.276 ***
	(0.001)	(0.000)	(0.000)	(0.005)	(0.157)	(0.006)
Staff	−0.006 ***	−0.003 ***	−0.005 ***	0.240 ***	0.209	0.139 ***
	(0.001)	(0.000)	(0.000)	(0.006)	(0.175)	(0.008)
Capstr	−0.003 *	−0.001	−0.003 **	0.042 ***	0.047	−0.110 ***
	(0.002)	(0.001)	(0.001)	(0.015)	(0.367)	(0.018)
Finlev	0.003 *	0.001	0.003 ***	0.017	0.003	−0.048 ***
	(0.001)	(0.001)	(0.001)	(0.012)	(0.307)	(0.015)
Roa	−0.011 ***	−0.006 ***	−0.007 **	0.204 ***	−0.636	1.984 ***
	(0.003)	(0.002)	(0.003)	(0.037)	(1.019)	(0.044)
Growth	0.014 ***	0.006	0.008 **	0.168 ***	1.308	−0.437 ***
	(0.005)	(0.004)	(0.004)	(0.053)	(1.184)	(0.066)
Constant	−0.025 ***	−0.012 ***	−0.012 **	4.956 ***	−4.954 ***	−0.561 ***
	(0.006)	(0.004)	(0.005)	(0.055)	(1.793)	(0.067)
N	325,900	325,900	325,900	171,761	2487	147,852
r2	0.556	0.552	0.525	0.934	0.814	0.916
r2_a	0.409	0.405	0.369	0.910	0.693	0.881
Enterprise, Year, Industry and City FE	Yes	Yes	Yes	Yes	Yes	Yes

Note: ***, ** and * mean significance at 1%, 5% and 10% levels in this table.

demonstrates the results of the front-end pollution management mechanism test for the GCP. The results in column (1) show that the coefficient of Treat*Post is significantly positive at the 5 percent level, indicating that the “Opinion” policy significantly enhances firms’ green innovation output. A comparison of the coefficients in columns (2) and (3) reveals that the number of green invention patents of enterprises increases significantly compared to green patents of utility models, i.e., the policy’s effect on the enhancement of the quality of enterprises’ green innovations is more obvious. Regression results in columns (4), (5) and (6) reveal a key finding: the GCP only significantly improves the energy efficiency of high-polluting enterprises. The apparent increase in energy use efficiency may be due to the fact that after the “Opinion” policy, enterprises carrying out technological innovation and equipment upgrading will use more efficient production lines, which will significantly increase the actual output per unit of pollutant. In addition, the use of clean energy such as natural gas is an effective way to reduce emissions compared to fossil energy such as coal, but the energy structure of enterprises has not improved significantly, probably because enterprises have not invested more in the introduction of clean energy due to the policy constraints on financing and have insufficient incentive to optimize and upgrade their own energy consumption structure.

The results in

Table 6. Testing of end-of-pipe pollution management mechanisms: pollution control equipment and treatment capacity.

Variable	(1)	(2)	(3)
	Sudremoval	Facility	Capacity
Treat*Post	0.021	0.034 ***	0.059 ***
	(0.027)	(0.005)	(0.018)
Age	0.004 ***	0.002 ***	0.003 ***
	(0.001)	(0.000)	(0.001)
Asset	−0.286 ***	0.066 ***	0.070 ***
	(0.018)	(0.003)	(0.012)
Staff	−0.169 ***	0.102 ***	0.157 ***
	(0.021)	(0.004)	(0.015)
Capstr	0.175 ***	0.024 **	0.076 **
	(0.049)	(0.010)	(0.034)
Finlev	−0.001	−0.022 ***	−0.035
	(0.042)	(0.008)	(0.029)
Roa	−2.044 ***	0.057 **	0.073
	(0.132)	(0.025)	(0.089)
Growth	0.032	−0.091 ***	−0.012
	(0.139)	(0.027)	(0.122)
Constant	2.857 ***	−0.561 ***	7.421 ***
	(0.194)	(0.035)	(0.130)
N	325,900	187,625	142,488
r2	0.845	0.854	0.760
r2_a	0.774	0.803	0.668
Enterprise, Year, Industry and City FE	Yes	Yes	Yes

Note: *** and ** mean significance at 1% and 5% levels in this table.

show that GCP positively affects Facility and Capacity in high-polluting enterprises, and both are significant at the 1% level. However, the effect of GCP on sulfur dioxide removal intensity is not significant. That is, the implementation of GCP increases the possibility of enterprises to use pollution treatment equipment, and at the same time improves the pollution treatment capacity of the equipment through technological upgrading, which proves the effect of the “Opinion” policy on the treatment of the end of the production of enterprises.

6. The Moderation Effect of Digital Economy on the Emission Reduction Effect of Green Credit Policy

6.1. Impact of DE on Corporate Pollution Reduction

This section first analyzes the pollution reduction effect of DE. DE is measured based on internet development and digital financial inclusion. For the measurement of internet development at the city level, four indicators of Internet penetration rate, relevant employees, relevant output and mobile phone penetration rate are used, and the data are from China City Statistical Yearbook. For the development of digital financial inclusion, the China Digital Financial Inclusion Index is adopted, which is derived from the Digital Finance Research Center of Peking University. Through the method of principal component analysis, the data of the above 5 indicators are standardized and processed with dimensionality reduction, and the comprehensive development index of digital economy is obtained, which is recorded as the development index of digital economy (Digeco). The baseline regression results are shown in column (1) of

Table 7. The moderation effect of DE on the emission reduction effect of GCP.

Variable	(1)	(2)	(3)	(4)
	Polind	Polind	Polind	Polind
Digeco	0.058 ***	0.057 ***
	(0.015)	(0.016)
PrDige			32.767 ***	28.412 ***
			(3.439)	(2.498)
Treat*Post		−0.607 ***		0.131 ***
		(0.009)		(0.019)
Treat*Post *Digeco		−0.034 **
		(0.015)
Treat*Post *PrDige				−14.305 ***
				(2.665)
Age	0.002 ***	0.002 ***	0.003 ***	0.002 ***
	(0.000)	(0.000)	(0.001)	(0.001)
Asset	−0.089 ***	−0.089 ***	−0.106 ***	−0.054 ***
	(0.005)	(0.005)	(0.004)	(0.003)
Staff	−0.018 ***	−0.024 ***	−0.055 ***	−0.041 ***
	(0.005)	(0.005)	(0.004)	(0.003)
Capstr	0.022	0.023	0.027 **	0.015 **
	(0.016)	(0.016)	(0.012)	(0.007)
Finlev	0.000	0.002	−0.011	−0.017 **
	(0.014)	(0.014)	(0.011)	(0.007)
Roa	−0.647 ***	−0.604 ***	−0.764 ***	−0.758 ***
	(0.043)	(0.043)	(0.029)	(0.027)
Growth	−0.320 ***	−0.312 ***	0.684 ***	0.810 ***
	(0.075)	(0.074)	(0.064)	(0.051)
Constant	1.966 ***	2.119 ***	2.013 ***	1.174 ***
	(0.058)	(0.057)	(0.050)	(0.031)
N	273,221	273,221	132,679	132,679
r2	0.778	0.783	0.823	0.829
r2_a	0.698	0.705	0.741	0.749
Enterprise, Year, Industry and City FE	Yes	Yes	Yes	Yes

Note: *** and ** mean significance at 1% and 5% levels in this table.

. The regression coefficient is significantly positive at the 1% level, indicating that the development of DE has significantly increased the pollution emissions of industrial enterprises.

This paper further selects the instrumental variable of DE, namely the digital word frequency ratio (PrDige), to further verify the robustness of the regression results. By sorting out the report texts of provinces and municipalities across the country, this study sorts out relevant digital texts from aspects of digital technology and digital application and calculates the proportion of digital words as the instrumental variable of urban DE development for regression estimation. The results are shown in column (3) of Table 7, and the regression results are still significantly positive. The development of the DE has had a negative impact on enterprises’ pollution reduction. The possible reason is that the use of a large number of digital technology equipment in enterprises leads to a substantial increase in power consumption. At present, China has a high proportion of coal electricity, increasing power consumption, rising coal consumption and increasing pollution emissions of enterprises. Among them, the application of digital technology in mining industry has increased the mining scale of mineral resources and caused excessive consumption of resources, which has a negative impact on environmental protection.

6.2. The Impact of DE on the Emission Reduction Effect of GCP

In order to verify the research hypothesis 4 proposed in the theoretical analysis above, that is, DE can moderate the impact of GCP on environmental pollution, this study adds the interaction terms between GCP and Digeco, and GCP and PrDige. The regression results are shown in columns (2) and (4) of Table 7. The coefficients of the interaction terms are both significantly negative, indicating that the synergistic emission reduction effect of DE and GCP is obvious. In regions with higher development level of DE, the effect of GCP on promoting enterprise pollution emission reduction is more significant.

In the process of digital transformation, enterprises need to reset production equipment and increase production through increasing resource input and energy consumption, which will increase pollution emissions of enterprises. The credit rationing channel of GCP can make up for the deficiencies of micro enterprises such as lack of behavioral incentives and unfavorable policy environment and realize efficient allocation of capital in green economic development. Although the development of DE has a negative impact on environmental protection, the capital allocation function of GCP makes the interests of various investment and financing entities linked to environmental protection. By encouraging the innovation of environmental protection technology of industrial enterprises, the efficiency of energy use is improved, and the pollution emission of enterprises is reduced. Moreover, the GCP restricts the use of funds, which will accelerate the digital industry to support the development of green industries through digital intelligent technology, help industrial upgrading, and achieve pollution reduction. In short, the comprehensive effect of the collaboration between DE and GCP is conducive to the ecological environment, which verifies the research hypothesis 4.

7. Conclusions and Policy Suggestions

7.1. Conclusions

With the gradual development of China’s green economic system, the importance of the ecological environment for economic development has begun to take root in people’s hearts. GCP combines financial services and environmental protection and shoulders the important mission of optimizing credit rationing structure and improving enterprises’ pollution control ability. With the matching samples of China’s industrial enterprise and the “Opinions” issued in 2007, the pollution abatement effect and influence mechanism of GCP are studied. On the basis of literature combing and status quo analysis, the impact of GCP on pollutant emissions is theoretically investigated, and then a DID model is built to test the GCP’s inhibitory effect on environmental pollution. The mechanism test model is applied to reveal the mechanism of GCP on pollution reduction. The synergistic emission reduction effect of the DE is analyzed as well. Based on the above-mentioned research, the following conclusions are drawn:

First, by constructing a DID model, this study found that “Opinions” significantly reduced the pollution emission level of high-polluting firms. This conclusion is supported by a series of robustness tests, such as excluding other policy effects and adjusting the policy implementation window. The results of heterogeneity analysis further reveal that GCP has a particularly significant effect on the environmental performance of highly polluting firms with strong financing constraints and state-owned firms, and that for capital-intensive industries, more highly polluting, high-emission and high-energy-consuming industrial enterprises may be clustered, resulting in a large concentration of pollutants, which hampers the policy’s abatement effect.

Second, using the mechanism test model, it is proved that GCP can improve the ability of enterprises to control the front-end management and end-of-pipe management, realizing the pollution reduction in enterprises in terms of both “treating the symptoms” and “treating the root causes”. Specifically, incentivizing enterprises to increase green innovation output through credit channels, helping enterprises to improve energy use efficiency and controlling the formation of potential pollutants at the source of production are the “root causes” of pollution prevention and control. GCP can increase the scale of investment in pollution treatment equipment and, thus, enhance the pollution treatment capacity of the equipment, so as to process and treat pollutants at the end of production, which is the “symptomatic” way of pollution prevention and control.

Thirdly, the analysis of the moderation effect model shows that the development of regional DE can positively moderate the pollution reduction effect of GCP. The DE can reduce pollution by promoting green technology innovation in enterprises.

7.2. Policy Suggestions

With the conclusions of this study, we can deeply understand the impact of GCP on enterprises’ pollution, the mechanism of action and the heterogeneity of its pollution reduction effect. Finally, the following recommendations are put forward.

Firstly, differentiated incentives should be implemented. In view of the significant differences in the scale of green credit in China between different regions and commercial banks, as well as the regional heterogeneity of policy effects, it is recommended to gradually improve environmental governance policies and relevant laws and regulations. The government should implement more refined and operable GCP to promote the balanced development of green credit across the country. The government needs to consider the differences between different industries in the implementation process, especially to strengthen the guidance of non-state-owned enterprises and capital-intensive industries. Financial institutions should further improve the green credit approval standards, fully consider the credit risk and environmental risk of enterprises, give enterprises different loan sizes and preferential interest rates and optimize the green credit approval process by adopting differentiated management strategies for enterprises of different natures and credibility status. Banks can increase the tilt of credit resources to environmentally friendly enterprises through credit channels, orderly promote the exit and transformation of highly polluting industries and reduce enterprise pollution emissions.

For highly polluting industries (such as steel, cement, chemicals, etc.), banks should force enterprises to provide third-party environmental risk assessment reports to quantify pollutant emissions and emission reduction technology paths. Adopt the “one-vote veto system for environmental protection” and stop granting credit to enterprises that fail to meet the national ultra-low emission standards or fail to complete rectification within the deadline. Establish a database for ranking industrial pollution intensity and gradually withdraw credit support for the bottom 20% of enterprises. For low-polluting industries (such as new energy, ecological agriculture, etc.), banks should prioritize the allocation of credit resources, shorten the credit approval process, and allow future carbon revenue to be pledged. Provide “green credit + technical consulting” bundled services and jointly evaluate the ecological benefits of projects with third-party institutions.

Secondly, enterprise governance and green technological innovation should be strengthened. Enterprises should actively disclose environmental information, and through increased data disclosure of environmental pollution indicators such as wastewater, waste gas, solid emissions, etc., they can send positive signals to society and improve the green reputation of enterprises, so that banks and the government can fully understand the environmental information of the enterprises and, thus, provide them with more credit capital support. Enterprises can also analyze the statistics of their own environmental information, urging enterprises to actively carry out rectification and transformation. GCP can improve environmental pollution by increasing green innovation output. R&D innovation reflects the core competitiveness of enterprises. The importance of green innovation projects should be fully recognized, and enterprises should be encouraged to increase green investment. The financial support for the realization of green technological innovation should be provided actively.

Thirdly, Chinese government and banks should take full advantage of the synergistic emission reduction effects of the digital economy and green credit. The government should reduce energy consumption by subsidizing enterprises’ digital transformation, while incorporating digital levels into green credit approval criteria. Banks should develop dynamic green credit interest rate incentives based on the degree of digital transformation and pollution reduction potential of enterprises, such as lower interest rates for enterprises that adopt artificial intelligence powered energy efficiency management systems. Banks should introduce digital assessment tools (such as carbon asset management platforms) to quantify the emission reduction effects of enterprises and serve as core indicators for green credit approval. For highly digital industries, banks should lower green lending rates and increase credit lines. For low-digital industries, banks’ green credit should include mandatory digital transformation clauses: new loans need to promise to complete key aspects of digitalization within a certain period of time; otherwise, interest rate penalty clauses will be triggered.