Synergy or Counteraction: Can Multiple Environmental Policies Promote High-Quality Green Transformation of Enterprises?—A Comprehensive Assessment Based on Double Machine Learning Algorithms

Abstract

As environmental issues grow increasingly complex and multifaceted, the synergistic effects of environmental policies and their implementation methods have become central to the environmental policy system. This paper analyzes panel data from all A-share-listed companies in China between 2013 and 2022 and aims at comprehensively evaluate the role and impact of command-and-control, market-incentive, and public-participation environmental policies along with their combinations on corporate green transformation by using a double machine learning method. The results indicate that (1) all three types of environmental policies positively influence corporate green transformation, with market-based policies having the most pronounced effect; (2) the synergistic effects of policy combinations further enhance corporate green transformation, especially the combination of market-incentive and public-participation environmental policies; (3) heterogeneity analysis highlights differences in the effects of these environmental policies and their combinations on corporate attributes and regional factors; and (4) mechanism analysis indicates that green innovation and financial constraints indirectly drive the high-quality green transformation of enterprises.

1. Introduction

In light of the interconnected crises of climate change, biodiversity loss, and environmental pollution, enterprises globally face the challenge of achieving high-quality green transformation. Under the Paris Agreement, countries will announce their updated emission reduction targets for the next decade at the UN Climate Change Conference. At the end of 2020, in response to the requirements of the Paris Agreement, the United Nations called on major emitters to achieve net-zero greenhouse gas emissions by 2050 and limit the rise in the global temperature to no higher than 1.5 degrees Celsius compared with industrialized levels. However, due to various factors such as time and the level of available technology, it is difficult to achieve the net-zero emission goal. If companies can reduce carbon emissions and achieve a green transition through the use of reasonable and effective environmental policies, such as command-and-control environmental policies, market-incentive environmental policies, and public-participation environmental policies, and their combinations, it will make a significant contribution to the global climate problem. On 22 September 2024, the United Nations adopted the Future Compact, which aims to accelerate the transition to a fossil-free world while advancing the Sustainable Development Goals and the Paris Agreement. Globally, China’s environmental quality remains low. As one of the largest energy consumers in the world, China has proactively set the “dual carbon” target of peaking carbon emissions by 2030 and achieving carbon neutrality by 2060. Enterprises, as the main consumer of resources and the main polluter of the ecological environment, are pivotal in realizing the “dual carbon” target and the main force of carbon reduction. The critical question, therefore, is how enterprises can initiate their green transformation. In 2021, China’s State Council issued guiding opinions aimed at promoting the green upgrading of various industries and infrastructure, with an emphasis on energy conservation, environmental protection, clean production, and the use of renewable energy. The Guiding Opinions on Accelerating the Green Development of the Manufacturing Industry propose building a leading force for the green development of the manufacturing industry and establishing a benchmark cultivation system that connects the “comprehensive benchmark” of green manufacturing with the “single-item benchmark” in specific sectors. The white paper on Green Development in the New Era of China emphasizes promoting the green transformation in traditional industries, promoting the greening of the entire value chain of agriculture, industry, and the service industry. It calls for technological-, model-, and standards-based innovation, while improving the greening level of traditional industries through the establishment of a target evaluation and assessment system for green development. Furthermore, it emphasizes strictly implementing the main responsibility of enterprises and the regulatory responsibility of the government. Given this backdrop, two crucial questions arise:

RQ1: Can the integration of command-and-control, market-incentive, and public-participation policies effectively drive high-quality green transformation in enterprises?

RQ2: What are the underlying mechanisms through which these policies operate?

Although the existing literature emphasizes the importance of policy synergy, they often focus on analyzing individual policy instruments, overlooking the synergistic effects of multiple policy instruments. There is also a lack of systematic empirical analysis on how environmental policies drive the high-quality green transformation of enterprises through internal mechanisms such as green innovation and financing constraints. Porter (1991) [1] believes that environmental regulation can force enterprises to improve green innovation and pollution control by increasing environmental costs. Petroni et al. (2019) [2] show that greater environmental regulatory pressure encourages firms to pursue green transformation and upgrading, driving the green transformation of enterprises. Aghion et al. (2012) [3] proves that carbon tax policies drive R&D in green technologies within the automobile industry. Poulsen and Lema (2017) [4] emphasized that green supply chain innovation promotes product and industrial upgrades, facilitating green transformation. Thompson et al. (2004) [5] noted that banks develop green lending standards and make loan decisions based on firms’ environmental performance. Matisoff (2015) [6] found that public-participation programs, such as environmental management systems, labeling and certification initiatives, and information disclosure mechanisms, may mitigate market barriers, thereby reducing firms engaging in myopic or sub-optimal behaviors. However, some scholars have emphasized the synergistic effects of diverse environmental policies. Lindberg, Markard, and Andersen et al. (2019) [7] pointed out that policy portfolios focus on understanding how multiple complex policies work together and must be analyzed or evaluated. The current discussions of the European Union (EU) and the Organization for Economic Co-operation and Development (OECD) on sustainable transformation face the need to understand how to combine the multiple objectives of multi-level, multi-instrument, and policy to address the complexity of their issues (Diercks, 2019; Solorio, 2011; Edmondson, 2019) [8,9,10]. In large countries like China (Liu et al., 2017) [11], there are significant differences in economic development among different regions, with varying economic foundations, resource endowments, levels of environmental pollution, and technological innovation capabilities (Hattori, 2017) [12]. In addition, the mechanisms of the effectiveness of different environmental regulation tools vary (Stavins, 1996) [13]. So, it is necessary for local governments in different regions to implement different environmental policies at different stages of economic development to improve the environmental governance efficiency. Not all policy combinations produce positive synergies (Borrás and Edquist, 2013) [14], and some may even have the opposite effect (Howlett and Rayner, 2017) [15]. Therefore, carefully designed policy combinations are essential to meet the specific implementation conditions, help companies achieve green transitions, and solve global climate challenges.

This paper attempts to reveal the mechanism of environmental policies promoting the green transformation of enterprises from a theoretical perspective. Empirically, it selects data from the Environmental Protection Law as a representative of command-and-control environmental policies, the Environmental Protection Tax Law of the People’s Republic of China as a representative of market-incentive environmental policies, and ISO 14001:2015 [1] environmental management system certification as a representative of public-participation environmental policies. Employing a double machine learning model, this study quantitatively evaluates the effectiveness of these policies in facilitating high-quality green transformation, addressing the limitations of existing research that often focuses on individual policies and their direct impacts. Furthermore, it analyzes the benefits of policy synergy for green development. This study contributes to the broader field of green development. Additionally, it provides insights that can assist policymakers in formulating more effective environmental regulations to expedite high-quality green transformation, respond to global calls for sustainability, and foster high-quality green economic growth.

Specifically, the panel data of all A-share-listed companies in China from 2013 to 2022 are used, and the double machine learning method is innovatively used for empirical tests. The findings indicate that the command-and-control environmental policy, market-incentive environmental policy, public-participation environmental policy, and their combinations can significantly promote the green transformation of enterprises. Additionally, green innovation and financing constraints mediate the effectiveness of these environmental policies. Moreover, variations in enterprise attributes and regional differences contribute to the differing impacts of these policies.

The contributions of this paper are as follows: Firstly, it innovatively incorporates a double machine learning algorithm into policy evaluation, analyzing the effectiveness of command-and-control, market-incentive, and public-participation environmental policies, as well as their combinations, on corporate green transformation. Secondly, while the existing literature primarily investigates command-and-control environmental policies and market-incentive environmental policies, this study integrates public-participation policies within the same analytical framework, providing a comprehensive comparison. Thirdly, this paper employs green total factor productivity as a metric to assess the high-quality green transformation of enterprises, comprehensively considering the entire spectrum of corporate production operations and management indicators to assess high-quality green transformation.

2. Institutional Background

In light of the urgent challenge of global climate change, carbon emission reduction and green development have become common concerns of the international community. Governments, international organizations, and enterprises of all countries are taking active measures to respond to climate change and promote the process of sustainable development. According to a report by the United Nations Environment Program, global carbon emissions must decrease by 7.6% each year in the next ten years in order to achieve the 1.5 °C temperature control target. The “2023 Sustainable Development Financing Report” released by the United Nations calls for the formulation of a new generation of sustainable industrial policies supported by comprehensive national planning, expanding investment, and promoting the establishment of a green, low-carbon, and circular development economic system to achieve sustainable industrial transformation, narrow the widening development gap between countries, and jointly achieve global sustainable development goals. At the same time, the “Paris Agreement” signed in 2016 clearly states that all countries should control the increase in the global average temperature to within 2 °C compared to pre-industrial levels, and strive to control the increase in temperature to within 1.5 °C. China has taken effective measures in environmental governance. Since the “Environmental Protection Law of the People’s Republic of China” was passed in 2014, it has been continuously revised and updated, changing the traditional old industrial production model of “polluting first and treating later”. The new “Environmental Protection Law” enhances environmental supervision and enforcement, increases the pollution prevention responsibilities of enterprises, and holds them legally accountable for environmental damage. However, the law’s impact varies among enterprises due to differing levels of pollution.

To accelerate global green development, marketization tools such as environmental tax reform, the implementation of taxes or subsidies, and the establishment of performance targets have been widely advocated. The European Union, for instance, has employed public procurement and funding reviews to ensure that funds are used for environmental protection. The United States requires companies to disclose information on the environmental impact of their operating areas and encourages companies to develop voluntary programs to reduce the environmental impact of their supply chains. In 2011, China’s carbon emission trading pilot was the first to launch local carbon emission trading market pilot projects in seven provinces, Beijing, Tianjin, Shanghai, Chongqing, Guangdong, Hubei, and Shenzhen, accumulating valuable experience for the establishment of a national carbon market. In 2018, the “Environmental Protection Tax Law of the People’s Republic of China” was implemented, stipulating tax reduction and exemption measures, and establishing a positive incentive mechanism for enterprises in the green transformation process of “pay more for more pollution, pay less for less pollution, and pay nothing for no pollution”, as well as a dynamic tax adjustment mechanism adjusted according to the actual situation. Each province determines specific tax rates based on its environmental carrying capacity, pollutant emission levels, and broader economic, social, and ecological development goals. The tax rates in six provinces, including Beijing, Tianjin, Hebei, Shanghai, Jiangsu, and Henan, are at a high level, while 12 provinces, including Heilongjiang, Liaoning, Jilin, Zhejiang, Anhui, Fujian, Jiangxi, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang, have determined the tax amount according to the lower limit.

To advance global environmental governance and public participation, the United Nations launched the “Race to Zero” campaign, which was launched on “World Environment Day” on 5 June 2020. At the same time, domestic environmental protection laws and relevant laws and regulations such as the Atmospheric Pollution Prevention and Control Law also provide legal protection for public participation. Through the government’s release of environmental monitoring information, environmental protection authority departments solicit opinions, questionnaires, symposiums, expert demonstration meetings, hearings, etc., allowing the public to participate in environmental governance. Additionally, individuals can voice their environmental concerns via social media, letters, and petitions. To assist organizations in minimizing their environmental impact and ensuring regulatory compliance, the International Organization for Standardization (ISO) has developed the ISO14001 environmental management system certification applicable to organizations of any type and size. In recent years, ISO14001 environmental management system certification has become an added advantage for enterprises’ bidding projects. For example, Ligao (Shandong) New Energy Technology Co., Ltd. obtained five system certifications including ISO14001 in August 2024, which not only enhanced the company’s market competitiveness, but also demonstrated the company’s commitment to high-standard management.

To accelerate the implementation of green development goals and complete the “14th Five-Year” energy-saving and carbon-reduction binding indicators, the State Council of China issued the “2024–2025 Energy Saving and Carbon Reduction Action Plan”, encouraging high-energy-consuming projects to use non-fossil energy. The high-quality green transformation of enterprises is not only an effective way to deal with global climate change, but also an inevitable choice to achieve high-quality development. To further advance the green and low-carbon transformation in the energy sector, in May 2024, the National Energy Administration released the “Compilation of Typical Cases of Green and Low-Carbon Transformation of Energy”, which includes 23 case studies from various provinces and regions, serving as valuable references for promoting green and low-carbon energy transformation across different sectors.

3. Literature Review and Hypotheses Proposed

Command-and-control environmental policies regulate the environmental behavior of enterprises through mandatory laws and standards by the government. Neoclassical economics argues that environmental policy regulations increase the cost of institutional compliance for enterprises, thereby inhibiting the innovation ability of enterprises through the crowding-out effect. In order to survive, high-polluting enterprises will transfer the cost of environmental protection taxes by misappropriating technology research and development funds (Karmaker et al., 2021) [16], thus crowding out the resources of enterprises for green innovation. Conversely, in 1995, Porter and Van der Linde (1995) [17] proposed the Porter hypothesis, which believes that reasonable environmental regulations can effectively guide enterprises in pollution control and technological innovation. Klemetsen (2018) [18] demonstrated that a sufficiently high cost of non-compliance would significantly promote green technological innovation in enterprises. Barbera and McConnell (1990) [19] found that the cost of emission reduction has a significant negative impact on the productivity of heavily polluting industries in the United States, leading to a 10% to 30% decrease in productivity. In contrast to these views, some scholars have identified a non-linear relationship between command-and-control environmental policies and green innovation. Song et al. (2020) [20] studied the U-shaped relationship between environmental regulation and green innovation by using data from 30 regions in China from 2009 to 2017. When the intensity of environmental regulation continues to increase, its effect will change from inhibition to promotion. Liao, ZJ (2018) [21] found that command-and-control tools have a significant positive impact on eco-organizational innovation, which in turn enhances the corporate reputation. Xie et al. (2019) [22] found that environmental regulations can increase the proportion of investment in new energy power generation.

Based on the above, we propose Hypothesis 1: command-and-control environmental policies can promote the green transformation of enterprises.

Market-incentive environmental policies leverage economic mechanisms, such as environmental taxes and fees, to encourage enterprises to undergo green transformations. For example, using the carbon emission trading pilot policy as a quasi-natural experiment, studies have found that market-incentive environmental policies can effectively reduce carbon emissions, reduce environmental pollution (Cheng et al., 2024) [23], improve the corporate environmental performance (Yu et al., 2022) [24], and improve urban energy utilization (Hong et al., 2022) [25]. In the short term, enterprises can obtain part of the input cost compensation from the government’s innovation incentive subsidies, which is conducive to enterprises carrying out green technological innovation. However, in the long run, the implementation of market-incentive environmental policies alone is not enough to promote green technological innovation in enterprises (Shi and Li, 2022) [26]. Additionally, the implementation of market-incentive environmental policies can incur high costs and may disproportionately affect various enterprises and individuals. For example, pollution charges may impose a “double burden” on enterprises, requiring them to pay residual pollution taxes in addition to paying for pollution reduction. Zhang et al. (2020) [27] found that green innovation, including green management innovation and green production innovation, and environmental information disclosure can significantly reduce corporate financing constraints, and the interaction between the two can further improve the financing conditions of enterprises. Moreover, green credit policies, which include loans, interest subsidies, and tax incentives, internalize environmental costs and influence behavioral choices in the credit market. These measures can bolster the substantive innovation incentives of enterprises, improve the willingness of banks to provide credit, and thus it can largely encourage enterprises to achieve green transformation (Sun et al., 2019) [28] and promote the development of green finance.

Based on the above, we propose Hypothesis 2: market-incentive environmental policies can promote the green transformation of enterprises.

Public-participation environmental policies emphasize societal involvement and oversight in environmental protection, evolving in response to the negative incentives posed by mandatory laws and regulations (Blackman, 2008) [29]. The Chinese government has enhanced the transparency of environmental policies through information disclosure and public participation mechanisms. The corporate environmental information disclosure system requires enterprises to disclose their environmental performance and accept social supervision. In addition, the government encourages public engagement in environmental protection activities, such as volunteer services and green consumption, to raise public awareness and motivate enterprises to meet social environmental needs while promoting green production and green management. At the same time, Riedl et al. (2017) [30] found that companies with a good environmental performance are more likely to be favored by investors, which helps alleviate corporate financing constraints. Compared with the command-and-control and the market-incentive policy tools, public-participation environmental policies are based on trust between enterprises and the government, reducing the cost of direct government regulation and giving enterprises greater flexibility and autonomy, which enables managers to choose technologies and R&D that are more suitable for their own situation. The main aims are investment to improve the environmental performance (Camisón, 2010) [31], to minimize internal costs in pollution control and emission reduction (Wang et al., 2024) [32], and to promote the green transformation of Chinese manufacturing enterprises (Li et al., 2019) [33]. In terms of whether enterprises effectively carry out green transformation, Rondinelli and Vastag (2000) [34] found that although the empirical research results on the impact of ISO14001 certification on the corporate performance are still inconclusive, at least in theory, ISO14001 can significantly improve environmental management. Christmann and Taylor (2006) [35] found that the proactiveness of enterprises in obtaining certification is positively correlated with consumers’ environmental awareness.

Based on the above, we propose Hypothesis 3: public-participation environmental policies can promote the green transformation of enterprises.

With the rise of environmental issues as a key policy area, policies that focus on single tools (Bachus and Vanswijgenhoven, 2018) [36] or binary interactions (Weber et al., 2014) [37] are no longer able to meet the needs of public policy in addressing ecological challenges that often involve different sectors and policy areas. Policy portfolio analysis addresses issues related to the coherence of multiple objectives, the alignment of multiple instruments, and the alignment of objectives and instruments, and is essential to understanding environmental performance improvement (Kern and Howlett, 2009) [38]. Liao (2018) [39] found that the combination of command-and-control, market-based, and information-based tools can reduce government costs, improve policy effectiveness, and promote enterprise eco-product innovation. Command-and-control policies facilitate corporate environmental technology innovation, while certain market-incentive instruments, such as pollution charges and government subsidies, also promote this innovation. Li et al. (2024) [40] found that a policy mix of penalties and subsidies can significantly improve a company’s environmental performance, and this impact is enhanced with the increase in fines and subsidies. Zhang and Fun (2024) [41] found that the combination of command-and-control environmental policies and public-participation environmental policies can significantly promote green development in cities. Leal et al. (2019) [42] examines the environmental policy mix between tradable permits and emission taxes and found that when the excess burden of taxation is significant, tradable permits policy with tax treatment is efficient at enhancing welfare in the presence of a consumer-friendly firm. Greco et al. (2022) [43] examined data from the Mannheim Innovation Panel on German firms and found that in the short and long term, the positive impact of a cross-instrumental policy mix on process ecological innovation is greater than that of general innovation policy instruments. Chen and Wu (2022) [44] found that the policy mix has a better implementation effect on carbon emission reduction, personal income, and travel costs. Zhou et al. (2020) [45] found that an environmental policies mix can increase the green total factor productivity of enterprises and promote the development of green industries.

Based on the above, we propose Hypothesis 4: the combination of command-and-control environmental policies and market-incentive environmental policies can promote the green transformation of enterprises.

Hypothesis 5: the combination of command-and-control environmental policies and public-participation environmental policies can promote the green transformation of enterprises.

Hypothesis 6: the combination of market-incentive environmental policies and public-participation environmental policies can promote the green transformation of enterprises.

Hypothesis 7: the combination of three types of environmental policies can promote the green transformation of enterprises.

4. Research Design

4.1. Data Sources and Filtering

This paper uses panel data of all A-share-listed companies in China from 2013 to 2022, including the green total factor productivity of enterprises, policies, the number of green patent applications of enterprises, and socio-economic data. The control variables and observed variables are sourced from National Bureau of Statistics of China (NBS), the China Environmental Statistics Yearbook, the China Environmental Yearbook, the China Statistical Yearbook, and China Economic Database. Data on three representative environmental policies are obtained from the State Council, the National Development and Reform Commission, and provincial government websites. Information on green patent applications comes from the State Intellectual Property Office (SIPO) of the People’s Republic of China. The data on ISO14001 environmental certification of enterprises come from CSMAR.

To ensure the authenticity and reliability and to enhance the universality and effectiveness of the conclusions, the obtained samples were screened and processed: First, delete enterprise samples with less than 10 years of listing time, that is, insufficient financial data. In addition, since R studio is used for data preprocessing, the missing values have a greater impact on the results, so the enterprise samples with insufficient financial data are excluded. Second, considering that illegal enterprises have a greater impact on the overall sample, ST enterprises are excluded. At the same time, non-real enterprises such as financial enterprises are excluded. Third, in order to eliminate the influence of extreme values and outliers, and prevent the interference of discrete data on the results, Winsorize processing is performed on the 1% and 99% quantiles of continuous variables.

4.2. Variables and Measurements

4.2.1. Core Explanatory Variables

• Measurement of the command–control environmental policy.

The “Environmental Protection Law” was revised in 2014 and officially implemented in 2015, emphasizing that enterprises must comply with environmental protection laws and regulations in the production process, take effective measures to reduce pollutant emissions, improve resource utilization efficiency, and promote green production methods. For example, Article 32 of the “Environmental Protection Law” stipulates that enterprises shall conduct environmental impact assessments in accordance with the law and take corresponding pollution prevention and control measures based on the assessment results. Article 44 clearly states that enterprises that violate environmental protection laws and regulations will be investigated for their administrative or criminal responsibility in accordance with the law. Some scholars have found that the “Environmental Protection Law” has an impact on multiple aspects of enterprises, such as environmental protection investment (Pan et al., 2024) [46], green innovation (Liu et al., 2021) [47], and financing constraints.

Therefore, this study uses the “Environmental Protection Law”, officially implemented in 2015, as a measure of the enforcement intensity of command-and-control environmental policies. According to the “Guidelines for Environmental Information Disclosure of Listed Companies” issued by the Ministry of Environmental Protection of China on 14 September 2010, enterprises are classified as heavy polluters and non-heavy polluters. The Chinese government classifies 16 categories of industries, including thermal power, iron and steel, cement, electrolytic aluminum, coal, metallurgy, chemical industry, petrochemical industry, building materials, paper making, brewing, pharmaceuticals, fermentation, textile, tanning, and mining as heavy polluters. Heavy polluting enterprises are used as the experimental group, and non-heavy polluting enterprises are used as the control group. Heavy polluting enterprises in 2015 and thereafter are 1, and non-heavy polluting enterprises before 2015 are 0.

2.

Measurement of market-incentive environmental policy.

The “Environmental Protection Tax Law of the People’s Republic of China” (hereinafter referred to as the “Environmental Protection Tax Law”) has been officially implemented since 1 January 2018. This law uses economic means to encourage enterprises to reduce pollutant emissions, promote resource conservation and efficient use, and promote enterprises to achieve green development. Its core is to impose economic penalties on enterprises that emit pollutants through taxation, thereby encouraging enterprises to reduce pollution emissions and improve environmental quality. The law specifies the tax standards for four categories of pollutants: air pollutants, water pollutants, solid waste, and noise, and the tax rates are adjusted according to the type and emission of pollutants. The more pollutants an enterprise emits, the higher the environmental protection tax it has to pay, and vice versa. This mechanism effectively internalizes environmental costs, prompting enterprises to pay more attention to environmental protection in the production process.

Since the implementation of the “Environmental Protection Tax Law”, enterprises have actively responded to national policies and continuously improved their environmental performance through technological innovation and management optimization. Many enterprises have begun to implement green supply chain management, select environmentally friendly materials, optimize production processes, and reduce energy consumption and waste generation. Guided by the “Environmental Protection Tax Law”, local governments have formulated corresponding environmental regulation policies and incentive measures in combination with local conditions, providing policy support for the green transformation of enterprises. For example, some local governments have introduced policies such as green credit and tax incentives to encourage enterprises to adopt clean production technologies and equipment, and improve the resource recycling rate.

Therefore, this study uses the “Environmental Protection Tax Law”, officially implemented in 2018, as a measure of the enforcement intensity of market-incentive environmental policies. Heavy polluting enterprises are used as the experimental group, and non-heavy polluting enterprises are used as the control group. Heavy polluting enterprises in 2018 and thereafter are 1, and non-heavy polluting enterprises before 2018 are 0.

3.

Measurement of public-participation environmental policy.

Referring to the approach of Arimura et al. (2008) [48], this paper uses ISO14001 environmental management system certification as a tool for public-participation environmental policies. The ISO14001 certification system started with the “Environmental Management System Specification and Use Guide” issued by the State Bureau of Technical Supervision in 1996, and is applicable to all types of organizations, including companies, institutions, and government agencies. Obtaining this certification means that the organization has reached international standards in environmental management and can ensure that the pollutants in the enterprise’s production process are effectively controlled to meet relevant standards. Some scholars have found that ISO14001 environmental certification has an impact on multiple aspects of enterprises, such as environmental management (Daddi et al., 2015) [49], environmental responsibility (POTOSKI and PRAKASH, 2013) [50], and total factor productivity (Treacy et al., 2019) [51].

Therefore, this paper will use whether the enterprise is on the list of ISO14001 environmental management system certification as a proxy variable for public-participation environmental policies, taking the value 0 before the enterprise obtains ISO14001 environmental certification, and taking the value 1 after obtaining ISO14001 environmental certification.

4.

A combination of three environmental policies.

Under the implementation of the above three types of policies, enterprises all have values from 0 to 1 at different time nodes, and the combination of the three types of policies can be made quantitatively to sum them.

4.2.2. Explained Variable

Green total factor productivity is the main driving force for transforming the economic development mode under energy-saving and emission-reduction constraints. It is an important indicator to measure the quality of an enterprise’s growth and development and the degree of green transformation. For the degree of high-quality green transformation of enterprises, the green total factor productivity of enterprises is selected as a proxy variable for quantification.

This article adopts a directional distance function and ML productivity index that considers undesirable outputs to measure the green total factor productivity of enterprises. Suppose that a certain system has n decision-making units $\left({\mathrm{DMU}}_{\mathrm{j}}, \mathrm{j}=1,2,\dots ,\mathrm{n}\right)$, and each decision-making unit has three types of input–output indicators, including m inputs (I = 1, 2, …, m), s1 expected outputs, and s2 undesirable outputs, and is represented by a vector, $\mathrm{x}\in {ℝ}^{\mathrm{m}}, {\mathrm{y}}^{\mathrm{g}}\in {ℝ}^{{\mathrm{s}}_1}, {\mathrm{y}}^{\mathrm{b}}\in {ℝ}^{{\mathrm{s}}_2}$, and define matrix X, Yg, Yb as follows: $\mathrm{X}=\left[{\mathrm{x}}_1,{\mathrm{x}}_2,\dots ,{\mathrm{x}}_{\mathrm{n}}\right]\in {ℝ}^{\mathrm{m}\times \mathrm{n}}$; $\mathrm{Y}=\left[{\mathrm{y}}_1^{\mathrm{g}},{\mathrm{y}}_2^{\mathrm{g}},\dots ,{\mathrm{y}}_{\mathrm{n}}^{\mathrm{g}}\right]\in {ℝ}^{\mathrm{s}1\times \mathrm{n}}$; $\mathrm{Y}=\left[{\mathrm{y}}_1^{\mathrm{b}},{\mathrm{y}}_2^{\mathrm{b}},\dots ,{\mathrm{y}}_{\mathrm{n}}^{\mathrm{b}}\right]\in {ℝ}^{\mathrm{s}2\times \mathrm{n}}$. The directional distance function is introduced, and it is expected to reduce undesirable outputs while increasing desirable outputs. The directional distance function is defined as:

$$\begin{array}{c}{\overrightarrow{\mathrm{D}}}_0\left(\mathrm{x},\mathrm{y},\mathrm{b};{\mathrm{g}}_{\mathrm{y}},-{\mathrm{g}}_{\mathrm{b}}\right)=\sup \left\{\mathsf{\beta }:\left(\mathrm{y}+{\mathsf{\beta }\mathrm{g}}_{\mathrm{y}},\mathrm{b}-{\mathsf{\beta }\mathrm{g}}_{\mathrm{b}}\right)\in \mathrm{p}\left(\mathrm{x}\right)\right\}\end{array}$$

(1)

Among them, g is the direction vector, $\mathrm{g}=\left(\mathrm{gy},-\mathrm{gb}\right)$, and β is the distance function value. The directional distance function of a certain decision-making unit k in period t can be solved by the following linear programming:

$$\begin{array}{c}\overrightarrow{{\mathrm{D}}_0^{\mathrm{t}}}\left({\mathrm{x}}_{\mathrm{k}}^{\mathrm{t}},{\mathrm{y}}_{\mathrm{k}}^{\mathrm{t}},{\mathrm{b}}_{\mathrm{k}}^{\mathrm{t}};{\mathrm{y}}_{\mathrm{k}}^{\mathrm{t}},-{\mathrm{b}}_{\mathrm{k}}^{\mathrm{t}}\right)=\max \mathsf{\beta }\end{array}$$

(2)

After obtaining the directional distance function by solving the linear programming, the ML index is further defined according to the directional distance function. The ML index from period t to period t + 1 is obtained as:

$${\mathrm{ML}}_{{\mathrm{TFP}}_{\mathrm{t}}^{\mathrm{t}+1}}{\displaystyle \frac{\underset{\left(1+{\mathrm{D}}_0^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}},{\mathrm{b}}^{\mathrm{t}};{\mathrm{y}}^{\mathrm{t}},-{\mathrm{b}}^{\mathrm{t}}\right)\right)}{\to }}{\underset{\left(1+{\mathrm{D}}_0^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1},{\mathrm{b}}^{\mathrm{t}+1};{\mathrm{y}}^{\mathrm{t}+1},-{\mathrm{b}}^{\mathrm{t}+1}\right)\right)}{\to }}}\ast {\displaystyle \frac{\underset{\left(1+{\mathrm{D}}_0^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}},{\mathrm{b}}^{\mathrm{t}};{\mathrm{y}}^{\mathrm{t}},-{\mathrm{b}}^{\mathrm{t}}\right)\right)}{\to }}{\underset{\left(1+{\mathrm{D}}_0^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1},{\mathrm{b}}^{\mathrm{t}+1};{\mathrm{y}}^{\mathrm{t}+1},-{\mathrm{b}}^{\mathrm{t}+1}\right)\right)}{\to }}}$$

(3)

4.2.3. Mediating Variables

• Green innovation

Through green innovation, enterprises fulfill their social responsibility toward the environment. Corporate green innovation capability refers to the ability of enterprises to actively adopt green technologies, green products, and green management innovation measures in the process of economic development, and to achieve sustainable development such as reducing environmental pollution, saving resources, and improving economic and social benefits through technological innovation, process optimization, and product research and development. Green innovation capabilities include technological innovation capabilities, management innovation capabilities, and institutional innovation capabilities.

According to existing research, this paper selects the number of green patent applications of enterprises to measure the green innovation of enterprises.

• Financing constraints

Nandy and Lodh (2012) [52] proposed that in order to avoid environmental risks, banks may reduce the loan of enterprises with poor environmental performance, so as to allocate loan funds more reasonably, which is of vital importance to promoting enterprises to improve their green innovation level and transform to green. Porter and Linde (1995) [17] believe that environmental regulation can force enterprises to improve their level of green innovation and participate in pollution control by increasing environmental costs, and thus put forward the Porter hypothesis.

For financing constraints, this paper uses the corresponding formula of −0.737 Size + 0.043 Size² − 0.04 Age with enterprise size and enterprise age as variables. The idea of measuring financing constraints quantitatively comes from Kaplan et al. (1997) [53]. First, the degree of financing constraints of enterprises is qualitatively divided according to the financial status of the enterprises in the finite sample, and then the quantitative relationship between the degree of financing constraints and the variables reflecting the enterprise characteristics is drawn, that is, the financing constraints index. Representative measurement methods include KZ index (Lamont et al., 2001) [54], SA index (Fee et al., 2009) [55], and WW index (Whited et al., 2006) [56]. In order to avoid the interference of endogeneity, Hadlock and Pierce (2009) [57] used the KZ method, and used the financial report of the enterprise to classify the type of financing constraint of the enterprise, only using two variables that do not change much with time and have strong exogeneity: enterprise scale and enterprise age to construct SA index: −0.737 Size + 0.043 Size² − 0.04 Age.

4.2.4. Controlled Variable

Referring to the research of other scholars in related fields, this paper selects the following control variables: enterprise size (Size), measured by the natural logarithm of the company’s total assets; asset–liability ratio (LEV), measured by dividing total liabilities by total assets; net asset income rate (ROA), measured by dividing net profit by total assets; current ratio (Liquid), measured by dividing current assets by current liabilities; and two dummy variables, industry (Industry) and year (Year). The table of variable definitions for this study is shown in

Table 1. Variable definition table.

Variable	Definition	Measurement
Lawtreat	Environmental Protection Law implemented in 2015	Heavy polluting enterprises in 2015 and thereafter are given 1, and 0 otherwise
Taxtreat	Environmental Protection Tax Law implemented in 2018	Heavy polluting enterprises in 2018 and thereafter are given 1, and 0 otherwise
ISO	ISO14001 Environmental Certification	Value 1 after the enterprise obtains ISO14001 environmental certification, 0 otherwise
LawTax	Environmental Protection Law and Environmental Protection Tax Law	Value 1 when simultaneously affected by environmental protection law and environmental protection tax, 0 otherwise
LawISO	Environmental Protection Law and ISO14001	Value 1 when simultaneously affected by environmental protection law and environmental certification, 0 otherwise
TaxISO	Environmental Protection Tax Law and ISO14001	Value 1 when simultaneously affected by environmental protection tax and environmental certification, 0 otherwise
Combination	A combination of three environmental policies	Summed up the values of three types of policies
Pullo	Heavily polluting enterprises or not	Value 1 for heavy pollution enterprises and 0 for non-heavy pollution enterprises
GI	Number of enterprise green patent applications	Take the natural logarithm of the total number plus 1 of green invention patents applied for by the enterprise
SA	Financing constraints	−0.737 Size + 0.043 Size2 − 0.04 Age
GreenTFP	Green total factor productivity of enterprises	A non-radial SBM-ML exponential measure was used
Size	Enterprise scale	Take the natural logarithm of the total assets of the company
LEV	Asset–liability ratio	Total liabilities/Total assets
ROA	Return on Assets	After tax net profit/Total assets
Liquid	Current ratio	Current assets/Current liabilities
Year		Dummy variable

.

Descriptive statistics of the selected variables are shown in

Table 2. Descriptive statistics for variables.

	N	Mean	Std. Dev.	Median	Min	Max
LEV	15,125	0.43	0.34	0.42	0.01	13.71
ROA	15,125	0.87	2.54	0.37	0.00	167.54
Liquid	15,125	0.93	0.07	0.96	0.32	1.00
Year	15,125	0.13	2.96	0.14	3	22.39
Size	15,125	22.18	1.33	21.98	17.05	28.64
GreenTFP	15,125	1.02	0.10	1.03	0.80	1.18
Pullo	15,125	0.40	0.49	0.00	0.00	1.00
Lawtreat	15,125	0.28	0.45	0.00	0.00	1.00
Taxtreat	15,125	0.42	0.49	0.00	0.00	1.00
ISO	15,125	0.04	0.20	0.00	0.00	1.00
LawTax	15,125	0.15	0.36	0.00	0.00	1.00
TaxISO	15,125	0.03	0.16	0.00	0.00	1.00
LawISO	15,125	0.02	0.13	0.00	0.00	1.00
Combination	15,125	0.01	0.10	0.00	0.00	1.00
GI	15,125	12.18	47.59	3.00	0.00	1257.00
SA	15,125	0.01	0.12	0.00	−0.50	0.81

.

4.3. Model Specification

This paper aims to study the impact of the three environmental policies and their combinations on the high-quality green transformation of enterprises. Different from most current studies that use traditional causal inference models to evaluate policy effects, this paper uses a double machine learning model (DML) to assess environmental policy effects.

More and more scholars are paying more attention to DML (Chernozhukov et al., 2018, Athey et al., 2019, Knittel and Stolper, 2021) [58,59,60]. Gao et al. (2024) [61] quantified the impact of the green credit policy on the green total factor productivity of enterprises through DML. Dongmei Wang et al. (2024) [62] used DML for empirical tests and found that the environmental information disclosure policy has significantly improved the carbon total factor productivity.

DML utilizes advanced machine learning algorithms to capture the complex nonlinear relationship (Yang and Wang, 2022) [63] between dependent variables and covariates and potential high-order interaction effects. When faced with a large number of control variables or covariates, traditional statistical methods may fail due to the problem of dimensional disasters, while DML can effectively estimate environmental policies in a high-dimensional environment by using efficient data dimensionality reduction technology and model selection methods. In addition, when dealing with causal effect estimation, DML can overcome the influence of confounding variables and provide unbiased and consistent causal effect estimation even in the presence of unobserved confounding factors. Moreover, the estimator constructed under the DML framework is usually based on a rigorous theoretical foundation, which ensures that the policy evaluation results obtained when certain conditions are met have statistical significance and reliability. DML flexibly selects different predictors according to the actual research needs, and allows for the processing of different types of dependent variables and various types of covariates. At the same time, by separating prediction and causal effect estimation, this method can reduce the impact of data noise or outliers and enhance the robustness of environmental policy evaluation results. Therefore, the use of DML can make up for the deficiencies of traditional models, and it is more suitable for this research.

Referring to Chernozhukov et al. (2018) [58], the partially linear model is constructed as follows:

$$\begin{array}{c}\mathrm{Enterprise}=\mathrm{Policy}{\mathsf{\theta }}_0+{\mathrm{g}}_0\left(\mathrm{X}\right)+\mathrm{U}, \mathrm{E}\left[\mathrm{U}|\mathrm{X},\mathrm{Policy}\right]=0\end{array}$$

(4)

$$\begin{array}{c}\mathrm{Policy}={\mathrm{m}}_0\left(\mathrm{X}\right)+\mathrm{V}, \mathrm{E}\left[\mathrm{V}\mid \mathrm{X}\right]=0\end{array}$$

(5)

$$\mathrm{X}=\left({\mathrm{X}}_1,\dots ,{\mathrm{X}}_{\mathrm{P}}\right)$$

Formula (4) is the main equation. Enterprise is the explained variable, indicating the green total factor productivity of the enterprise. Policy is the processing variable which, respectively, represents the implementation of command-and-control environmental policies, market-incentive environmental policies, public-participation environmental policies, and the combination of the three policies. After the policy is implemented, the value is 1, otherwise it is 0. X represents the multivariate control variable vector, and U and V are the disturbance terms, which are the main regression coefficients that we want to estimate. Formula (5) shows the dependence of the processing variable policy on the control variable X, which helps to describe and eliminate the regularization bias. It can be seen from Equations (4) and (5) that the multidimensional control variable vector X affects the processing variable policy through the function h(X), and affects the explained variable enterprise through the function g(X).

Since the specific form of it is unknown, its estimated value can be obtained through the double machine learning algorithm; U is the error term and its conditional mean is 0. The estimation of Equation (4) can be obtained:

$${\check{\mathsf{\theta }}}_0={\left({\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum }_{\mathrm{i}\in \mathrm{I}}{\hat{\mathrm{V}}}_{\mathrm{i}}{\mathrm{Policy}}_{\mathrm{i}}\right)}^{-1}{\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum }_{\mathrm{i}\in \mathrm{I}}{\hat{\mathrm{V}}}_{\mathrm{i}}\left({\mathrm{Enterprise}}_{\mathrm{i}}-{\hat{\mathrm{g}}}_0\left({\mathrm{X}}_{\mathrm{i}}\right)\right)$$

(6)

Among them, $\mathrm{i}$ represents the i-th observation value, $\mathrm{I}$ represents the total observation value, and $\mathrm{n}$ is the sample size.

5. Empirical Tests

5.1. Benchmark Regression Results

Table 3. Benchmark regression results.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.049 *** (0.002)	0.114 *** (0.001)	0.025 *** (0.001)	0.077 *** (0.002)	0.045 *** (0.002)	0.092 *** (0.001)	0.073 *** (0.003)

Note: The values in parentheses are robust standard errors, and ***, represent significance levels of 1%.

reports the regression results of the impact of command-and-control environmental policies, market-participatory environmental policies, public-participatory environmental policies, and their combinations and comprehensive policies on the green transformation of enterprises. The results in Table 3 show that the regression coefficients of the green total factor productivity of enterprises and the three types of environmental policies and their combinations are all positive, indicating that the three types of environmental policies and their combinations have a positive effect on the green transformation rate of enterprises. The coefficient of environmental protection law is 0.049, indicating that environmental protection regulations have a significant positive impact on the green transformation of enterprises. The coefficient of environmental tax is 0.114, which is the most significant impact on the green transformation of enterprises among all single policies, indicating that market-incentive environmental policies, such as tax incentives, have a more direct and effective role in promoting the green transformation of enterprises. The coefficient of participatory environmental policy is 0.025; although relatively small, it is still significant, indicating that public participatory environmental policy tools also play a certain role in the green transformation of enterprises. The coefficient of the combination of environmental protection law and environmental protection tax is 0.077, indicating that the joint implementation of the two is more effective than the separate implementation. The coefficient of the combination of environmental protection law and participatory environmental protection policy is 0.045. The results show that the combination of command-and-control environmental policy and public-participation environmental policy can also promote the green transformation of enterprises to a certain extent. The coefficient of the combination of environmental tax and participatory environmental protection policy is 0.092, which is the highest among all combinations of environmental policies, indicating that the combination of market-incentive environmental policies and public-participation environmental policies is the most effective. The coefficient of the comprehensive environmental policy is 0.073, indicating that the implementation of the combination of the three types of environmental policies has a significant positive impact on the green transformation of enterprises. At the same time, all the regression coefficients are significant at the statistical level of 1%, which indicates that the research results are highly credible.

5.2. Heterogeneity Tests

Furthermore, the issue of concern in this study is whether the policy tools have heterogeneous effects among enterprises with different attributes. According to the attributes of enterprises, the sample is divided into state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs), and heterogeneity tests are conducted on them. From the data results in

Table 4. Heterogeneity analysis of SOEs and non-SOEs.

		Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	SOEs	0.034 *** (0.002)	0.101 *** (0.001)	0.020 *** (0.002)	0.058 *** (0.002)	0.037 *** (0.002)	0.076 *** (0.002)	0.057 *** (0.003)
	non-SOEs	0.078 *** (0.002)	0.129 *** (0.002)	0.034 *** (0.002)	0.105 *** (0.003)	0.065 *** (0.003)	0.113 *** (0.003)	0.097 *** (0.004)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

, it can be seen that all coefficients are significant at the significance level of 1%. The coefficient of environmental protection law for SOEs is 0.034, and that for non-SOEs is 0.078, which indicates that environmental protection law has a positive impact on both types of enterprises, but non-SOEs respond more strongly to environmental protection law. The environmental tax coefficient of SOEs is 0.101, and that of non-SOEs is 0.129, which is higher, indicating that environmental tax has a more obvious incentive effect on non-SOEs. The coefficient of participatory environmental protection policy for SOEs is 0.020, and that for non-SOEs is 0.034, which is slightly higher, indicating that non- SOEs tend to achieve green transformation through market mechanisms and public participation. In the heterogeneity of the policy combination, for the combination of environmental protection law and environmental protection tax, the coefficient of SOEs is 0.058, and that of non-SOEs is 0.105, and the response of non-SOEs is stronger, indicating that non-SOEs are more likely to do so. When faced with the dual effects of regulations and economic incentives, it is more capable of flexibly adjusting its business strategy to achieve green transformation. The combination of environmental protection law and participatory environmental protection policy has a coefficient of 0.037 for SOEs and 0.065 for non-SOEs, which also shows that non-SOEs respond more significantly. The coefficient of the combination of environmental tax and participatory environmental protection policy for SOEs is 0.076, and that for non-SOEs is 0.113, and this combination is particularly effective in non-state-owned enterprises. The coefficient of the comprehensive environmental policy for SOEs is 0.057, and that for non-SOEs is 0.097, which is higher, indicating that under the influence of the comprehensive policy, the green transformation of non-SOEs is more significant.

For further study of whether the policy tools have heterogeneous effects among enterprises in different regions, this paper divides the 30 provinces in China into the eastern, central, and western regions for heterogeneity analysis, and the results are shown in

Table 5. Heterogeneity analysis in the eastern regions.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.054 *** (0.005)	0.095 *** (0.005)	0.042 *** (0.005)	0.074 *** (0.008)	0.05 *** (0.006)	0.085 *** (0.006)	0.076 *** (0.011)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

,

Table 6. Heterogeneity analysis in the middle regions.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.064 *** (0.003)	0.096 *** (0.004)	0.041 *** (0.003)	0.085 *** (0.005)	0.057 *** (0.004)	0.091 *** (0.004)	0.083 *** (0.006)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

and

Table 7. Heterogeneity analysis in the western regions.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.046 *** (0.002)	0.124 *** (0.001)	0.022 *** (0.002)	0.075 *** (0.002)	0.044 *** (0.002)	0.094 *** (0.002)	0.071 *** (0.003)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

, respectively. In the eastern region, the coefficient of environmental protection law is 0.054, the coefficient of environmental protection tax is 0.095, and the coefficient of participatory environmental protection policy is 0.042. These results show that command-and-control, market-incentive, and public-participation environmental policies all have a significant positive impact on the green transformation of enterprises in the eastern region. The impact of market-incentive environmental policies is slightly higher than that of command-and-control and public-participation environmental policies. At the same time, the coefficients of the policy combinations are all relatively high, especially the combination of environmental taxes and other policies, such as environmental taxes and participatory environmental protection policies with a coefficient of 0.085, indicating that in the eastern region, the combination of market incentives and public participation is more effective. In the central region, the coefficient of environmental protection law is 0.064, that of environmental protection tax is 0.096, and that of participatory environmental protection policy is 0.041, indicating that command-and-control, market-incentive, and public-participation environmental policies all have significant positive impact on the green transformation of enterprises in the central region. At the same time, the coefficient of the combination of environmental protection law and environmental protection tax is 0.085, which is the highest combination coefficient, indicating that in the central region, the combination of command-and-control and market-incentive environmental policies is an effective strategy to promote the green transformation of enterprises. In the western region, the coefficients for environmental protection law, environmental protection tax, and participatory environmental protection policy are 0.046, 0.124, and 0.022, respectively. The environmental protection tax exerts the most significant impact, suggesting that enterprises in the western region are more reliant on market-based incentives, like tax incentives, to attract investment and foster industrial growth. The combination coefficient of environmental tax and participatory environmental protection policy is 0.094, which is the highest combination coefficient, indicating that in the western region, the combination of market incentives and public participation is equally important.

5.3. Mediator Effect Analysis

From the above analysis, it can be seen that all three environmental policies and their combinations have a significant impact on the green transformation of enterprises. How does the former affect the green transformation of enterprises? In this regard, this paper conducts a mediation effect test on green innovation and financing constraints, and the results are shown in

Table 8. Analysis of mediation effect: green innovation.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GI	0.049 *** (0.006)	0.117 *** (0.005)	0.024 *** (0.005)	0.077 *** (0.008)	0.046 *** (0.008)	0.093 *** (0.006)	0.073 *** (0.001)
GreenTFP	0.049 *** (0.002)	0.116 *** (0.002)	0.025 *** (0.002)	0.077 *** (0.003)	0.046 *** (0.003)	0.093 *** (0.002)	0.073 *** (0.04)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

and

Table 9. Analysis of intermediary effect: financing constraints.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
SA	0.049 *** (0.006)	0.117 *** (0.005)	0.024 *** (0.005)	0.077 *** (0.008)	0.046 *** (0.003)	0.093 *** (0.006)	0.073 *** (0.01)
GreenTFP	0.049 *** (0.002)	0.116 *** (0.002)	0.025 *** (0.002)	0.077 *** (0.003)	0.046 *** (0.008)	0.093 *** (0.002)	0.073 *** (0.004)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

. First, the coefficient of environmental protection law on green innovation is 0.049, indicating that command-and-control tools have positively promoted green innovation. Then, controlling for green innovation, the coefficient of environmental protection law on the total factor green productivity of enterprises is 0.049, indicating that even if the mediating effect of green innovation is considered, command-and-control tools still have a significant positive impact on the green transformation of enterprises. The coefficients of environmental tax, participatory environmental protection policies, and policy combinations on green innovation and total factor green productivity of enterprises are also significantly positive, indicating that command-and-control environmental policies, market-incentive environmental policies, public-participation environmental policies, and their combinations enhance enterprises’ green productivity indirectly by promoting green innovation. Green innovation is an intermediate variable that helps enterprises achieve high-quality green transformation.

The results in Table 9 show that the coefficient of environmental protection law on financing constraints is 0.049, indicating that command-and-control tools promote the green transformation of enterprises by easing financing constraints. Controlling for financing constraints, the coefficient of environmental protection law on the total factor green productivity of enterprises is 0.049, indicating that easing financing constraints is a way for command-and-control tools to promote the green transformation of enterprises. The coefficients of environmental tax, participation in environmental protection policies, and policy combinations for financing constraints and total factor green productivity of enterprises are also significantly positive, indicating that command-and-control environmental policies, market-participatory environmental policies, public-participation environmental policies, and their combinations enhance the promotion of the enterprises’ green transformation by easing financing constraints.

5.4. Robustness Test

The empirical analysis results show that command-and-control environmental policies, market-incentive environmental policies, public-participation environmental policies, their combinations, and comprehensive policies all significantly promote the green transformation of enterprises. In order to ensure the robustness of the empirical results, this paper adopts three methods: replacing the machine learning model, replacing the explained variable, and adding control variables for a robustness test.

To verify the robustness of the benchmark regression results and ensure that they are not caused by the assumptions of a specific model, the double machine learning model is replaced with a random forest model. The random forest model is an ensemble learning method, which improves the prediction accuracy and controls overfitting by constructing multiple decision trees and voting or averaging. The data in

Table 10. Replace the machine learning model.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.059 *** (0.002)	0.066 *** (0.001)	0.017 *** (0.001)	0.052 *** (0.002)	0.044 *** (0.002)	0.063 *** (0.001)	0.041 *** (0.002)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

show that after using the random forest model, the impact of command-and-control, market-incentive, public-participation environmental policies, and their combinations on the green transformation of enterprises is still significant, and the size and sign of the coefficients and the benchmark regression results are similar, indicating that the main results are not due to bias in model selection. At the same time, the empirical results remain robust after being replaced by random forest models, which enhances the credibility of the benchmark regression results.

To verify the impact of environmental policies on the green transformation of enterprises from different perspectives, referencing to Huang et al. (2019) [64], the explained variable is replaced by green technology efficiency (GTEC). Green technology efficiency is an indicator of the resource utilization efficiency of enterprises in the green production process. The data in

Table 11. Replace the explained variable.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GTEC	0.048 *** (0.002)	0.115 *** (0.001)	0.024 *** (0.001)	0.078 *** (0.002)	0.046 *** (0.002)	0.093 *** (0.001)	0.076 *** (0.002)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

show that even if the explained variable is replaced by green technology efficiency, the impact of command-and-control, market-incentive, public-participation environmental policies, and their combinations on the green transformation of enterprises is still significant, and the empirical results are still robust. This shows that environmental policies not only promote the green transformation of enterprises as a whole, but also improve the technical efficiency of enterprises in green production. The impact of environmental policies on the green transformation of enterprises is multifaceted.

To further ensure the accuracy of the results and exclude other possible factors that may affect the green transformation of enterprises, the control variable return on the invested capital is added. The above selected enterprise size, enterprise age, ownership structure, asset–liability ratio, net asset income rate, and current ratio as control variables. The return on invested capital is an important indicator to measure the efficiency of enterprise capital utilization. This article adds the control variable return on invested capital, as well as the number of employees and the number of executives in the regression. The results are shown in

Table 12. Add control variables.

	Lawtreat	Taxtreat	ISO	LawTax	LawISO	TaxISO	Combination
GreenTFP	0.048 *** (0.002)	0.115 *** (0.001)	0.024 *** (0.001)	0.078 *** (0.002)	0.046 *** (0.002)	0.093 *** (0.001)	0.076 *** (0.002)

Note: The values in parentheses are robust standard errors, and *** represent significance levels of 1%.

. After adding the control variables, the positive effects of command-and-control, market-incentive, public-participatory environmental policies, and their combinations on the green transformation of enterprises still exist. This result also further verifies the robustness of the benchmark regression results, indicating that the positive impact of environmental policies on the green transformation of enterprises is not due to the interference of other unconsidered factors.

6. Discussions

6.1. Command-and-Control Environmental Policy

Command-and-control environmental regulation contributes to the green transformation of enterprises. These policies compel firms to advance technologically, triggering the “Porter effect”, which enhances competitiveness while ensuring compliance with environmental standards (André et al., 2009) [65]. Moreover, the mediating effect caused by the increase in green R&D investment and the masking effect brought about by the increase in operating costs can trigger a race to the top in green innovation among regions (Feng et al., 2021) [66], induce process-oriented green technology innovation and promoting the green technology of the entire industry. Conversely, it leads to higher borrowing costs, imposing financial and operational risks that diminish enterprises’ motivation to fulfill their green responsibilities (Yu et al., 2021) [67]. When those regulated can perceive that the increase in environmental protection production costs can be overcompensated by the net income, environmental regulatory tools are effective and efficient (Korhonen et al., 2015) [68]. Environmental regulation reduces corporate financing constraints by mitigating agency problems and attracting more investors, thereby improving corporate performance. In addition, command-and-control environmental policies use mandatory laws and regulations to constrain corporate information disclosure behaviors (Lapologang and Zhao, 2023) [69], which can suppress corporate information “greenwashing” behaviors, improve information transparency, and promote enterprises to take their own risks. By intensifying penalties for illegal activities by third-party environmental testing agencies, local governments can effectively regulate monitoring behavior, mitigate collusion between enterprises and testing agencies to evade pollution penalties, and further encourage green transformation among businesses.

6.2. Market-Incentive Environmental Policy

Market-incentive environmental policy tools can significantly influence corporate behavior. Firstly, governments can utilize environmental taxes to adjust production costs, encouraging enterprises to fulfill their environmental responsibilities and promote green transformation. When the tax burden surpasses the costs of pollution control, firms are likely to adopt cleaner production technologies and advanced pollution control equipment to reduce emissions (Karmaker et al., 2021) [16]. Therefore, environmental taxes will drive enterprises to actively seek technological innovations related to energy conservation and emission reduction, and increase their investment in green technology innovation, thereby accelerating the high-quality green transformation of enterprises. Secondly, the government can establish an emissions trading market to incentivize pollution control and green innovation. The Coase theorem advocates that externalities can be corrected through negotiations between the parties concerned, so as to maximize social efficiency (Ronald H. Coase, 1960) [70]. The emission trading market mainly internalizes the external cost of environmental pollution through the price mechanism to restrict the total amount of pollution discharged by enterprises (Baudry et al., 2021) [71]. Since the sale of surplus permits can generate economic benefits, coupled with the control of total emissions (Carmona et al., 2009) [72], enterprises will consciously seek the greening of production processes and end-of-pipe governance technology innovation under the drive of profit maximization goals, so as to reduce their own generation and emissions of pollutants (Lanoie et al., 2008) [73], and trade the remaining emission rights to other enterprises with emission reduction needs, which has a positive impact on the green technology innovation and high-quality green transformation of enterprises. Additionally, the government can offer environmental subsidies to address the funding gap for green innovation. Subsidies or tax refunds can effectively promote remanufacturing, enabling manufacturers to achieve maximum profits and a higher environmental performance (Shu et al., 2017) [74]. Institutional theory shows that organizations that comply with external institutional rules are more likely to survive and develop further (Dacin et al., 2002) [75]. According to the institutional theory, the subsidies provided by the government to the green innovation activities of enterprises can effectively alleviate the financing constraints of enterprises (Yu et al., 2021) [67] and reduce the research and development costs and risks of green processes of enterprises.

6.3. Public-Participation Environmental Policy

Public-participation environmental policy tools can significantly enhance the high-quality green transformation of enterprises. Firstly, public resistance to pollution and government propaganda for environmental protection promote the diffusion of ecological technology innovation (Zhang et al., 2019) [76], driving the development of corporate green technological innovation. The key role of public-participation environmental policies may lie in transforming social norms of corporate behavior, enhancing companies’ responsiveness to environmental issues, and reshaping their understanding of stakeholder responsibility. Given the lower marginal cost of pollution control, non-participants may be more effective in improving their environmental behavior by imitating industry leaders than participating companies (Matisoff, 2015) [77]. Secondly, the government encourages corporate green transformation and improves environmental governance outcomes by reaching self-originating environmental agreements with enterprises. The government sets higher environmental targets and stricter environmental responsibility requirements for enterprises. After weighing the costs of environmental obligations and the potential benefits, enterprises establish their own voluntary energy-saving agreement organizations, leading to increased resource utilization, reduced pollutant emissions, and ultimately, the high-quality green transformation of enterprises. Additionally, the government assigns environmental labels and certifications to enterprises, reduces information and search costs, which helps enterprises overcome market barriers and allows them to try new policy tools and technologies, promotes policy learning (Bui and Kapon, 2012) [78], and may provide companies with the ability to develop and disseminate incremental improvements (Börkley et al., 1998) [79]. The environmental department tags the environmental performance of heavily polluting enterprises, with the “heavy pollution tag” providing suggestions for green innovation and investment for related enterprises, facilitating dynamic adjustments in corporate green development strategies (Lu et al., 2023) [80]. Environmental certifications not only enhance corporate information transparency, but also promote technological progress by playing an incentive role (Jiang et al., 2020) [81], improving the corporate environmental performance, alleviating financing constraints, and increasing total factor productivity.

6.4. Environmental Policy Combinations

The combination of command-and-control and market-incentive environmental policies is more effective than either policy implemented in isolation. When these policy instruments are combined, they create a synergistic effect (“1 + 1 > 2”) that enhances the corporate environmental product and process innovation, ultimately improving the corporate performance (Liao et al., 2018; Cai et al., 2023) [39,82]. Deterrence theory holds that the severity of punishment is an essential component of effective law enforcement (Becker, 1968) [83]. If the penalties for environmental violations outweigh the costs of compliance, profit-oriented businesses will have an incentive to comply. Compared with punitive measures, incentive tools can compensate enterprises for the losses caused by the positive externalities of environmental governance activities, and motivate enterprises to participate in environmental practices (Lin et al., 2015) [84]. Additionally, increasing relevant subsidies and penalties can promote the formation of a long-term mutually beneficial strategy between the government and enterprises (Zhu et al., 2007) [85], and stimulate the production of renewable energy (Dou et al., 2021) [86].

The combination of command-and-control and public-participation policies fosters greater corporate green transformation than public-participation policies alone, but is less effective than command-and-control policies on their own. This may result from the significant capital and technological investments required to meet stringent environmental regulations. In contrast, participatory policies allow for more flexible resource allocation adjustments. When corporate resources are limited, firms often prioritize compliance with mandatory regulations over seeking additional incentives from participatory measures. Additionally, the effectiveness of public-participation policies depends on the market demand for green products and public awareness of environmental issues. If either is lacking, the incentive effect of participatory policies diminishes. However, the implementation and enforcement of command-and-control regulations tend to be stricter, ensuring compliance. In contrast, the effectiveness of participatory policies relies on corporate voluntarism and transparency, leading to less pronounced outcomes compared to mandatory regulations. Corporate responses to environmental policies are also linked to long-term strategies and goals. Firms recognizing that compliance with environmental regulations can enhance their brand image and market competitiveness may prioritize these regulations over the incentives offered by public-participation policies.

The combination of market-incentive environmental policies and public-participation environmental policies is particularly effective. Participatory policies are adaptable to the specific needs of various companies and industries, while environmental taxes provide a stable and universal incentive mechanism. This combination ensures policy coherence while maintaining the flexibility necessary to address diverse circumstances. Signal theory holds that environmental subsidies can generate positive certification effect signals, thereby amplifying the legitimacy of enterprises among external investors in a market with mature mechanisms (Yan and Li, 2018) [87], enhancing the environmental image and reputation of enterprises, and enhancing the competitiveness of enterprises. Relying solely on one type of policy increases compliance risks for companies; however, a multifaceted approach helps distribute risk. Additionally, this policy combination effectively tackles market failures and government dysfunctions, compensating for and enhancing the limitations of individual policy tools, leading to improved emission reductions and social welfare (Li and Lin, 2013) [88], as well as enhanced efficiency and equity.

6.5. Enterprise Attributes

In the pursuit of corporate operation and sustainable development, non-SOEs tend to attach greater importance to the substantial economic benefits that ESG performance can bring. Although SOEs often bear the burden of policy implementation, such as supporting employment goals, they also receive more policy support from the government (Lin et al., 1998) [89]. They are usually more proactive in disclosing their ESG practices and achievements to the public, aiming to enhance their market performance and overall performance through such transparency (Xie et al., 2019) [22]. Market-incentive environmental regulations, such as China’s carbon trading policy, significantly increase the technological innovation activities of non-SOEs while reducing the low-quality technological innovation behaviors of SOEs. In China, SOEs play a leading role in the economy. It is necessary to strengthen the supervision of SOEs’ innovation investment and guide SOEs to improve the quality of innovation. When designing environmental policies, policymakers should consider the characteristics and needs of different types of enterprises. For SOEs, increased guidance and support may be needed to improve their response to environmental policies. In contrast, policies aimed at non-SOEs should focus on creating a favorable market environment and incentive mechanisms to facilitate their green transformation.

6.6. Regional Differences

Command-and-control environmental policies have a similar impact across the eastern, central, and western regions of China. However, market-incentive environmental policies have a similar impact on the eastern and central regions, but the impact is the greatest in the western region. The eastern region, being more economically developed, has likely reached a certain threshold in responding to environmental policies, which diminishes the additional benefits of market incentives compared to those in the western region. In contrast, the western region is in a stage of accelerated development and has a more urgent demand for investment and economic growth. Market-incentive environmental policies such as tax incentives and subsidies can more effectively attract capital into the green field and promote green transformation. In addition, the western region may face more severe capital constraints. Market-incentive environmental policies can effectively reduce the financial burden on enterprises by providing fiscal subsidies, tax breaks, and other measures, thus stimulating enterprises to increase environmental investment, thereby having a greater impact in the western region. Public-participation environmental policies have a similar impact on the eastern and central regions, but the impact is smaller in the western region. The eastern region, with its more developed economy, higher education levels, and greater environmental awareness among residents, facilitates greater engagement in public-participation policies. Although the economic development level of the central region is slightly lower than that of the eastern region, it has also been growing rapidly in recent years, and the public’s environmental awareness has gradually improved. In contrast, the economic development level of the western region is relatively low, and the public’s environmental awareness and participation are not as good as in the eastern and central regions. At the same time, due to the relatively rich educational resources and more developed information dissemination channels in the eastern and central regions, the public can more easily obtain environmental information and participate in environmental policies more effectively. In addition, the effective implementation of public-participation environmental policies requires active promotion and extensive publicity by the government. Governments in the eastern and central regions may invest more resources in policy implementation and publicity, thereby increasing public participation. In the western region, there are few enterprises with environmental labels. Not many people pay attention to the disclosure of environmental incidents, making it difficult to create public opinion pressure on serious pollution incidents or enterprises. And due to limited resources in the western region, insufficient policy implementation and publicity efforts have led to lower public participation. Based on the above analysis, policymakers should consider regional differences when formulating environmental policies. We can further strengthen the market-incentive and public-participation mechanisms in the eastern region, continue to use market-incentive environmental policies to guide the green transformation of enterprises in the central region, increase tax incentives in the western region, and encourage the combination of public-participation and market mechanisms.

6.7. Green Innovation and Financing Constraints

Green technology innovation promotes high-quality regional economic development by unleashing energy-saving and emission-reduction effects, promoting the clean-up of the industrial structure, and leading market demand (Li et al., 2024) [90]. Command-and-control environmental policies force enterprises to innovate through technological innovation to meet regulatory requirements by setting environmental standards. This pressure encourages the development of new green technologies and products, while financial institutions are often willing to provide funding for innovative projects, thereby alleviating financing constraints. Market-incentive environmental policies can reduce the cost of enterprises to obtain environmental protection technologies and funds. Through measures such as tax breaks and subsidies, they can reduce the financial burden on enterprises and improve their enthusiasm for investing in green technologies. Public-participation environmental policies enhance the environmental image and social responsibility of enterprises, enhance the confidence of consumers and investors in enterprises, and improve the competitiveness of enterprises in the market, thus promoting green financing for enterprises. Green innovation and financing constraints serve as crucial mediators in the impact of environmental policies on the green transformation of enterprises. Policymakers should consider how to promote green innovation and alleviate financing constraints through environmental policies, so as to more effectively promote enterprises to achieve high-quality green transformation. Concurrently, enterprises should prioritize investment in green innovation while exploring diverse financing channels to leverage the incentives and support provided by environmental policies for sustainable development.

7. Conclusions and Recommendations

This paper empirically examines the relationship between three types of environmental policies—command-and-control, market-incentive, and public-participation policies—and the green transformation of A-share-listed companies from 2013 to 2022. The research results show that (1) all three policy types significantly promote high-quality green transformation in enterprises; (2) market-incentive policies have the most pronounced effect on the green transformation process, while different policies exhibit varying levels of influence; (3) synergistic effects arise when combining various policy types, particularly with the integration of market-incentive and public-participation policies, which shows the greatest enhancement of green transformation; (4) heterogeneity analyses indicate that the impact of environmental policies varies across enterprise types and regions, where non-SOEs exhibit a stronger response to all policy types and their combinations, especially with the synergy of market-incentive and public-participation policies, and the effects of command-and-control policies on green transformation are relatively consistent across eastern, central, and western regions, with market-incentive policies having the greatest impact in the west, while public-participation policies are less effective in that region; and (5) green innovation and financing constraints emerge as key mediating variables, through which environmental policies indirectly facilitate high-quality green transformation by promoting innovation and alleviating financing challenges.

This paper puts forward the following suggestions: First, policymakers should strengthen the application of policy combinations and actively try different policy combinations. By integrating command-and-control policy tools, market-incentive policy tools, and public-participation policies, complementary and synergistic effects can be achieved. Policymakers should design and implement comprehensive environmental policy programs to make full use of the advantages of various policy tools, especially the combination of market-incentive and public-participation environmental policies, to achieve a more effective and efficient green transformation.

Second, policymakers should consider differences in enterprise types and regional characteristics when formulating differentiated and targeted environmental policies to enhance their effectiveness and adaptability. Non-SOEs are more responsive to environmental policies. Policymakers should actively focus on this area, provide customized support and incentive measures, and help non-SOEs overcome the obstacles in the process of green transformation. In economically developed eastern regions, emphasis should be placed on developing market-incentive and public participation mechanisms, while in central regions, market-incentive policies can guide corporate green transformation. In the western region, market-incentive measures should be used more to promote green transformation.

Third, the government should implement measures to foster green innovation and alleviate financing constraints for enterprises. The government should encourage enterprises to innovate in green through measures such as R&D funding and tax incentives, and, at the same time, strengthen intellectual property protection to provide a good external environment for enterprise innovation. Additionally, it is advisable for the government and financial institutions to provide more green financial products, reduce the financing cost of green projects, and reduce the financial risks of the green transformation of enterprises through financial tools such as green credit and green bonds.

In this study, the synergy effect of command-and-control environmental policies, market-incentive environmental policies, and public-participation environmental policies is evaluated, but more types of policies need to be richer and more in-depth in the follow-up research.