Internal and External Cultivation to Drive Enterprises’ Green Transformation: Dual Perspectives of Vertical Supervision and Environmental Self-Discipline

Abstract

Central Environmental Protection Inspection (CEPI) is a major step in China’s environmental vertical supervision reform. With the multi-period Difference-in-Differences method, we assess the impact of CEPI on enterprise green transformation. In addition, we further explore the impact of enterprise environmental self-discipline. The results show that CEPI significantly promotes enterprise green transformation, and this effect on governance is further strengthened by environmental self-discipline. The synergistic governance effect of compound environmental regulation is pronounced, particularly in enterprises lacking government–enterprise relationships and in areas covered by CEPI “look back” initiatives and where local governments rigorously enforce environmental laws. The mechanism analysis reveals that CEPI mainly promotes enterprise green transformation by improving executive green cognition, boosting investment in environmental protection, and enhancing green innovation efficiency. This study provides a fresh perspective on analyzing the governance impact of CEPI and provides valuable insights for improving multi-collaborative environmental governance systems.

1. Introduction

In recent years, the global climate environment has exhibited a tendency to deteriorate, with various extreme climates frequently appearing. This occurrence poses severe challenges to human survival and development, making green development a focus of common concern worldwide [1,2]. As the largest developing country in the world, China has continuously optimized its industrial structure and promoted green and low-carbon transformations in the economic field. Enterprises, as the micro-subjects of economic and social operations and pollution emissions, are the backbone of achieving sustainable development [3]. However, some enterprises aiming to maximize profits still maintain a development mode characterized by high energy consumption, high pollution, and high emissions. They even illegally discharge pollutants such as wastewater and waste gas, seriously impeding the green transformation of the economy. This situation arises because, under the institutional background of environmental decentralization, local governments often relax environmental supervision in pursuit of economic development. They indulge and shield environmental violations by enterprises, gradually becoming the “umbrella” for polluting enterprises [4,5]. To address the problem of central environmental policies being “distorted” due to insufficient motivation for environmental governance under the local management system, the Chinese government has established the vertical regulatory mechanism represented by central environmental inspection (CEPI). This mechanism directly supervises the environmental governance behavior of local governments at different levels and promotes enterprise green transformation.

Enterprise green transformation entails protecting the ecological environment while pursuing economic benefits. However, the characteristic of enterprises as “rational economic agents” dictates that enterprises are primarily driven by self-interest and are confronted with the challenge of effectively balancing the triple bottom line—economy, society, and environment. The lack of internal motivation for enterprises to fulfill environmental responsibilities leads to the prevalence of environmental default behavior [6]. Therefore, government-imposed environmental supervision faces inherent gaps in oversight, and relying solely on institutional pressure and other government interventions in environmental governance models may not fully constrain enterprise environmental behavior [7]. Environmental self-discipline can fill this gap. Enterprises as independent subjects, through self-regulation and self-restraint, can promote the development of environmental protection practices; this “self-adjustment” behavior is environmental self-discipline. ISO 14001 environmental management system certification currently stands as the most representative enterprise environmental self-discipline globally [8]. It has been proven to reduce compliance costs for enterprises in dealing with mandatory environmental regulations, optimize the allocation of internal resources, and promote green development [9,10]. It can be seen that the environmental self-discipline mechanism has become an indispensable part in the context of the transformation from unidimensional governmental governance to multi-dimensional governance between the government and the public and enterprises. The environmental governance strategy of heavily polluting enterprises is influenced by the internal and external coupling of environmental self-discipline and vertical supervision, and exploring the green transformation of enterprises in the context of vertical supervision cannot overlook the power of environmental self-discipline.

Existing research on CEPI has predominantly focused on its impacts on enterprise economic performance [11], environmental performance metrics [12], and the behavioral adjustments of local governments [13]. However, a critical gap remains in understanding whether and how the stringent pressure exerted by CEPI fundamentally drives enterprises towards a holistic green transformation—a strategic shift characterized by simultaneous reductions in pollution intensity and enhancements in efficiency. While CEPI’s “top-down” regulatory pressure is well recognized [4,14], the potential synergistic role of internal corporate initiatives, such as environmental self-discipline (e.g., voluntarily adopting ISO 14001), in amplifying this transformation effect has been overlooked. Previous studies have primarily examined the effects of CEPI or corporate responses in isolation, neglecting the interaction between external institutional pressures and internal environmental governance mechanisms. This study aims to fill this gap by addressing the following questions:

RQ1. Does the vertical supervision represented by CEPI affect the enterprises’ green transformation and what are the impact pathways?

RQ2. Does synergy exist between vertical supervision and environmental self-discipline and what are the influencing factors?

To investigate the policy impact of CEPI, which is a typical form of vertical supervision, and its combined effect with environmental self-discipline on enterprise green transformation, we chose A-share listed enterprises in heavily polluting industries in China from 2013 to 2020 as the research sample. The results show that CEPI significantly promotes enterprise green transformation and synergizes with environmental self-discipline. This conclusion holds even after robustness tests, such as changing the variable measurement method, the placebo tests, and propensity score matching. Research into the process by which CEPI pressure affects enterprises identifies three primary transmission mechanisms: executive green cognition, environmental protection investment, and green innovation efficiency. Further heterogeneity analysis indicates that vertical supervision and its synergistic effect with environmental self-discipline are more significant for enterprises in areas where government–enterprise relationship is less relevant, CEPI “looks back,” and local governments have high environmental enforcement power. The marginal contributions of the current study, compared to that of previous research, include the following: First, our research is based on CEPI, which enriches the existing evaluation perspective on the effect of CEPI by identifying the effect of vertical management on enterprise green transformation. Existing research has focused on the effects of CEPI on enterprise economic performance [15,16] and environmental performance [17,18]; nonetheless, few studies assess whether pressure from inspectors can promote enterprise green transformation. The current study focuses on whether CEPI stimulates internal environmental governance motivation among enterprises under the environmental decentralization system, providing more in-depth evidence for promoting the green development of polluting enterprises at the end of the environmental governance chain.

Second, the present study explores the impact of the joint environmental governance model of vertical supervision and environmental self-discipline on enterprise green transformation, offering which a fresh perspective for constructing a “multi-collaborative governance” ecological environment protection system. Existing research on CEPI has overlooked enterprise initiatives and solely focused on its direct impacts under the “top-down” model [19,20]. Our study clarifies that, under the impact of CEPI, enterprises actively engage in environmental self-discipline and that external regulation strengthens their recognition of environmental protection and promotes “down-top” corporate green transformation. Our study also confirms the coupled supportive relationship between external regulation and environmental self-discipline, which provides empirical evidence for the promotion of environmental “multiple governance”.

Third, the present study explores the mechanism underlying the effect of vertical supervision on enterprise green transformation and examines the boundary conditions of this transformation, as influenced by its compound effect with environmental self-discipline. Consequently, the “black box” of vertical supervision policy is elucidated. Existing studies either have concentrated on the governance effect of CEPI from the government perspective [21,22] or analyzed the environmental governance strategy adjustments of polluting enterprises facing CEPI [18,23]. Few scholars place government and enterprises within the same institutional framework and research scenario. Therefore, our study, framed by the “central government–local government–enterprise” model, examines how CEPI influences the environmental governance decisions made by polluting enterprises, with a particular focus on the proactive environmental responsibilities undertaken by these companies. Through this research, we provide empirical insights aimed at refining the CEPI system and fostering sustainable business practices.

The paper is structured as follows: Section 2 introduces the institutional context, theoretical background, and the development of the hypotheses. Section 3 outlines the research methodology. Section 4 presents the core empirical findings, followed by an analysis of the influence mechanisms in Section 5. In Section 6, we provide additional results from further analysis. Section 7 offers a summary of the study’s main findings, and Section 8 extends these conclusions to the environmental governance challenges faced by other developing countries.

2. Institutional Background, Theoretical Analysis, and Hypothesis Development

2.1. Institutional Background

2.1.1. Vertical Supervision: Central Environmental Protection Inspection

Vertical supervision emerges at a pivotal moment under the practical shortcomings of the territorial governance model. The promulgation of the Environmental Protection Law of the People’s Republic of China (PRC) in 1989 marks the formal establishment of the environmental decentralization system in China. However, in the context of environmental decentralization and economic prioritization, some local officials have prioritized local economic development at the expense of the environment. This situation has led to issues such as “collusion” between government and enterprises, “free-riding,” and so on [24,25]. To address the challenge posed by the lack of accountability of the main environmental governance body under the environmental decentralization system, China has implemented a vertical supervision system represented by CEPI.

CEPI is a mechanism for environmental governance led by the central government that inspects and rectifies environmental issues through in-depth investigations and the supervision of local governments and enterprises. Its primary targets include provincial party committees, governments, and relevant departments. However, during inspection and investigation, it may also be delegated to certain prefecture-level party committees and governments to directly investigate and understand enterprises, ensuring a comprehensive coverage of the “party committee, government, enterprise” framework. Different from previous environmental supervision endeavors by environmental departments, CEPI operates under central leadership, elevating environmental protection inspections from departmental governance to national governance [26].

The first round of CEPI covered 31 provinces (autonomous regions and municipalities) in China. To enhance governance effectiveness, the inspection teams performed a “look back” in 2018. This initiative vigorously rectified formalism and bureaucracy in ecological environmental protection and strengthened government accountability. Compared to the first round of CEPI, the second round of CEPI expanded its scope to include relevant departments under the State Council and central enterprises with considerable ecological environmental protection tasks as inspection targets. This expansion indicates that the breadth and depth of CEPI are constantly expanding and deepening. As of 4 January 2024, the first batch of the third round of CEPI has completed the inspection stage phase, suggesting that CEPI efforts are progressing toward normalization, institutionalization, and legalization (Figure 1).

2.1.2. Environmental Self-Discipline: ISO 14001 Certification

In contrast to environmental regulation, the enterprise environmental self-discipline mechanism refers to enterprises, as independent subjects, exercising self-restraint on their market behaviors and taking the initiative to participate in environmental protection. Engaging in the International Organization for Standardization (ISO) 14001 certification represents a common practice among enterprises to demonstrate environmental self-discipline. ISO released the ISO 14001 environmental management system standard in 1996. This standard adheres to the principle of voluntarism, allowing enterprise executives to decide whether to establish and implement the ISO 14001 environmental management system. China has the highest number of ISO 14001 certifications [27]. This number rapidly increased from 127,630 in 2015 to 415,619 in 2022 (Figure 2). This surge indicates the strong emphasis of China on green development and the implementation of strict environmental regulations, rendering ISO 14001 certification an attractive choice for enterprises seeking environmental legitimacy and strengthening their competitive advantage.

ISO 14001 follows a dynamic cyclic process consisting of “planning, implementation, inspection, and optimization.” First, enterprises need to identify and evaluate the applicable environmental regulations and existing environmental factors in their production and operational activities to form an initial environmental review. Second, referring to the results of the review and considering the economic and technological capacity of the organization, enterprises formulate environmental policies and objectives. They establish the framework of the environmental management system and develop an environmental management plan, implementing it. Third, eligible enterprises proactively apply to third-party certification bodies; upon a successful audit, they obtain ISO 14001 certification. Subsequently, an independent third-party organization conducts regular audits to ensure that its implementation complies with the required standards. Fourth, failure to meet ISO 14001 requirements during annual reviews may lead to the risk of certification suspension or revocation. Therefore, enterprises must regularly review the progress of environmental objectives and the implementation of the ISO 14001 environmental management system. They must make improvements based on changes in external and internal conditions. In summary, ISO 14001 certification represents a self-environmental restraint mechanism, which fully reflects the initiative of enterprise environmental governance. It can effectively reduce government supervision costs and improve overall environmental protection outcomes.

2.2. Theoretical Analysis and Hypothesis Development

2.2.1. Impact of Vertical Supervision on Enterprise Green Transformation

Under the institutional framework of environmental decentralization, local governments wield significant autonomy in environmental governance, allowing them to implement differentiated environmental governance policies [28,29]. However, granting excessive autonomy in environmental governance to local governments may instead become a “shield” for the pollution problem. Driven by economic incentives, local governments aiming to increase tax revenues may encourage enterprises to use their funds for production and operational activities that can generate quick profits instead of green investment with a long return cycle [30]. This approach may lead to local authorities turning a blind eye to or condoning environmental violations by enterprises within their jurisdictions or even lowering environmental regulatory standards or distorting the implementation of central environmental policies to attract investment from high-energy-consuming and highly polluting industries [31]. To address issues such as “formalism,” “collusion between enterprises and government,” and the “race to the bottom” caused by insufficient environmental governance motivation under decentralized systems, China has created a vertically integrated environmental supervision mechanism represented by CEPI. This mechanism combines measures such as mass-oriented governance, top-level authoritative intervention, and top-down mobilization.

Operating within the “central government–local government–enterprise” pressure transmission framework, CEPI urges local governments to strengthen environmental regulation and motivates enterprises to improve their environmental performance, fostering enterprise green transformation. Enterprise green transformation involves embracing green development as the main guiding concept, relying on green technological innovation to shift from resource-intensive, polluting practices to sustainable models characterized by intensive resource utilization and environmental protection. Ultimately, pollution reduction and productivity enhancement are achieved [32,33].

Based on institutional theory, as a form of coercive institutional pressure, CEPI reshapes the incentive structures for local governments and enterprises through political authority and accountability mechanisms. CEPI emphasizes the “one post and two responsibilities” working mechanism and strictly investigates environmental accountability. Inspection outcomes serve as crucial criteria for the assessment, evaluation, appointment, and dismissal of local government officials. CEPI imposes considerable environmental accountability and promotion pressure on local officials, fully mobilizing their enthusiasm for environmental governance [34]. However, the source of environmental problems lies within enterprises. If local governments intend to fulfill their assigned environmental assessment tasks during supervision, they must vertically transmit institutional pressures to local enterprises by formulating environmental policies, imposing penalties, and implementing other coercive measures. This compels enterprises to perceive and respond to governmental regulatory pressures in pursuit of operational legitimacy, thereby significantly elevating the corporate prioritization of environmental issues and propelling organizational behavioral adaptations to achieve environmental compliance.

Nonetheless, the concept of “rational economic agents” suggests that enterprises are primarily driven by self-interest [6]. When confronted with environmental protection pressures, local governments must consider whether enterprises opt for proactive environmental management decisions, such as increasing green investment and researching and developing green innovation technology, based on the balance between the costs and benefits of environmental protection investments. On the one hand, the authority and independence of CEPI effectively diminish the discretionary power of local governments in environmental governance and strengthen the rigidity of environmental law enforcement [22,35]. Under the strong political pressure from CEPI, local governments intensify environmental law enforcement, increasing compliance pressure on enterprises. Consequently, environmental violation costs exceed environmental governance costs. On the other hand, CEPI alleviates the information asymmetry between central and local governments by establishing a petition mechanism and publicly disclosing environmental violations. Consequently, the cost of public participation in environmental supervision is reduced, and public engagement in environmental governance is encouraged [23,36]. Public participation weakens collusion between local governments and enterprises, breaks the constraints of local protectionism, and shifts the burden of environmental pollution externalities onto enterprises [37]. Moreover, CEPI serves as a powerful environmental signal, lowering the risk of green innovation and increasing the expected returns on environmental investments [38]. Driven by the anticipated revenue effect, enterprises tend to increase green investment, research into and the development of green innovation technology, and other preventive environmental governance strategies to gain competitive advantages. Accordingly, the following hypothesis is formulated.

Hypothesis 1. 

Under restricted conditions, the implementation of CEPI significantly promotes enterprise green transformation.

2.2.2. Impact of Vertical Supervision and Environmental Self-Discipline on Enterprise Green Transformation

With the continuous development of environmental protection and the increasing strengthening of enterprise environmental responsibility, the solution to environmental problems cannot rely solely on new government regulations but also on the revival of enterprise morality. Enterprise environmental self-discipline is different from the traditional command-and-control model, emphasizing the principles of an autonomous system, coordination and self-governance. As a kind of “self-generated and self-initiated” environmental governance behavior, environmentally self-disciplined enterprises may implement more stringent environmental protection policies than their competitors in order to raise the barriers to entry in the industry and the potential costs for competitors and to gain more stakeholder attention [39]. However, in essence, environmental self-discipline is still a kind of flexible ethical obligation behavior that is mainly encouraging and voluntary, and some enterprises use behaviors such as “light green” and “greenwashing” to whitewash themselves and deceive their stakeholders [40]. The absence of regulatory controls has limited the utility of environmental self-disciplines.

The fundamental basis for the effectiveness of environmental self-discipline is coupled with regulatory support. The green development of enterprises cannot only rely on the enterprises themselves, but they also need to form a synergy with external regulation. Through the interpenetration of external regulation and internal self-discipline, enterprises can benefit from adopting environmental self-discipline in order to realize the role of environmental protection as a constraint on management and to guide management, shareholders and others to make decisions in favor of environmental protection and the green development of enterprises. In our study, we use CEPI as an external regulatory policy shock and participation in ISO 14001 certification to measure firms’ environmental self-discipline behavior. On the one hand, CEPI gradually transmits institutional pressure to local governments through top-down political mobilization, compelling them to enhance local environmental regulation [22]. Under more rigid local government environmental regulations, enterprises with ISO 14001 certification are less susceptible to spot checks by local authorities in environmental enforcement. Moreover, they can more easily obtain government support for their environmental performance. This support may come in the form of environmental subsidies, tax incentives, and access permits [41]. The reason is that ISO 14001 serves as a top-notch complement to government-mandated standards, indicating that enterprises can further improve their environmental performance on the basis of voluntarily meeting mandatory environmental protection requirements [42]. From the resource dependence perspective, ISO 14001 certification essentially represents an enterprise’s proactive construction of a resource exchange conduit with the government. By signaling regulatory compliance to reduce oversight risk, enterprises secure policy resource support in exchange. Therefore, to cope with pressure from local governments, enterprises become more inclined to enhance their environmental performance by seeking ISO 14001 certification.

Moreover, under the strict regulatory environment of CEPI, firms lacking ISO 14001 certification face heightened risks of non-compliance penalties and reputational damage. The certification acts as a preventive risk management tool, enabling the proactive identification and control of environmental risks. This reduces potential violation costs and frees resources for proactive green investment and innovation, rather than reactive crisis management [43]. Adopting ISO 14001 is thus a core strategy for minimizing environmental risk and ensuring sustainable development under regulatory pressure.

Meanwhile, CEPI implementation prompts stakeholders to recognize that only the sustainable development of enterprises aligns with long-term developmental interests, amplifying the legitimacy signaling effect of ISO 14001 certification [44]. As environmental awareness increases, enterprises with ISO 14001 certification convey to external stakeholders their proactive fulfillment of environmental responsibilities, improving the market acceptance of their products and gaining market legitimacy [39]. Thus, ISO 14001 certification alleviates information asymmetry between enterprises and external investors, allowing investors to gain a better understanding of enterprise environmental self-discipline behavior among enterprises, which can then form a green reputation distinct from that of competitors.

In summary, ISO 14001 certification, as a manifestation of enterprise environmental self-discipline, can effectively serve as a bridge for enterprises to interact with the government and the public in environmental governance. Under the influence of CEPI, enterprises with ISO 14001 certification gain easier access to investment, allowing them to allocate sufficient funds to green innovation activities. Ultimately, through green innovation technology, enterprises reduce their dependence on the original polluting production methods, realizing a mutually beneficial green transformation of the environment and the economy. Thus, the following research hypothesis is proposed.

Hypothesis 2. 

Under other conditions, the compound effect of CEPI and environmental self-discipline significantly promotes enterprise green transformation.

Based on the above, the logical framework of this study is shown in Figure 3.

3. Method

3.1. Measures

3.1.1. Dependent Variable

The dependent variable is enterprise green transformation, measured using two dimensions: emission reduction and efficiency improvement. To quantify enterprise green transformation, this study uses the logarithm of industrial sulfur dioxide emissions per unit output (Emission) and total factor productivity (Tfp), respectively. Among these approaches, certain methods have been identified for measuring enterprise total factor productivity. These techniques include the following: fixed effects (FE) method, generalized method of moments (GMM), ordinary least squares (OLS) method, semi-parametric Levinsohn–Petrin (LP) method, and Olley–Pakes (OP) method. Compared to other methods, the LP and OP techniques can more effectively address the issue of selective bias in measuring enterprise total factor productivity [45]. However, the OP method assumes a monotonic increasing relationship between investment and productivity, which may lead to the exclusion of samples with zero investment. Levinsohn and Petrin (2003) [46] improved the OP method by using intermediate inputs instead of investment indicators, effectively addressing endogeneity issues caused by productivity shocks. This enhancement yields more reliable results. Additionally, the LP method exhibits lower data dependency and is particularly suitable for China’s manufacturing sector samples where intermittent investment is prevalent. Therefore, this study calculates total factor productivity by using the LP method and conducts robustness tests by employing the OP technique. The calculation formula for the LP method is as follows:

LnY_it = α₀ + α₁lnL_it + α₂lnK_it + α₃lnM_it + θControl_it + ε_it

(1)

In Equation (1), subscripts i and t index enterprises and years, respectively. Total output (Y) derives from firm operating revenue. Labor input (L) is quantified using total employee compensation. Capital input (K) is computed from net fixed asset value. Intermediate input (M) represents the sum of operating costs, sales expenses, administrative expenditures, and financial outlays, net of current period depreciation and amortization charges.

3.1.2. Core Explanatory Variables

• Central Environmental Protection Inspection: In January 2016, the pilot phase of CEPI was launched in Hebei Province, officially commencing its first round in July 2016 (see

  Table 1. Implementation of Central Environmental Protection Inspection.

  Batch	Time	Inspector Province	Punished Enterprises (Units)
  Pilot	4 January 2016 to 5 February 2016	Hebei	200
  First batch	12 July 2016 to 19 August 2016	Inner Mongolia, Heilongjiang, Jiangsu, Jiangxi, Henan, Ningxia, Yunnan, and Guangxi	8160
  Second batch	24 November 2016 to 30 December 2016	Beijing, Shanghai, Hubei, Guangdong, Chongqing, Shanxi, and Gansu	12,184
  Third batch	24 April 2017 to 28 May 2017	Tianjin, Shanxi, Anhui, Liaoning, Fujian, Hunan, and Guizhou	24,299
  Fourth batch	7 August 2017 to 15 September 2017	Jilin, Zhejiang, Shandong, Hainan, Sichuan, Qinghai, Xinjiang, and Tibe	37,088

  Note: The data are sourced from the “Central Environmental Protection Inspection’s Feedback to Provinces of Inspection” published on the website of the Ministry of Ecology and Environment (http://www.mee.gov.cn/).

  ). Each province had inspection teams stationed in four batches to conduct inspections. In the research sample, the pilot city implementing CEPI is designated as the processing group. If an inspection group is stationed in the province where the company is registered, the virtual variable Treat is set to 1; otherwise, Treat is set to 0. This study focuses on the impact of CEPI implementation on annual enterprise green transformation; thus, the pilot period and the first two batches of CEPI are classified as Phase 1 (2016), whereas the third and fourth batches are classified as Phase 2 (2017). During the years 2016 or 2017, the virtual variable Post is set to 1; otherwise, Post is set to 0. The explanatory variable CEPI is constructed as Treat × Post, indicating that when an inspection team is stationed in the province where the company is registered, CEPI = 1 for both the current and future periods; otherwise, CEPI = 0.

2.

Enterprise environmental self-discipline: ISO 14001 Environmental Management System Certification is recognized as the most reliable indicator of enterprise environmental self-discipline [47,48]. Therefore, we base our evaluation of enterprise environmental self-discipline on whether the enterprise has obtained the ISO 14001 environmental management system certification. Specifically, if the enterprise obtains the ISO 14001 certification in the current year, ISO 14001 is set to 1; otherwise, ISO 14001 is set to 0.

3.1.3. Control Variables

Referring to the research by Wang et al. (2021) [18] and Zeng et al. (2023) [23], we incorporate control variables at the levels of enterprise ownership structure, financial characteristics, and comprehensive governance, as well as industry and province characteristic variables. Detailed variable definitions are provided in the Appendix A.

3.2. Sample and Data

To more accurately observe the implementation effects of the environmental pollution index, we select a research sample consisting of listed enterprises in heavily polluting industries from the A-share listed enterprises from 2013 to 2020. The starting point chosen is 2013, whereas the endpoint selected is 2020 for the following reasons: (1) The policy was implemented starting from 2016. For estimation purposes, the sample must be retained for a certain period before the policy implementation. (2) Considering the convening of the 18th National Congress, China’s focus on environmental issues markedly increased. Choosing 2013 as the starting point can reduce the estimation error caused by the abrupt shift in central environmental governance pressure to a certain extent. (3) The sampling interval cannot be excessively lengthy; otherwise, other strategies may interfere with the estimation results.

The research sample underwent the following treatments to ensure data validity: (1) Enterprises with abnormal trading statuses (ST, *ST, PT) were excluded. (2) Enterprises with abnormal or missing data were removed. (3) Continuous variables were winsorized at the 1% and 99% levels before descriptive statistics were conducted to eliminate outlier interference on the regression test. Last, we obtained 5542 sample observations. The data in this study were sourced from the CCER, RESET, and CSMAR databases.

3.3. Model

The following empirical models are constructed based on the research settings:

Y_it = β₀ + β₁CEPI_pt + γX_it + Year_t × Ind_j + Firm_i + ε_it

(2)

Y_it = β₀ + β₁CEPI_pt + β₂ISO14001_it + β₃CEPI_pt × ISO14001_it + γX_it + Year_t × Ind_j + Firm_i + ε_it

(3)

Equation (2) is used to evaluate the impact of vertical supervision on enterprise green transformation, and Equation (3) examines the combined effect of vertical supervision and enterprise environmental self-discipline on green transformation. The subscripts i, j, t, and p represent enterprise, industry, time, and province, respectively, with X representing other control variables. Considering the inconsistency in industry changes, this study controls for year-industry fixed effects (Year_t × Ind_j) to address unobservable factors at the industry level that vary over time, a more stringent approach than controlling solely for year (Year_t) and industry (Ind_j) fixed effects. Firm_i represents individual fixed effects, and ε_i,t denotes the random disturbance term. Emission_it and Tfp_it denote the pollution emission intensity and total factor productivity of enterprise i in year t, respectively. Emission_it and Tfp_it are the dependent variables to quantitatively examine enterprise green transformation from the dimensions of emission reduction and efficiency improvement.

4. Empirical Analysis and Research

4.1. Descriptive Statistics and Correlation Test

The average Emission is −7.801, with a maximum of −4.063 and a minimum of −11.701 (see

Table 2. Descriptive statistics.

Variables	Mean	Std	Max	Min	Median	P25	P75
Emission	−7.801	1.447	−4.063	−11.701	−7.633	−8.664	−6.832
TFP	8.431	0.978	10.890	5.533	8.354	7.794	9.016
CEPI	0.627	0.484	1	0	1	0	1
ISO 14001	0.286	0.452	1	0	0	0	1
Dual	0.233	0.423	1	0	0	0	0
Top1	0.354	0.151	0.806	0.095	0.333	0.240	0.450
Soe	0.411	0.492	1	0	0	0	1
Ratio	0.437	0.202	0.947	0.064	0.429	0.275	0.586
Roa	0.035	0.062	0.199	−0.248	0.032	0.010	0.064
Tobinq	1.862	1.116	6.966	0.843	1.488	1.161	2.117
Cflow	0.059	0.067	0.269	−0.162	0.059	0.021	0.097
Age	2.317	0.722	3.296	0.693	2.485	1.792	2.944
HHI B	0.140	0.093	0.444	0.038	0.140	0.049	0.173
ProGDP	10.472	0.794	11.615	6.704	10.530	10.068	11.076

). Tfp has an average value of 8.431, with a maximum of 10.890 and a minimum of 5.533. These results indicate a substantial difference between the Emission and Tfp levels in heavily polluting industries. The average value of CEPI is 0.627, suggesting the widespread coverage of CEPI. The average value of environmental self-discipline ISO14001 is 0.286, with a median value of 0, indicating that the majority of enterprises have yet to implement an environmental management system. In addition, the remaining control variables are consistent with existing research findings.

All correlation coefficients between the dependent variables and the main variables are considerably less than 0.6, indicating the absence of a significant multicollinearity problem (see Appendix A 

Table A2. Correlation analysis.

Variables	Emission	TFP	CEPI	ISO 14001	Dual	Top	Soe	Ratio	Roa	Tobinq	Cflow	Age	HHI B	ProGDP
Emission	1.000
TFP	−0.949 a	1.000
CEPI	0.048 a	0.066 a	1.000
ISO 14001	−0.038 a	0.041 a	0.028 b	1.000
Dual	0.191 a	−0.153 a	0.065 a	0.005	1.000
Top	−0.332 a	0.288 a	−0.064 a	−0.037 a	−0.058 a	1.000
Soe	−0.380 a	0.285 a	−0.104 a	−0.084 a	−0.283 a	0.255 a	1.000
Ratio	−0.427 a	0.338 a	−0.106 a	−0.047 a	−0.096 a	0.078 a	0.300 a	1.000
Roa	−0.083 a	0.148 a	0.101 a	0.042 a	0.031 b	0.102 a	−0.131 a	−0.419 a	1.000
Tobinq	0.433 a	−0.370 a	−0.100 a	−0.042 a	0.062 a	−0.157 a	−0.157 a	−0.250 a	0.097 a	1.000
Cflow	−0.212 a	0.204 a	0.060 a	0.038 a	−0.037 a	0.127 a	0.021	−0.134 a	0.444 a	0.003	1.000
Age	−0.314 a	0.259 a	0.019	−0.072 a	−0.231 a	−0.008	0.502 a	0.322 a	−0.179 a	−0.041 a	0.026 c	1.000
HHI B	−0.209 a	0.183 a	−0.119 a	−0.021	−0.070 a	0.225 a	0.130 a	0.030 b	−0.002	0.013	0.047a	0.139 a	1.000
ProGDP	0.066 a	0.002	0.243 a	0.105 a	0.163 a	−0.113 a	−0.314 a	−0.163 a	0.108 a	−0.035 a	0.066 a	−0.241 a	−0.156 a	1.000

Note: a, b, and c indicate significance at 1%, 5%, and 10% levels, respectively.

). In addition, a variance inflation factor test (VIF) was conducted, yielding an average VIF of less than 2, further verifying the absence of multicollinearity in this model.

4.2. Baseline Findings

4.2.1. Hypothesis 1

The estimation coefficient of CEPI is −0.093, which is significantly negative at 1%, indicating a decrease of 9.3% on average in the emission intensity of sulfur dioxide from enterprises after CEPI implementation. This finding demonstrates the effectiveness of CEPI in inhibiting pollution emissions. CEPI has an estimation coefficient of 0.050, which is significantly positive at 5%; that is, the total factor productivity of enterprises implementing CEPI has increased by 5% on average, emphasizing the role of CEPI in enhancing enterprise productivity (see Columns (1)–(2) of

Table 3. Regression results for Hypotheses 1 and 2.

Variables	(1)	(2)	(3)	(4)
	Emission	TFP	Emission	TFP
CEPI	−0.093 ***	0.050 **	−0.089 ***	0.049 **
	(−3.54)	(2.46)	(−3.19)	(2.20)
ISO 14001			0.001	−0.001
			(0.02)	(−0.05)
CEPI × ISO 14001			−0.103 ***	0.053 **
			(−3.55)	(2.18)
Dual	−0.013	0.024	−0.013	0.024
	(−0.30)	(0.83)	(−0.30)	(0.83)
Top1	−0.726 *	0.247	−0.726*	0.246
	(−1.92)	(0.99)	(−1.92)	(0.99)
Soe	0.137	−0.140*	0.137	−0.141 *
	(1.19)	(−1.72)	(1.19)	(−1.72)
Ratio	−1.057 ***	0.569 ***	−1.056 ***	0.569 ***
	(−7.36)	(4.73)	(−7.35)	(4.73)
Roa	−2.497 ***	2.284 ***	−2.496 ***	2.284 ***
	(−13.60)	(12.82)	(−13.53)	(12.79)
Tobinq	0.085 ***	−0.026 *	0.085 ***	−0.026 *
	(4.76)	(−2.03)	(4.77)	(−2.04)
Cflow	−0.818 ***	0.593 ***	−0.817 ***	0.593 ***
	(−4.83)	(4.29)	(−4.84)	(4.29)
Age	−0.458 ***	0.234 ***	−0.457 ***	0.234 ***
	(−6.84)	(4.18)	(−6.86)	(4.19)
HHI B	−3.462 ***	2.787 ***	−3.463 ***	2.787 ***
	(−3.47)	(3.70)	(−3.46)	(3.70)
ProGDP	0.180	−0.104	0.181	−0.104
	(1.00)	(−0.89)	−0.013	(−0.89)
_cons	−7.444 ***	8.204 ***	−7.455 ***	8.209 ***
	(−3.95)	(6.93)	(−3.94)	(6.91)
Firm FE	Y	Y	Y	Y
Year × Ind FE	Y	Y	Y	Y
N	5509	5291	5509	5291
R2	0.946	0.926	0.946	0.926

Note: (1) We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. (2) The difference in the number of observations between Emission (5509) and TFP (5291) arises because the Levinsohn–Petrin (2003) [46] estimator for TFP requires continuous production data, leading to the exclusion of firms with missing intermediate input values.

).

Economically, a one-standard-deviation increase in CEPI enforcement reduces emission intensity by 0.031 SDs (−0.093/1.447 × 0.484), equivalent to a 0.045-log-unit decrease (0.031 × 1.447). This translates to an average 4.395% reduction in SO₂ emissions per unit output (1 − e^−0.045). Similarly, the same intervention raises TFP by 0.025 SDs (0.050/0.978 × 0.484), corresponding to a 0.024-log-unit increase (0.025 × 0.978), yielding an average 2.429% productivity gain (e^0.024 − 1).

CEPI increases the pressure on local environmental protection and improves the commitment of local governments to environmental governance and pollution penalties. Under the higher penalty regime, the cost saved from reducing one unit of pollution emissions and other indirect benefits increase, indicating an increase in the marginal benefit of emission reduction [49]. Driven by economic interests, enterprises will prioritize proactive environmental governance measures, such as increasing environmental investments and green technological innovation, to improve environmental performance. Consequently, they can achieve a green transformation that strikes a balance between emission reduction and efficiency improvement. In summary, CEPI, an innovative environmental monitoring tool, can promote the green transformation of polluting enterprises in the inspection area by addressing two dimensions: reducing pollution emissions and enhancing total factor productivity. Thus, H1 is supported.

4.2.2. Hypothesis 2

The regression coefficient of ISO14001 on Emission and Tfp is not significant, suggesting that individual environmental self-discipline exerts no substantial impact on enterprise green transformation. However, the interaction term of CEPI × ISO14001 yields a significantly negative regression coefficient on Emission at the 1% level and a significantly positive regression coefficient on Tfp at the 5% level. This finding suggests that among enterprises in the areas where CEPI is implemented, those practicing environmental self-discipline experience an average decrease of 10.300% in sulfur dioxide emission intensity and a simultaneous increase of 5.300% in total factor productivity compared to those not practicing environmental self-discipline (see Columns (3)–(4) of Table 3). Furthermore, for firms subject to CEPI, ISO 14001 certification delivers incremental benefits: an additional 3.868% emission reduction (1 − e^{−0.103×0.383}) and a further 2.051% productivity gain (e^{0.053 × 0.383} − 1). The implication is that under the combined condition of environmental self-discipline, the promoting effect of vertical supervision on enterprise green transformation is further enhanced, indicating a synergistic effect between vertical supervision and environmental self-discipline. H2 is thus supported. This synergistic effect may be due to the government pressure exerted by CEPI, which aligns the environmental protection goals of enterprises practicing environmental self-discipline with those of local governments. Consequently, both the regulatory costs for the government and the costs of pollution control and emission reduction for enterprises are reduced, facilitating enterprise green transformation.

The aforementioned conclusions are more intuitively illustrated in Figure 4, which depicts the synergy between vertical supervision and environmental self-discipline.

4.3. Robustness Checks

4.3.1. Parallel Trend Test

A key assumption when applying the Difference-in-Differences method is that both the experimental and control groups follow parallel trends. Referring to Beck et al. (2010) [50], the current study uses the event study approach to examine this assumption (Figure 5). The results show that the coefficient of environmental protection supervision policy is not significant prior to the introduction of the CEPI policy. This suggests there was no notable difference between the treatment and control groups before the policy implementation, thus satisfying the parallel trend assumption. After the CEPI policy was implemented, the regression coefficient for environmental protection inspections became statistically significant. This indicates that the policy has contributed to the green transformation of enterprises. A possible explanation is that CEPI has heightened awareness among both local governments and businesses about the central government’s commitment to environmental protection through established channels. Consequently, enterprises will increase their investment in environmental protection, intensify green innovation, and enhance production efficiency.

4.3.2. Change Variable Measurement Method

• An alternative measure of the dependent variable: Based on the research conducted by Mao et al. (2022) [51] and Du and Li (2020) [52], the current study uses the logarithm of the sum of water and air pollution equivalents, as well as the total factor productivity calculated using the OP method, as replacements for the dependent variable.
• An alternative measure of the explanatory variable: We use the following three methods to change the explanatory variables—(1) The official ISO 14001 environmental management system is valid for three years. Therefore, if an enterprise obtains ISO 14001 environmental certification in a particular year, the value of ISO14001_2 for that year and the following two years is set to 1; otherwise, it is set to 0. (2) ISO14001_3 is assigned a value of 0 before the enterprise acquires ISO 14001 certification and is set to 1 in the year of certification and thereafter. (3) To eliminate endogenous problems associated with CEPI and ISO 14001, we adopt the approach proposed by Jiang et al. (2021) [48]. We measure the maturity of ISO 14001 by considering the implementation time of ISO 14001 when the CEPI policy is enforced. Specifically, if a company holds a valid ISO 14001 certification in the year of CEPI implementation and the subsequent two years, ISO14001_4 is set to 1; otherwise, it is set to 0. The regression results further affirm the reliability of the research conclusions (see

  Table 4. Regression results for the changing variable measurement method.

  Variables	An Alternative Measure of the Explained Variable				An Alternative Measure of the Explanatory Variable
  	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
  	Emission2	TFP_OP	Emission2	TFP_OP	Emission	TFP	Emission	TFP	Emission	TFP
  CEPI	−0.035 **	0.041 **	−0.028	0.042 *	−0.091 ***	0.050 **	−0.085 ***	0.043 *	−0.093 ***	0.051 **
  	(−2.31)	(2.09)	(−1.68)	(2.04)	(−3.09)	(2.17)	(−2.83)	(1.89)	(−3.39)	(2.32)
  ISO14001			0.006	0.004
  			(0.50)	(0.23)
  CEPI × ISO14001			−0.046 **	0.042 *
  			(−2.59)	(1.71)
  ISO 14001_2					−0.007	0.003
  					(−0.27)	(0.15)
  CEPI × ISO14001_2					−0.102 ***	0.055 **
  					(−3.73)	(2.22)
  ISO14001_3							−0.034	0.028
  							(−1.11)	(0.95)
  CEPI × ISO14001_3							−0.138 ***	0.089 **
  							(−3.54)	(2.52)
  ISO14001_4									0.008	−0.008
  									(0.21)	(−0.25)
  CEPI × ISO14001_4									−0.088 ***	0.046 *
  									(−2.95)	(1.90)
  Dual	0.003	0.022	0.003	0.022	−0.013	0.024	−0.014	0.024	−0.013	0.024
  	(0.13)	(0.86)	(0.13)	(0.86)	(−0.30)	(0.83)	(−0.31)	(0.85)	(−0.30)	(0.83)
  Top1	−0.214	0.032	−0.214	0.032	−0.725 *	0.246	−0.715 *	0.236	−0.726 *	0.246
  	(−1.36)	(0.13)	(−1.36)	(0.14)	(−1.92)	(0.99)	(−1.90)	(0.95)	(−1.92)	(0.99)
  Soe	0.048	−0.153 **	0.048	−0.153 **	0.138	−0.141 *	0.142	−0.145 *	0.137	−0.140 *
  	(1.13)	(−2.17)	(1.14)	(−2.16)	(1.20)	(−1.72)	(1.24)	(−1.76)	(1.19)	(−1.72)
  Ratio	−0.574 ***	0.402 ***	−0.573 ***	0.402 ***	−1.057 ***	0.569 ***	−1.056 ***	0.568 ***	−1.056 ***	0.569 ***
  	(−6.48)	(3.59)	(−6.46)	(3.60)	(−7.36)	(4.74)	(−7.35)	(4.73)	(−7.35)	(4.73)
  Roa	−1.360 ***	2.284 ***	−1.358 ***	2.285 ***	−2.499 ***	2.285 ***	−2.492 ***	2.279 ***	−2.497 ***	2.284 ***
  	(−11.55)	(13.48)	(−11.54)	(13.46)	(−13.64)	(12.88)	(−13.67)	(12.77)	(−13.57)	(12.82)
  Tobinq	0.066 ***	−0.017	0.066 ***	−0.017	0.085 ***	−0.026 *	0.085 ***	−0.026 **	0.085 ***	−0.026 *
  	(5.11)	(−1.28)	(5.12)	(−1.28)	(4.76)	(−2.03)	(4.73)	(−2.05)	(4.74)	(−2.03)
  Cflow	−0.421 ***	0.608 ***	−0.421 ***	0.608 ***	−0.817 ***	0.592 ***	−0.817 ***	0.592 ***	−0.817 ***	0.592 ***
  	(−3.55)	(4.06)	(−3.56)	(4.06)	(−4.83)	(4.30)	(−4.84)	(4.30)	(−4.79)	(4.26)
  Age	−0.259 ***	0.147 ***	−0.259 ***	0.147 ***	−0.456 ***	0.234 ***	−0.454 ***	0.231 ***	−0.458 ***	0.234 ***
  	(−7.83)	(2.84)	(−7.84)	(2.84)	(−6.89)	(4.19)	(−6.88)	(4.15)	(−6.88)	(4.18)
  HHI B	−1.830 ***	2.063 ***	−1.831 ***	2.064 ***	−3.464 ***	2.789 ***	−3.467 ***	2.789 ***	−3.462 ***	2.788 ***
  	(−3.95)	(3.85)	(−3.95)	(3.85)	(−3.47)	(3.70)	(−3.47)	(3.70)	(−3.46)	(3.68)
  ProGDP	0.159	−0.074	0.161	−0.073	0.181	−0.104	0.185	−0.108	0.181	−0.104
  	(1.33)	(−0.88)	(1.35)	(−0.87)	(1.00)	(−0.89)	(1.02)	(−0.92)	(1.00)	(−0.89)
  _cons	−0.312	6.656 ***	−0.336	6.649 ***	−7.447 ***	8.206 ***	−7.492 ***	8.251 ***	−7.447 ***	8.207 ***
  	(−0.26)	(8.21)	(−0.27)	(8.16)	(−3.94)	(6.92)	(−3.96)	(6.94)	(−3.94)	(6.93)
  Firm FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  N	5509	5291	5509	5291	5509	5291	5509	5291	5509	5291
  R2	0.916	0.907	0.916	0.907	0.946	0.926	0.946	0.926	0.946	0.926

  Note: (1) We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. (2) Water pollution indicators include emissions of COD, NH₃-N, TN, and TP. Air pollution indicators cover SO₂, NOx, and soot. Pollutant emissions were converted into standardized pollution equivalents, summed, and then log-transformed to comprehensively quantify corporate pollution levels.

  ).

4.3.3. Placebo Test

• The random allocation of treatment groups: We randomly assigned treatment and control groups through bootstrap sampling and estimated the baseline model. For enhanced robustness, this process was repeated 1000 times to generate the distribution of t-values for the t-value estimates of the regression coefficients (Figure 6). Most of the t-values from the random samples are close to zero, with only a few estimates showing t-values greater than those from the baseline regression. This suggests that the impact of CEPI on the green transformation of enterprises is not affected by potential unobserved variables.

2.

Changing the implementation time: The Plan on Supervision and Inspection of Environmental Protection (Trial) passed in 2015 explicitly proposes to establish a national environmental protection supervision system. To further eliminate interference from environmental policies before 2016 and to assess whether enterprises exert any anticipatory psychological effects on CEPI, this study regenerates variable CEPI_2 by advancing the policy implementation by one year. Subsequently, it incorporates this adjusted variable into the baseline model for regression analysis (see

Table 5. Regression results for the placebo test (one period ahead of policy).

Variables	(1)	(2)	(3)	(4)
	Emission	TFP	Emission	TFP
CEPI_2	−0.030	0.006	−0.026	0.004
	(−0.88)	(0.21)	(−0.72)	(0.15)
ISO14001			0.003	−0.003
			(0.09)	(−0.11)
CEPI_2 × ISO14001			−0.037	0.007
			(−0.88)	(0.19)
Dual	−0.013	0.024	−0.013	0.024
	(−0.31)	(0.84)	(−0.31)	(0.84)
Top1	−0.723 *	0.246	−0.723 *	0.246
	(−1.91)	(0.99)	(−1.91)	(0.99)
Soe	0.138	−0.141 *	0.139	−0.141 *
	(1.20)	(−1.73)	(1.20)	(−1.73)
Ratio	−1.053 ***	0.567 ***	−1.053 ***	0.567 ***
	(−7.35)	(4.73)	(−7.33)	(4.73)
Roa	−2.492 ***	2.283 ***	−2.491 ***	2.282 ***
	(−13.47)	(12.90)	(−13.44)	(12.90)
Tobinq	0.084 ***	−0.025 *	0.084 ***	−0.025 *
	(4.72)	(−1.99)	(4.73)	(−2.00)
Cflow	−0.815 ***	0.590 ***	−0.814 ***	0.590 ***
	(−4.77)	(4.26)	(−4.77)	(4.26)
Age	−0.457 ***	0.234 ***	−0.457 ***	0.234 ***
	(−6.84)	(4.17)	(−6.86)	(4.17)
HHI B	−3.455 ***	2.781 ***	−3.456 ***	2.781 ***
	(−3.47)	(3.70)	(−3.47)	(3.70)
ProGDP	0.174	−0.100	0.175	−0.101
	(0.96)	(−0.85)	(0.96)	(−0.85)
_cons	−7.417 ***	8.198 ***	−7.428 ***	8.204 ***
	(−3.88)	(6.82)	(−3.87)	(6.79)
Firm FE	Y	Y	Y	Y
Year × Ind FE	Y	Y	Y	Y
N	5509	5291	5509	5291
R2	0.946	0.926	0.946	0.926

Note: We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, and * indicate significance at 1%, and 10% levels, respectively.

). The estimated coefficient of CEPI_2 is not statistically significant at the 10% confidence level. This finding indicates that the enhancing effect of CEPI on enterprise green transformation is not affected by relevant policies before inspection and that enterprises have no corresponding anticipations regarding CEPI implementation.

4.3.4. Exclusion of Impact of Other Environmental Regulatory Policies

• The second round of CEPI was conducted in six batches from 2019 to 2022. Thus, enterprises covered by this round during the sample period were excluded, and regression analysis was rerun.
• Key pollution units announced by the Ministry of Ecology and Environment are subject to vertical supervision from national and provincial ecological environmental departments. Thus, units announced during the sample period were excluded, and regression analysis was rerun.
• The inspection of Air Pollution Prevention and Control (2017–2018) significantly reduces pollution emissions [53]. Regression analysis is repeated after observations subject to this policy are excluded. Despite these adjustments, the regression results remain consistent with the baseline regression results (see

  Table 6. Regression results after excluding the influence of other environmental regulatory policies.

  Variables	The Second Round of CEPI				Key Pollution Units				The Inspection Over Air Pollution Prevention and Control (2017−2018)
  	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
  	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP
  CEPI	−0.085 ***	0.045 **	−0.080 ***	0.043 *	−0.090 ***	0.054 **	−0.077 **	0.046	−0.086 ***	0.044 *	−0.082 ***	0.042 *
  	(−3.30)	(2.18)	(−2.93)	(1.94)	(−2.86)	(2.15)	(−2.36)	(1.69)	(−3.08)	(1.95)	(−2.77)	(1.75)
  ISO14001			0.006	−0.004			0.027	−0.017			0.001	−0.000
  			(0.22)	(−0.21)			(0.77)	(−0.60)			(0.03)	(−0.00)
  CEPI × ISO14001			−0.095 ***	0.044 *			−0.107 ***	0.063 **			−0.094 ***	0.048 *
  			(−3.00)	(1.72)			(−3.00)	(2.05)			(−3.01)	(1.71)
  Dual	−0.031	0.035	−0.031	0.035	0.001	0.019	0.001	0.019	−0.006	0.018	−0.006	0.018
  	(−0.71)	(1.16)	(−0.71)	(1.15)	(0.01)	(0.57)	(0.02)	(0.56)	(−0.12)	(0.56)	(−0.12)	(0.56)
  Top1	−0.810 **	0.307	−0.810 **	0.307	−0.681 *	0.299	−0.681 *	0.298	−0.594	0.228	−0.594	0.228
  	(−2.14)	(1.25)	(−2.13)	(1.25)	(−1.77)	(1.06)	(−1.77)	(1.06)	(−1.35)	(0.78)	(−1.35)	(0.78)
  Soe	0.146	−0.150	0.146	−0.150	0.150	−0.149	0.151	−0.150	0.134	−0.136	0.134	−0.136
  	(1.14)	(−1.68)	(1.14)	(−1.68)	(1.10)	(−1.43)	(1.11)	(−1.43)	(1.06)	(−1.53)	(1.07)	(−1.53)
  Ratio	−1.014 ***	0.548 ***	−1.013 ***	0.548 ***	−1.154 ***	0.680 ***	−1.152 ***	0.679 ***	−1.219 ***	0.656 ***	−1.218 ***	0.656 ***
  	(−7.30)	(4.34)	(−7.29)	(4.34)	(−7.40)	(5.24)	(−7.35)	(5.22)	(−8.98)	(5.47)	(−8.95)	(5.48)
  Roa	−2.457 ***	2.268 ***	−2.455 ***	2.267 ***	−2.624 ***	2.364 ***	−2.619 ***	2.360 ***	−2.456 ***	2.256 ***	−2.455 ***	2.256 ***
  	(−12.28)	(11.80)	(−12.24)	(11.78)	(−11.22)	(11.87)	(−11.23)	(11.82)	(−11.29)	(10.92)	(−11.20)	(10.87)
  Tobinq	0.084 ***	−0.023 *	0.084 ***	−0.023 *	0.086 ***	−0.030 **	0.086 ***	−0.030**	0.063 ***	−0.011	0.063 ***	−0.011
  	(4.87)	(−1.82)	(4.86)	(−1.83)	(4.42)	(−2.16)	(4.46)	(−2.19)	(4.49)	(−0.93)	(4.51)	(−0.93)
  Cflow	−0.759 ***	0.517 ***	−0.758 ***	0.517 ***	−0.882 ***	0.616 ***	−0.883 ***	0.617 ***	−0.769 ***	0.556 ***	−0.768 ***	0.556 ***
  	(−4.60)	(3.68)	(−4.60)	(3.68)	(−3.99)	(3.56)	(−4.02)	(3.57)	(−4.54)	(3.87)	(−4.54)	(3.86)
  Age	−0.476 ***	0.246 ***	−0.475 ***	0.246 ***	−0.414 ***	0.200 ***	−0.414 ***	0.200 ***	−0.392 ***	0.189 ***	−0.391 ***	0.189 ***
  	(−6.62)	(4.12)	(−6.65)	(4.14)	(−6.46)	(3.93)	(−6.48)	(3.95)	(−6.28)	(3.57)	(−6.28)	(3.57)
  HHI B	−3.715 ***	3.013 ***	−3.716 ***	3.012 ***	−3.359 ***	2.710 ***	−3.355 ***	2.707 ***	−3.485 ***	2.695 ***	−3.485 ***	2.695 ***
  	(−3.41)	(3.82)	(−3.40)	(3.83)	(−3.39)	(3.58)	(−3.38)	(3.56)	(−3.18)	(3.31)	(−3.18)	(3.30)
  ProGDP	0.142	−0.092	0.144	−0.093	0.212	−0.124	0.215	−0.126	−0.014	−0.089	−0.012	−0.090
  	(0.76)	(−0.71)	(0.77)	(−0.71)	(1.04)	(−0.94)	(1.06)	(−0.96)	(−0.07)	(−0.52)	(−0.06)	(−0.52)
  _cons	−6.970 ***	8.016 ***	−6.992 ***	8.024 ***	−7.730 ***	8.376 ***	−7.775 ***	8.408 ***	−5.429 **	8.075 ***	−5.445 **	8.083 ***
  	(−3.62)	(6.10)	(−3.61)	(6.08)	(−3.68)	(6.34)	(−3.69)	(6.36)	(−2.65)	(4.65)	(−2.63)	(4.63)
  Firm FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  N	5139	4938	5139	4938	4425	4267	4425	4267	4599	4406	4599	4406
  R2	0.947	0.927	0.947	0.927	0.939	0.918	0.939	0.918	0.941	0.920	0.941	0.920

  Note: (1) We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively. (2) The data of key pollution units are collected from the Ministry of Ecological Environment of China and the public information collation of each prefecture-level city. The data of the inspection over Air Pollution Prevention and Control (2017) are sourced from the “Intensified Supervision Plan for Comprehensive Air Pollution Prevention and Control in Autumn and Winter of 2017–2018 in Beijing, Tianjin, and the Surrounding Areas” published on the website of the Ministry of Ecology and Environment (http://www.mee.gov.cn/).

  ).

4.3.5. Other Robustness Tests

• Cluster adjustment: In the empirical tests mentioned earlier, standard errors were adjusted at the provincial level. To further address concerns such as heteroscedasticity and serial correlation, this study uses robust standard error clustering at the enterprise level.
• Propensity score matching: This study employs propensity score matching to select samples from two groups, mitigating endogeneity issues arising from sample self-selection bias. Specifically, we use indicators, such as the joining together of two jobs (Dual), ownership concentration (Top), the asset–liability ratio (Ratio), return on assets (Roa), operating cash flow (Cflow), enterprise age (Age), and the Herfindahl index (HHI_B) to select control cities for each treated city with a 1:1 nearest neighbor matching algorithm.
• The exclusion of direct-controlled municipalities. Direct-controlled municipalities (Beijing, Shanghai, Tianjin, and Chongqing) operate at an administrative level different to that of prefecture-level cities. Their economic scale, pollution emissions, and innovation capabilities may differ significantly, potentially influencing the results of this study. Therefore, we exclude samples from direct-controlled municipalities from the full sample and re-estimate the regression analysis. The results after regression remain consistent with the previous findings (see

  Table 7. Regression results for other robustness tests.

  Variables	Cluster Adjustment								Propensity Score Matching				Exclusion of Direct-Controlled Municipalities
  	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)
  	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP
  CEPI	−0.093 ***	0.050 **	−0.089 ***	0.049 **	−0.092 ***	0.050 **	−0.089 ***	0.049 **	−0.105 **	0.072 *	−0.105 **	0.082 **	−0.083 **	0.043 *	−0.074 **	0.039
  	(−3.46)	(2.28)	(−3.18)	(2.13)	(−3.62)	(2.41)	(−3.24)	(2.18)	(−2.67)	(2.02)	(−2.67)	(2.09)	(−2.65)	(1.88)	(−2.24)	(1.58)
  ISO14001			0.001	−0.001			0.001	−0.001			0.017	0.022			0.012	−0.006
  			(0.03)	(−0.05)			(0.02)	(−0.04)			(0.55)	(0.84)			(0.42)	(−0.29)
  CEPI × ISO 14001			−0.103 ***	0.053 **			−0.103 ***	0.053 **			−0.085 *	0.066 **			−0.095 ***	0.046 *
  			(−3.23)	(2.04)			(−3.31)	(2.19)			(−1.81)	(2.12)			(−3.17)	(1.83)
  Dual	−0.013	0.024	−0.013	0.024	−0.013	0.024	−0.013	0.024	−0.012	0.009	−0.012	0.009	−0.031	0.030	−0.031	0.030
  	(−0.52)	(1.21)	(−0.52)	(1.21)	(−0.36)	(0.81)	(−0.36)	(0.81)	(−0.15)	(0.24)	(−0.15)	(0.26)	(−0.67)	(0.93)	(−0.67)	(0.93)
  Top1	−0.726 ***	0.247	−0.726 ***	0.246	−0.726 **	0.247	−0.726 **	0.246	−0.908	0.104	−0.903	0.109	−0.804 *	0.311	−0.804 *	0.311
  	(−3.69)	(1.62)	(−3.69)	(1.62)	(−2.40)	(1.17)	(−2.40)	(1.17)	(−1.61)	(0.32)	(−1.60)	(0.34)	(−2.04)	(1.25)	(−2.04)	(1.25)
  Soe	0.137	−0.140*	0.137	−0.141 *	0.137	−0.140	0.137	−0.141	0.266	−0.176	0.266	−0.176	0.122	−0.141	0.122	−0.141
  	(1.55)	(−1.86)	(1.56)	(−1.86)	(1.06)	(−1.28)	(1.07)	(−1.28)	(1.21)	(−1.58)	(1.20)	(−1.58)	(0.93)	(−1.54)	(0.93)	(−1.55)
  Ratio	−1.057 ***	0.569 ***	−1.056 ***	0.569 ***	−1.057 ***	0.569 ***	−1.056 ***	0.569 ***	−0.961 ***	0.770 ***	−0.962 ***	0.769 ***	−1.050 ***	0.562 ***	−1.050 ***	0.562 ***
  	(−10.33)	(6.46)	(−10.33)	(6.46)	(−7.34)	(4.60)	(−7.34)	(4.61)	(−4.67)	(4.40)	(−4.67)	(4.39)	(−6.90)	(4.33)	(−6.90)	(4.33)
  Roa	−2.497 ***	2.284 ***	−2.496 ***	2.284 ***	−2.497 ***	2.284 ***	−2.496 ***	2.284 ***	−2.268 ***	2.326 ***	−2.271 ***	2.331 ***	−2.426 ***	2.158 ***	−2.426 ***	2.157 ***
  	(−14.08)	(13.66)	(−14.07)	(13.65)	(−12.72)	(12.38)	(−12.74)	(12.39)	(−6.32)	(10.40)	(−6.27)	(10.44)	(−11.72)	(11.66)	(−11.72)	(11.66)
  Tobinq	0.085 ***	−0.026 **	0.085 ***	−0.026 **	0.085 ***	−0.026 *	0.085 ***	−0.026 *	0.091 ***	−0.015	0.092 ***	−0.016	0.079 ***	−0.018	0.079 ***	−0.018
  	(6.78)	(−2.52)	(6.77)	(−2.52)	(4.97)	(−1.95)	(4.96)	(−1.95)	(3.18)	(−0.94)	(3.18)	(−0.98)	(4.54)	(−1.48)	(4.54)	(−1.49)
  Cflow	−0.818 ***	0.593 ***	−0.817 ***	0.593 ***	−0.818 ***	0.593 ***	−0.817 ***	0.593 ***	−0.554 **	0.425*	−0.557 **	0.423 *	−0.753 ***	0.549 ***	−0.753 ***	0.549 ***
  	(−5.78)	(4.96)	(−5.78)	(4.96)	(−5.10)	(4.53)	(−5.10)	(4.53)	(−2.31)	(1.75)	(−2.29)	(1.76)	(−4.34)	(3.91)	(−4.34)	(3.91)
  Age	−0.458 ***	0.234 ***	−0.457 ***	0.234 ***	−0.458 ***	0.234 ***	−0.457 ***	0.234 ***	−0.505 ***	0.228 ***	−0.505 ***	0.231 ***	−0.472 ***	0.229 ***	−0.472 ***	0.229 ***
  	(−9.88)	(6.05)	(−9.88)	(6.05)	(−7.18)	(4.42)	(−7.19)	(4.42)	(−4.75)	(2.81)	(−4.74)	(2.83)	(−6.54)	(3.70)	(−6.54)	(3.71)
  HHI B	−3.462 ***	2.787 ***	−3.463 ***	2.787 ***	−3.462 ***	2.787 ***	−3.463 ***	2.787 ***	−3.500 ***	3.260 ***	−3.507 ***	3.264 ***	−3.696 ***	2.989 ***	−3.696 ***	2.989 ***
  	(−6.45)	(6.24)	(−6.45)	(6.24)	(−4.19)	(4.40)	(−4.19)	(4.40)	(−3.14)	(3.00)	(−3.17)	(3.02)	(−3.10)	(3.36)	(−3.10)	(3.36)
  ProGDP	0.180 *	−0.104	0.181*	−0.104	0.180	−0.104	0.181	−0.104	0.115	−0.160	0.116	−0.154	0.270	−0.143	0.270	−0.144
  	(1.69)	(−1.56)	(1.70)	(−1.57)	(0.83)	(−0.91)	(0.84)	(−0.91)	(0.42)	(−1.16)	(0.43)	(−1.14)	(1.33)	(−1.10)	(1.33)	(−1.11)
  _cons	−7.444 ***	8.204 ***	−7.455 ***	8.209 ***	−7.444 ***	8.204 ***	−7.455 ***	8.209 ***	−6.754**	8.655 ***	−6.769**	8.586 ***	−8.254 ***	8.549 ***	−8.291 ***	8.566 ***
  	(−6.66)	(11.83)	(−6.67)	(11.84)	(−3.29)	(6.92)	(−3.29)	(6.93)	(−2.38)	(6.23)	(−2.38)	(6.28)	(−3.87)	(6.46)	(−3.88)	(6.45)
  Firm FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y
  N	5509	5291	5509	5291	5509	5291	5509	5291	2583	2547	2583	2547	4843	4675	4843	4675
  R2	0.946	0.926	0.946	0.926	0.946	0.926	0.946	0.926	0.954	0.937	0.954	0.937	0.942	0.922	0.942	0.922

  Note: We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.

  ).

5. Mechanism Analysis

The preceding theoretical analysis indicates that CEPI, as the most stringent environmental policy in China, can potentially correct the deviations in the implementation of local government environmental policies and affect the environmental governance strategies of enterprises. Thus, this research investigates the impact of CEPI implementation on the green cognition of executives, environmental protection investment, and green innovation efficiency from three dimensions: cognition, input, and output. This approach allows for a further investigation of the environmental behavior of enterprises under the pressure of CEPI.

5.1. Green Cognition of Executives

Cognitive theory emphasizes that executive cognition serves as the antecedent variable of enterprise behavior. Specifically, the subjective cognition of executives directly affects enterprise behavior and decision-making [54]. In response to external factors and stimuli, executives base their decisions on their comprehension and cognition, which then determines enterprise behavior choices [55,56].

As a signaling mechanism, CEPI underscores the commitment of the central government to protect the environment [38]. As CEPI continues to exert environmental pressure on enterprises, enterprise executives become increasingly aware of the importance of environmental issues for the sustainable development of their businesses, improving their green cognition. This heightened awareness can prompt enterprises to take proactive measures in environmental protection, enhancing the environmental performance of enterprises [53]. Specifically, green technological innovation exhibits uncertain characteristics, such as high investment and high risk [57]. Under external institutional pressure from CEPI, executives with a high level of green awareness recognize the importance of green development and the market opportunities it presents. They integrate environmental protection into enterprise strategic objectives and demonstrate a willingness to invest their limited resources and capabilities toward environmental innovation activities, such as environment-friendly technologies and equipment. These efforts ultimately promote green transformation.

To examine the mechanism of executive green cognition, we adopt a methodology based on the study by Duriau et al. (2007) [58]. Specifically, we select keywords such as “environmental protection strategy,” “work,” “governance,” and “education” to conduct a word frequency analysis on the annual reports of listed enterprises. We then take the logarithm of the resulting frequency count and add one to measure executive green cognition (Cognition). The dependent variable in the benchmark regression model is replaced by Cognition to assess the level of green focus in enterprise management decision-making. Column (1) of

Table 8. Regression results for the mechanism analysis.

Variables	(1)	(2)	(3)
	Cognition	Investment	Efficiency
CEPI	0.054 **	0.675 *	0.383 *
	(2.15)	(1.87)	(1.78)
Dual	−0.009	−0.198	0.106
	(−0.46)	(−0.75)	(0.47)
Top1	−0.026	1.448	−1.413
	(−0.14)	(1.36)	(−1.02)
Soe	−0.198 **	0.495	−0.915
	(−2.63)	(0.77)	(−1.53)
Ratio	0.113	1.164	0.329
	(1.18)	(1.40)	(0.49)
Roa	0.135	3.155 *	0.848
	(0.76)	(1.94)	(0.74)
Tobinq	−0.008	−0.177	0.158 **
	(−0.87)	(−1.65)	(2.51)
Cflow	0.124	−0.655	−0.238
	(1.04)	(−0.55)	(−0.20)
Age	0.005	0.380	−0.208
	(0.18)	(1.04)	(−0.69)
HHI B	−0.454	−2.046	−5.104
	(−1.23)	(−0.74)	(−1.64)
ProGDP	−0.060	0.611	−0.750
	(−0.83)	(0.97)	(−1.05)
_cons	1.044	7.626	11.667
	(1.40)	(1.14)	(1.54)
Firm FE	Y	Y	Y
Year × Ind FE	Y	Y	Y
N	5125	3390	5096
R2	0.627	0.547	0.668

Note: (1) We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. **, and * indicate significance at 5%, and 10% levels, respectively. (2) Green innovation efficiency (Efficiency) is measured by the ratio of green innovation output to innovation input. Specifically, this paper approximates green innovation input using the firm’s annual R&D expenditures. Green innovation output for listed companies is measured by the natural logarithm of the sum of the annual number of green invention patents, utility model patents, and design patents applied for by the listed company, plus one.

shows the regression results concerning the senior management perception of green initiatives. The regression results indicate that after CEPI implementation, Cognition is increased by 5.40% and is significant at the 5% level. These findings suggest that CEPI promotes enterprise green transformation by enhancing green awareness among senior executives and steering enterprises toward sustainable green development.

5.2. Environmental Protection Investment

CEPI emphasizes the shared responsibility between the party and the government to ensure consistency in the rights and responsibilities of local governments in environmental protection [22]. Thus, CEPI reconstructs the incentive structure, which can significantly increase the pressure of local governments on environmental governance, further fostering the commitment of enterprises to environmental governance [59]. In the short term, enterprises are compelled to acquire pollution prevention and control equipment and incur expenses associated with pollution treatment to swiftly comply with pollution reduction requirements [18]. Over time, stringent environmental supervision internalizes environmental costs, compelling enterprises to invest in green innovation technologies to improve production efficiency and bolster competitive advantages [60,61]. In summary, when confronted with severe regulatory pressure from CEPI, enterprises tend to increase investments in environmental protection to pursue sustainable development goals. This move prioritizes long-term benefits over short-term cost savings, achieving mutually beneficial results for environmental performance and business performance.

To examine the mechanism of environmental protection investment, we aggregate data on projects directly related to environmental protection—such as desulfurization projects, waste gas treatment, and cleaner production—from the detailed construction items that are in progress in the annual reports of listed enterprises in heavily polluting industries.

We then take the logarithm to reflect changes in environmental protection investment. Column (2) in Table 8 shows that after CEPI implementation, enterprise investment in environmental protection (Investment) increases significantly at the 10% level, emphasizing the role of CEPI in bolstering enterprise environmental protection investment.

5.3. Green Innovation Efficiency

CEPI significantly improves the intensity of local government environmental regulation through the authority of the central government and the shared responsibility between the party and the government. With intensified environmental regulation, the cost of pollution control and penalties for enterprises gradually exceed the research and development cost of green technology innovation [62]. To achieve environmental compliance, enterprises prioritize reducing dependence on polluting production methods by adjusting the input of production factors to improve the efficiency of green innovation [60]. Moreover, CEPI widely mobilizes the public to participate in environmental governance. This political mobilization is likely to persist beyond the end of the supervised action but may eliminate the institutional barriers and ensure permanent public supervision. Public participation raises environmental pressure and non-compliance risks for enterprises, improving their environmental protection initiatives and fostering green technological innovation [23]. Meanwhile, CEPI underscores the commitment of the central government to ecological civilization construction, reducing market risks for green innovation products and encouraging enterprises to actively engage in green research and development. This approach promotes continuity in green innovation activities and enhances green innovation efficiency [38]. In summary, under the dual pressure of the CEPI system and public participation, enterprises tend to pursue green technological innovation, improve green innovation efficiency, and ultimately realize the dual benefits of pollution reduction and productivity improvement.

To measure the input and output of green innovation technology, we adopt the method employed by Zhou et al. (2023) [63]. We use the green innovation output per unit of innovation input to quantify green innovation efficiency (Efficiency). Column (3) in Table 8 shows that after CEPI implementation, Efficiency increases significantly at the 10% level, increasing by 38.300%. These findings suggest that CEPI implementation promotes the green innovation efficiency of enterprises and supports the mechanism of green innovation efficiency.

6. Heterogeneity Analysis

Based on the preceding analysis, this study further examines the impact of enterprise, policy, and regional differences. It also conducts heterogeneity analysis from three perspectives: the government–enterprise relationship, CEPI “look back,” and environmental enforcement. This approach aids in determining the impact of vertical supervision on enterprise green transformation and delineating the boundaries of its synergistic effect with environmental self-discipline.

6.1. Government–Enterprise Relationship

Political connection is a significant method by which government may exert influence over enterprise behavior. It is also an important informal system and relationship resource within enterprises [64]. Thus, we designate the government–enterprise relationship (PC) as a virtual variable and perform grouping regression based on PC. When enterprise executives serve or have served as party representatives, NPC deputies, CPPCC members or government officials at all levels, PC is set to 1; otherwise, it is set to 0. The results show that after the impact of CEPI, enterprises without government–enterprise relationships exhibit a more noticeable decrease in pollution emission intensity, a larger increase in total factor productivity, and a more significant synergistic effect of vertical supervision and environmental self-discipline (see

Table 9. Regression results for the heterogeneity of government–enterprise relationships.

Variables	Government–Enterprise Relationship (PC = 1)				No Government–Enterprise Relationship (PC = 0)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP
CEPI	0.002	−0.012	−0.014	0.001	−0.121 ***	0.071 ***	−0.117 ***	0.071 **
	(0.04)	(−0.20)	(−0.25)	(0.01)	(−3.65)	(2.86)	(−3.51)	(2.70)
ISO14001			−0.087	0.077 *			0.010	−0.004
			(−1.60)	(2.02)			(0.26)	(−0.16)
CEPI × ISO 14001			−0.041	0.028			−0.124 ***	0.067 **
			(−0.63)	(0.48)			(−3.73)	(2.60)
Dual	0.011	0.004	0.014	0.001	0.018	−0.003	0.018	−0.003
	(0.15)	(0.07)	(0.19)	(0.03)	(0.31)	(−0.08)	(0.31)	(−0.08)
Top1	−0.480	−0.009	−0.491	0.003	−1.088 **	0.581 *	−1.086 **	0.581 *
	(−0.91)	(−0.03)	(−0.94)	(0.01)	(−2.22)	(1.98)	(−2.21)	(1.97)
Soe	0.599	−0.603	0.595	−0.599	0.149	−0.116	0.150	−0.116
	(1.22)	(−1.21)	(1.20)	(−1.20)	(1.13)	(−1.24)	(1.13)	(−1.24)
Ratio	−0.639 ***	0.372 **	−0.657 ***	0.388 **	−1.143 ***	0.574 ***	−1.143 ***	0.574 ***
	(−3.59)	(2.47)	(−3.66)	(2.55)	(−6.80)	(3.60)	(−6.80)	(3.60)
Roa	−2.382 ***	2.200 ***	−2.410 ***	2.222 ***	−2.409 ***	2.164 ***	−2.409 ***	2.163 ***
	(−7.26)	(9.55)	(−7.28)	(9.44)	(−9.73)	(8.56)	(−9.75)	(8.55)
Tobinq	0.005	0.013	0.002	0.016	0.081 ***	−0.019	0.081 ***	−0.019
	(0.21)	(0.68)	(0.07)	(0.87)	(4.25)	(−1.41)	(4.25)	(−1.41)
Cflow	−0.664 **	0.651 **	−0.680 **	0.668 **	−0.706 ***	0.444 ***	−0.704 ***	0.445 ***
	(−2.41)	(2.45)	(−2.47)	(2.51)	(−4.15)	(3.11)	(−4.16)	(3.12)
Age	−0.258 **	0.042	−0.256 **	0.038	−0.529 ***	0.315 ***	−0.529 ***	0.315 ***
	(−2.08)	(0.47)	(−2.09)	(0.44)	(−5.72)	(4.72)	(−5.72)	(4.73)
HHI B	−0.481	0.435	−0.498	0.450	−4.103 ***	3.259 ***	−4.101 ***	3.257 ***
	(−0.97)	(0.71)	(−1.01)	(0.73)	(−3.44)	(3.78)	(−3.44)	(3.78)
ProGDP	−0.374	0.165	−0.375	0.166	0.330 *	−0.213*	0.332 *	−0.213 *
	(−1.34)	(0.68)	(−1.37)	(0.71)	(1.93)	(−1.95)	(1.93)	(−1.95)
_cons	−2.768	6.348 **	−2.720	6.303 **	−8.588 ***	8.971 ***	−8.610 ***	8.975 ***
	(−0.93)	(2.44)	(−0.93)	(2.50)	(−4.79)	(7.99)	(−4.78)	(7.99)
Firm FE	Y	Y	Y	Y	Y	Y	Y	Y
Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y
N	1544	1505	1544	1505	3875	3703	3875	3703
R2	0.972	0.964	0.973	0.964	0.948	0.927	0.948	0.927

Note: We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.

).

When regulated enterprises are connected with the government, a close relationship is created, often resulting in a mutually beneficial community. In the context of economic competition, government officials may weaken environmental supervision, acquiesce with or even connive in the environmental pollution behavior of polluting enterprises, and guide politically connected enterprises to invest their funds in short-term profitable projects to reduce investment in environmental protection, thereby increasing local fiscal revenue [52]. Meanwhile, the government conceals the environmental pollution behavior of politically connected enterprises. Before inspectors enter, the local government will be aware of the arrival time and notify polluting enterprises within the jurisdiction that have a government–enterprise relationship to reduce production or even stop production. When faced with CEPI rectification requirements, enterprises can reduce violation penalties through close political connections. Consequently, the cost of pollution violations becomes lower than the cost of investing in environmental protection [65]. Owing to their profit-driven nature, enterprises opt to violate pollution regulations, thus impeding enterprise green transformation.

6.2. CEPI “Look Back”

The “look back” Environmental Protection inspection involves a random evaluation of whether the rectification targets issued by CEPI in various areas have been attained and whether the rectification tasks have been accomplished. It aims to ensure that local governments fulfill their rectification responsibilities. Thus, we perform grouping regression based on whether the place of business registration undergoes the “look back” process (see

Table 10. Regression results for the heterogeneity of CEPI “look back”.

Variables	CEPI “Look Back” (Look Back = 1)				CEPI “Look Back” (Look Back = 0)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP
CEPI	−0.081 *	0.047	−0.070	0.045	−0.104	0.074	−0.110	0.070
	(−1.93)	(1.53)	(−1.61)	(1.33)	(−1.69)	(1.60)	(−1.72)	(1.40)
ISO14001			−0.002	0.006			0.020	−0.024
			(−0.05)	(0.29)			(0.32)	(−0.56)
CEPI × ISO 14001			−0.113 ***	0.062 **			−0.073	0.058
			(−3.53)	(2.26)			(−0.89)	(0.95)
Dual	−0.059	0.050	−0.059	0.049	0.061	−0.007	0.063*	−0.009
	(−1.12)	(1.32)	(−1.12)	(1.32)	(1.80)	(−0.27)	(1.92)	(−0.32)
Top1	−0.925 **	0.396	−0.925 **	0.396	0.422	−0.484	0.425	−0.486
	(−2.14)	(1.41)	(−2.15)	(1.41)	(1.42)	(−1.65)	(1.42)	(−1.67)
Soe	0.102	−0.133	0.103	−0.133	0.075	−0.072	0.073	−0.071
	(0.81)	(−1.41)	(0.81)	(−1.41)	(0.67)	(−0.77)	(0.66)	(−0.75)
Ratio	−1.060 ***	0.520 ***	−1.061 ***	0.520 ***	−1.300 ***	0.803 ***	−1.302 ***	0.802 ***
	(−6.15)	(3.28)	(−6.15)	(3.28)	(−6.40)	(4.09)	(−6.37)	(4.03)
Roa	−2.436 ***	2.165 ***	−2.438 ***	2.166 ***	−2.546 ***	2.579 ***	−2.548 ***	2.570 ***
	(−10.11)	(11.04)	(−10.19)	(11.06)	(−8.73)	(6.30)	(−8.54)	(6.17)
Tobinq	0.090 ***	−0.027 *	0.089 ***	−0.027 *	0.072 *	−0.031	0.072 *	−0.031
	(4.41)	(−1.76)	(4.38)	(−1.76)	(1.92)	(−1.10)	(1.92)	(−1.11)
Cflow	−0.763 ***	0.462 ***	−0.760 ***	0.461 ***	−0.934 *	0.824 ***	−0.931 *	0.822 ***
	(−4.26)	(2.92)	(−4.24)	(2.91)	(−2.21)	(3.35)	(−2.19)	(3.32)
Age	−0.486 ***	0.247 ***	−0.484 ***	0.246 ***	−0.332 ***	0.137	−0.332 ***	0.138
	(−5.48)	(3.22)	(−5.51)	(3.22)	(−4.41)	(1.72)	(−4.28)	(1.70)
HHI B	−4.965 ***	3.943 ***	−4.976 ***	3.949 ***	−1.546 *	1.095	−1.556 *	1.096
	(−3.57)	(4.17)	(−3.59)	(4.20)	(−1.85)	(1.49)	(−1.86)	(1.49)
ProGDP	0.027	−0.070	0.030	−0.070	−0.341	0.278	−0.337	0.277
	(0.11)	(−0.33)	(0.12)	(−0.33)	(−1.64)	(1.77)	(−1.62)	(1.77)
_cons	−5.491 **	7.640 ***	−12.410 ***	7.645 ***	−2.948	4.907 **	−2.984	4.928 **
	(−2.13)	(3.58)	(−4.77)	(3.56)	(−1.35)	(2.92)	(−1.37)	(2.95)
Firm FE	Y	Y	Y	Y	Y	Y	Y	Y
Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y
N	3738	3583	3738	3583	1752	1686	1752	1686
R2	0.941	0.920	0.941	0.920	0.965	0.948	0.965	0.948

Note: We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.

). The findings show that the regression coefficient of vertical supervision CEPI on the Emission intensity of enterprise pollution is significantly negative at the 10% level; meanwhile, the regression coefficient to total factor productivity Tfp is not significant. One possible reason is that certain areas still encounter false rectification problems within the environmental protection supervision system. The implementation of the “looking back” mechanism has imposed persistent supervision pressure on local governments, which compels them to rectify environmental violations, thus reducing enterprise pollution emissions. However, stringent environmental law enforcement pressures enterprises to rectify concerns immediately, leaving no buffer period. Enterprises may prioritize pollution control methods with rapid results and require low-cost short-term investment, diverting funds away from green technology innovation. Such an approach is not conducive to improving the production efficiency of enterprises.

Columns (3)–(4) in Table 10 indicate that the regression coefficient of CEPI × ISO14001 has a significant negative regression coefficient on Emission at the 1% level and a significant positive regression coefficient on Tfp at the 5% level. These findings suggest the inadequacy of external regulatory pressure for enterprise green transformation and the need for proactive change among enterprises. Environmental self-discipline within enterprises emphasizes a strategic change in their environmental protection concept, which is embodied in the enterprise environmental management system and green production. Under the scrutiny of the “look back” mechanism by CEPI, enterprises implementing the ISO 14001 standard typically demonstrate organizational capabilities in environmental protection practices such as pollution prevention and environmental product design [66]. In addition, the implementation of the ISO 14001 standard by enterprises improves the efficiency of raw materials and energy usage. This enhancement leads to increased production efficiency, promoting enterprise green transformation [67].

6.3. Environmental Enforcement

Environmental enforcement (Enforce) is an important factor influencing the effectiveness of environmental regulation, and its intensity varies from region to region. To assess whether these differences can reflect the impact of vertical supervision and environmental self-discipline on enterprise green transformation, we classify provinces based on the median number of environmental administrative punishment cases into two groups: regions with strict environmental enforcement (Enforce = 1) and those with weak environmental enforcement (Enforce = 0).

The results show that in regions with high environmental enforcement, vertical supervision and its synergistic effect with environmental self-discipline exert a pronounced impact on enterprise green transformation (see

Table 11. Regression results for the heterogeneity of environmental enforcement.

Variables	Strict Environmental Enforcement (Enforce = 1)				Strict Environmental Enforcement (Enforce = 1)
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Emission	TFP	Emission	TFP	Emission	TFP	Emission	TFP
CEPI	−0.108 ***	0.064 **	−0.101 ***	0.063 *	−0.053	0.030	−0.041	0.021
	(−3.46)	(2.23)	(−3.04)	(1.93)	(−0.94)	(0.59)	(−0.73)	(0.40)
ISO14001			−0.015	0.006			0.073	−0.046
			(−0.28)	(0.17)			(1.37)	(−1.10)
CEPI × ISO 14001			−0.140 ***	0.072 **			0.002	−0.004
			(−3.06)	(2.18)			(0.03)	(−0.06)
Dual	0.013	−0.003	0.013	−0.003	−0.050	0.048	−0.051	0.048
	(0.24)	(−0.10)	(0.24)	(−0.10)	(−0.53)	(0.59)	(−0.53)	(0.59)
Top1	−0.559	0.277	−0.558	0.278	−1.100	0.193	−1.096	0.201
	(−1.40)	(1.06)	(−1.40)	(1.07)	(−1.18)	(0.33)	(−1.18)	(0.35)
Soe	0.094	−0.137	0.096	−0.137	0.012	−0.034	0.005	−0.029
	(0.81)	(−1.59)	(0.83)	(−1.61)	(0.08)	(−0.33)	(0.03)	(−0.28)
Ratio	−1.080 ***	0.524 ***	−1.081 ***	0.524 ***	−1.103 ***	0.678 ***	−1.099 ***	0.674 ***
	(−5.22)	(2.90)	(−5.22)	(2.90)	(−4.57)	(3.03)	(−4.59)	(2.99)
Roa	−2.470 ***	2.218 ***	−2.469 ***	2.218 ***	−2.143 ***	2.066 ***	−2.132 ***	2.057 ***
	(−9.00)	(8.39)	(−9.03)	(8.37)	(−8.99)	(7.78)	(−8.89)	(7.68)
Tobinq	0.076 ***	−0.025	0.076 ***	−0.024	0.091 **	−0.039	0.093 **	−0.041
	(3.06)	(−1.40)	(3.01)	(−1.39)	(2.19)	(−1.28)	(2.23)	(−1.33)
Cflow	−0.720 ***	0.357 **	−0.719 ***	0.357 **	−0.692 **	0.748 ***	−0.696 ***	0.748 ***
	(−3.21)	(2.10)	(−3.22)	(2.09)	(−2.88)	(4.21)	(−2.92)	(4.26)
Age	−0.573 ***	0.377 ***	−0.569 ***	0.376 ***	−0.182	−0.038	−0.195	−0.031
	(−5.57)	(4.62)	(−5.68)	(4.68)	(−1.44)	(−0.34)	(−1.55)	(−0.27)
HHI B	−3.662 ***	3.053 ***	−3.659 ***	3.052 ***	−1.799	1.712	−1.755	1.681
	(−3.42)	(3.20)	(−3.41)	(3.21)	(−0.84)	(1.33)	(−0.84)	(1.34)
ProGDP	−0.040	−0.092	−0.040	−0.092	0.076	0.103	0.078	0.102
	(−0.25)	(−0.59)	(−0.24)	(−0.59)	(0.19)	(0.48)	(0.20)	(0.48)
_cons	−11.786 ***	7.763 ***	−11.793 ***	7.763 ***	−13.952 ***	6.840 ***	−13.972 ***	6.847 ***
	(−6.96)	(4.79)	(−6.90)	(4.78)	(−3.49)	(3.17)	(−3.50)	(3.18)
Firm FE	Y	Y	Y	Y	Y	Y	Y	Y
Year × Ind FE	Y	Y	Y	Y	Y	Y	Y	Y
N	3686	3522	3686	3522	1634	1581	1634	1581
R2	0.954	0.931	0.954	0.931	0.962	0.955	0.963	0.955

Note: We report t-statistics in parentheses and cluster robust standard errors at the province level in this table. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.

). This impact is attributed to strict environmental enforcement being a necessary condition for the effective implementation of environmental policies [68]. Under the institutional pressure of CEPI, local governments become aware of the central environmental protection goals and policy directives, leading to an increase in penalties for environmental pollution. Strict regulatory pressure on adherence to environmental regulation “forces” enterprise green transformation [69]. Moreover, environmental self-discipline also plays a synergistic role. In regions with rigid environmental enforcement, enterprises exhibiting environmental self-discipline may have a heightened awareness of environmental protection and adopt proactive measures to drive green transformation efforts [48].

7. Discussion

7.1. Main Findings

A “multiple collaborative governance” mechanism involving the government as the leader, enterprises as the primary actors, and the public as participants must be established. In this context, the impact and transmission mechanism of environmental protection inspections on enterprise green transformation need to be investigated within the collaborative framework of vertical supervision and enterprise environmental self-discipline. This research considers CEPI as an exogenous policy shock and regards heavily polluting enterprises in cities under supervision as the research object to assess the impact of vertical supervision and its coordination with environmental self-discipline on enterprise green transformation and subsequently identifies its internal mechanism. This study reveals that environmental vertical supervision significantly promotes the green transformation of heavily polluting enterprises, and a synergistic effect exists between vertical supervision and environmental self-discipline. The mechanism analysis indicates that vertical supervision can drive enterprise green transformation by improving executive green cognition, increasing environmental protection investment, and improving green innovation efficiency. Heterogeneous characteristics are observed in certain dimensions, including government–enterprise relationships, CEPI “look back” initiatives, and the intensity of environmental law enforcement. These factors influence the synergistic effects of vertical supervision and environmental self-discipline.

7.2. Theoretical Contributions

This study makes three key theoretical contributions to the environmental governance literature: First, this study breaks away from prior isolated research on either government regulation or corporate self-discipline, empirically revealing for the first time a significant synergistic effect between external vertical supervision and internal environmental self-discipline [4,10]. Contrary to substitutionary perspectives, environmental self-discipline functions as an amplifying mechanism that enhances regulatory efficacy [70]. It provides strong support for institutional theory propositions on how coercive pressures activate and enhance internal normative mechanisms and operationalizes the concept of “multi-stakeholder collaborative governance” in corporate green transformation.

Second, the research elucidates the mediating pathways through which regulatory pressure translates into green transformation outcomes. By validating executive green cognition, environmental protection investment, and green innovation efficiency as critical transmission mechanisms, it addresses a fundamental gap in prior work that emphasize regulation–performance correlations while obscuring intermediate processes [17,20]. The integration of cognitive theory, resource-based logic, and innovation dynamics provides a granular framework for understanding how institutional pressures catalyze sustainable strategic adaptation.

Third, this study significantly enriches institutional and political economy perspectives by demonstrating that the effectiveness of synergistic governance is constrained by key boundary conditions. The research finds that strong government business ties weaken the effect, while the synergistic effect is enhanced during CEPI “look back” inspections, and local environmental enforcement intensity is also crucial. This indicates that the effectiveness of synergistic governance highly depends on the firm’s institutional environment.

7.3. Policy and Managerial Implications

Based on the aforementioned conclusions, this study presents several implications. First, for governmental bodies, we recommend integrating top-down enforcement with bottom-up corporate self-discipline through incentive-aligned regulatory instruments. These should include tax relief, expedited permitting, and procurement preferences for ISO 14001-certified firms, coupled with central–local coordination mechanisms that embed environmental KPIs in officials’ performance evaluations. Concurrently, regulatory resources should prioritize corruption-vulnerable sectors where government–enterprise collusion subverts environmental goals while institutionalizing “look back” audits to combat formalism and ensure policy fidelity.

Then, corporate leaders must reconceptualize environmental management as strategic infrastructure rather than compliance overhead. The proactive adoption of certified environmental management systems builds organizational resilience and reputational capital, particularly when coupled with executive compensation structures tied to environmental KPIs and early investments in green innovation that leverage regulatory synergies for competitive advantage.

Finally, governments and firms should co-construct third-party accountability infrastructures featuring public environmental disclosure platforms that transparently report certifications, violations, and emissions data. Such systems empower stakeholders to differentially reward sustainable performers and sanction laggards while enabling firms to credibly signal environmental legitimacy.

8. Conclusions

This study, while rooted in the unique political context of China, provides valuable theoretical contributions that are highly relevant to environmental governance in other developing economies grappling with similar institutional challenges. Economic decentralization coupled with strong local protectionism represents a common challenge across emerging economies such as India, Brazil, and Mexico [71,72]. In these contexts, subnational governments frequently prioritize economic growth objectives over stringent environmental enforcement. Our findings demonstrate that vertical supervision mechanisms, exemplified by China’s CEPI, offer a viable strategy to counteract collusion between local regulators and polluting enterprises. Such mechanisms achieve this by imposing credible top-down accountability, facilitating bottom-up public scrutiny through transparent complaint systems, and enforcing meaningful political sanctions for non-compliance, thereby disrupting entrenched local interests.

Furthermore, achieving sustainable development in developing nations necessitates a hybrid governance model integrating mandatory oversight with voluntary corporate standards. Vertical supervision provides the essential foundation for regulatory compliance, yet its effectiveness is significantly amplified when combined with market-based incentives and robust public participation. This integrated approach fosters true co-governance. Ultimately, a resilient environmental governance system requires the collective agency of governments, enterprises, and citizens. Such a multi-dimensional framework transcends the limitations of purely top-down regulation and embeds sustainability principles within both corporate strategy and societal norms.

This study recognizes its limitations, which can be potentially addressed in future research. First, in addition to CEPI and ISO 14001 certification, other types of regulation in vertical environmental regulation and enterprise environmental self-regulation exist. These other types and their impact on enterprise green transformation may be investigated in future research to verify the universality of the current conclusion. Second, other techniques to measure enterprise green transformation enterprises have been identified. In future studies, the measurement system for enterprise green transformation could incorporate factors such as environmental training for employees and cultivating green awareness. Third, the impact mechanism examined in this study may not provide a sufficiently comprehensive view. Subsequent research could start with the dynamic game strategy among the central government, local government, and enterprises within the context of system pressure to analyze other potential mechanisms.