Does an Environmental Protection Tax Promote or Inhibit the Market Value of Companies? Evidence from Chinese Polluting Companies

Abstract

This study takes the environmental protection tax (EPT) implemented in China in 2018 as the policy background and systematically examines the impact mechanism and boundary conditions of EPT on the market value of listed companies in the polluting industries. The results indicate that EPT significantly inhibits Tobin’s Q of polluting companies. A one-unit increase in EPT leads to a 0.274-unit decrease in Tobin’s Q. The heterogeneity test reveals that the EPT shock exhibits a spatial gradient effect of “Eastern > Central > Western > Northeastern”. The rigidity of the tax system is stronger than that of the pollution discharge fee, and the effect on non-heavily polluting industries is stronger than that on heavily polluting industries. Mechanism analysis shows that while corporate financial flexibility can buffer against short-term EPT shocks, R&D investment and patent quality expose an “innovation trap” characterized by high investment but low conversion efficiency, largely determined by the type of innovation pursued. By elucidating the multiple moderating and mediating mechanisms at play, this study constructs an integrated “institutional pressure-resource constraints-market feedback” model, thereby providing a new analytical framework for environmental economics in emerging markets.

1. Introduction

Confronted with the global climate crisis and escalating pollutant emissions, the international community is increasingly emphasizing the critical importance of sustainable development for businesses [1,2,3,4]. Consequently, the impact of environmental regulation policies on economic entities has emerged as a cutting-edge topic of shared interest among academics and practitioners worldwide [5,6,7,8]. International studies generally suggest that environmental protection tax (EPT) policies can achieve a win-win outcome for both environmental and economic performance through the mechanism outlined in the “Porter Hypothesis” [9,10,11,12,13,14]. However, the specific effects are moderated by multiple factors, including industry characteristics, policy design, and market maturity. Notably, capital markets are growing increasingly sensitive to how environmental risks are priced into company valuations [15,16,17,18,19], making the correlation between corporate environmental performance and market value a significant research stream in corporate finance [20,21].

China’s environmental governance model exhibits distinct institutional features. The formal implementation of the Environmental Protection Tax Law in 2018 signified a major shift in the nation’s regulatory approach, moving from administrative commands toward market-based economic incentives and constraints. While current research in China has predominantly focused on the emission reduction effects of the EPT [22,23,24], there is a lack of systematic investigation into its micro-level transmission mechanisms—particularly its impact pathways and boundary conditions concerning corporate market value [25,26,27]. This research gap is especially pronounced against the backdrop of deepening the registration system reforms for polluting firms and improving the pricing efficiency in capital markets. There is thus a pressing need for in-depth studies that are firmly grounded in China’s unique institutional context.

Against this backdrop, this study aims to address the following core research question: Does China’s EPT promote or inhibit the market value of polluting companies? Furthermore, what are the underlying mechanisms and boundary conditions that govern this relationship? To answer these questions, the study sets forth three primary objectives: (1) to empirically test the direct impact of EPT intensity on corporate market value (Tobin’s Q); (2) to explore the moderating effects of company-specific factors on the EPT-value nexus; and (3) to examine the specific mediating pathways through which the EPT affects polluting companies’ market value.

The study is necessitated by both practical policy needs in EPT implementation and theoretical development. Its theoretical significance is threefold: first, by constructing a theoretical framework linking EPT to corporate value, it bridges the disciplinary divide between environmental economics and corporate finance, deepening the understanding of corporate value reconstruction under institutional pressure [28,29]. Second, by incorporating the context of China’s transitional economy, it expands the theoretical boundaries of environmental regulation research in emerging markets [30,31]. Finally, it provides providing an Eastern perspective for global sustainability studies. The primary innovation lies in its novel empirical approach: it pioneers the direct investigation of EPT’s impact on the specific corporate value by constructing a “EPT intensity-corporate behavior-market value” model. By introducing key moderating (e.g., debt-to-asset ratio, profit margin, R&D investment) and mediating variables (patent quality), it reveals the effect’s boundary conditions and transmission mechanisms, thereby filling a theoretical gap in the micro-mechanisms of environmental regulation in emerging markets and providing Chinese empirical evidence for policy design.

This study begins with an introduction to the EPT policy and the research agenda. It then moves to a literature review and theoretical framework to establish the foundation for the analysis. The subsequent section explains the research design and data, followed by the presentation of empirical results. The paper then discusses the implications of these findings and, finally, concludes by summarizing the key insights.

2. Literature Review and Theoretical Basis

2.1. EPT Evolution Process

Porter & Linde [32] proposed the Porter hypothesis, arguing that environmental regulation can force companies to innovate and enhance competitiveness, and its theoretical framework laid the foundation for subsequent research. For example, Chenghao et al. [33] revealed that fiscal growth significantly positively drives EPT revenue, verifying the positive fiscal impact of EPT reform. Chenghao & Mayburov [34] empirically proved that China’s EPT has generated economic dividends. Lin et al. [35] revealed that corporate environmental regulatory pressure promotes supplier green innovation through supply chain transmission and promotes external cost transfer internally. Liu et al. [36] found that the dual pilot policy synergy significantly promotes green technology innovation and corrects technology positive externalities and pollution negative externalities through mechanisms such as innovation subsidies and industrial transformation. Gyamfi et al. [37] found that green finance, EPT and innovative technologies significantly reduce carbon emissions in E7 countries, while Kinyar & Bothongo [38] found that EPT is more effective in reducing emissions than environmental protection spending.

There are also some negative studies. For example, Pan & Zhang [39] concluded that EPT inhibits R&D investment of heavily polluting companies, challenging the traditional incentive hypothesis. Yan et al. [40] revealed that China’s environmental protection tax inhibits the green productivity of companies in the short term, challenging the Porter hypothesis. Erdogan [41] revealed that Germany’s EPT is not a green fiscal tool. Wang et al. [42] found that China’s environmental fee-to-tax reform significantly inhibits the total factor productivity of companies and other significant negative evidence.

Some studies yield mixed conclusions. For instance, Qian et al. [43] believes that China’s levy of VOCs environmental protection tax effectively reduces emissions and synergistically reduces multiple pollutants, but impacts the macroeconomy.

To intuitively display the relevant research on EPT, this study conducted a bibliometric analysis and counted the contribution of EPT as a keyword. Figure 1 was generated from a search of the Web of Science Core Collection using the terms “China” and “EPT”, and the years were from 2012 to 2024. There was a total of 717 documents. Then VOSviewer 1.6.20 was used to generate a network diagram of all keyword contributions. Figure 2 is its kernel density distribution diagram.

Figure 1 and Figure 2 reveal the core keywords occupying central positions, such as “policy”, “environmental protection”, and “tax”, indicating their high frequency and density in the literature, which signifies prominent research hotspots. The term “policy” is closely linked with “government” and “management”, underscoring the significance of policy formulation and governance in this field. Similarly, “environmental protection” is associated with “pollution” and “emissions”, reflecting a research focus on the implementation and impacts of protective measures. The lines, color-coded by year (with blue representing 2019 and orange 2022, as shown in the legend), illustrate a temporal evolution. Notably, attention to “green innovation” or related concepts has intensified in recent years. This trend suggests that, alongside growing environmental awareness and technological advancement, researchers are increasingly concentrating on how technological innovation can facilitate the achievement of green and sustainable development goals.

2.2. The Relationship Between the Evolution of Corporate Value and Taxation

Tobin’s Q ratio was first proposed by Nicholas Kaldor in 1966 [44] and later promoted by James Tobin in 1976 [45]. This ratio measures the ratio of the market value of physical assets to the replacement cost, reflecting the relationship between the market valuation of corporate assets and the cost of reproduction. The development of this theory provides profound insights into understanding asset pricing, investment decisions, and economic growth.

The study of corporate value was first introduced to international research by Min & Prather [46], and revealed that the announcement of international joint ventures generally produces a positive stock price effect, especially for polluting companies with high Tobin Q values and low free cash flow. Thoma [47] constructed a comprehensive value index of trademark indicators, it is revealed that its positive correlation with Tobin Q is significantly higher than that of traditional intangible assets such as patents and R&D, confirming the key role of trademark portfolio strategy in enhancing the market value of companies, and providing a new dimension for IP management.

In recent years, research has paid more attention to heterogeneity: Jia et al. [48] revealed that the EU carbon border adjustment mechanism significantly impacted the stock prices of high-pollution and indirect emission companies in China. Senna & de Araújo Moxotó [49] revealed that, according to the Brazilian carbon efficiency index, high-carbon emission companies have better financial performance and can improve value performance.

In addition, Yuan et al. [50] confirmed that China’s carbon trading pilot policy effectively alleviates information asymmetry in the stock market and effectively corrects corporate value judgments by improving carbon information disclosure, releasing carbon market signals and strengthening government participation. Zhou et al. [51] found that under the carbon tax policy, the brand extension strategy forces regulatory price adjustment through social impact regulation, balances market expansion and cannibalization effects, and improves environmental benefits while damaging corporate profits.

The association mechanism between taxation and corporate value has been a research focus in recent years. Empirical studies based on Tobin’s Q show divergence: some studies support a positive effect. For example, Hasan et al. [52] found that high organizational capital companies increase corporate value through tax avoidance, and shareholders regard it as an efficiency-enhancing strategy. This effect is particularly pronounced in polluting companies with good governance and strict financial constraints, tax avoidance increases future cash flow and tax haven investment, verifying the role of organizational capital investment value in promoting tax efficiency. Tang et al. [53] found that the development of financial technology increases corporate value by promoting corporate tax avoidance, reduces tax risks and enhances the efficiency of tax incentives. Ran & Ji [54] revealed that China’s Environmental Protection Tax Law enhances corporate reputation by promoting green investment, verifying the effectiveness of the EPT policy. Lv et al. [55] revealed that corporate tax avoidance damages corporate value.

On the contrary, some studies support negative effects. For instance, Jing et al. [56] revealed that China’s environmental tax law exacerbates tax avoidance by polluting companies in high-tax areas, verifying the unintended consequence of tax avoidance to enhance corporate value, indicating that EPT has a negative impact on corporate value.

There are also actual situations where positive and negative effects exist. For example, Ostad & Mella [57] studied that corporate tax information is more value-relevant to investors during the right-leaning government, and it is not significant during the left-leaning period, and verified for the first time that party political orientation significantly regulates the information content of tax policies.

Research on the relationship between taxation and corporate value has been developing continuously, but there is a lack of research on the impact of EPT on polluting corporate value. There is a major research gap in this regard. The reason is that environmental regulation is also a research hotspot that has developed in recent years. Especially for China’s newly released EPT in 2018, it is difficult to establish a timely research stream on its link with polluting corporate value.

However, this study innovatively establishes a micro-study on the EPT’s impact on market value pricing. Similarly, to more intuitively display the relevant research on the relationship between EPT and corporate value, a bibliometric analysis was conducted, and the contribution of keywords such as EPT and Tobin’s Q was counted. The search was performed through the Web of Science Core collection, and there were two search terms, namely “EPT” and “Tobin’s Q”. The search results showed that there were no documents, so the scope was enlarged, and the search terms were “Tax” and “Tobin’s Q”. The time was also from 2012 to 2024, and 49 documents were retrieved and screened. Finally, VOSviewer 1.6.20 was used to generate a word cloud diagram of all keyword contributions, as shown in Figure 3, and Figure 4 is its kernel density distribution diagram.

Figure 3 and Figure 4 present the bibliometric analysis of the relationship between tax and corporate value (proxied by Tobin’s Q). The centrality of nodes such as “firm performance”, “Tobin’s Q”, and “agency costs” indicates that a significant body of research focuses on corporate financial performance. The temporal evolution, represented by line colors from 2014 to 2024, reveals a growing scholarly interest in the connections between Tobin’s Q and concepts like taxes, policy, and “financial constraints”, marking them as emerging research trends. Figure 4, the kernel density distribution, further corroborates that keywords like “firm performance”, “Tobin’s Q”, and “corporate governance” occupy high-density areas, confirming them as core research foci. In contrast, keywords such as “tax” and “policy” appear relatively isolated in the network or reside in low-density areas of the kernel density map. This pattern highlights a promising direction for future research, particularly concerning the impact of the EPT on corporate value, and underscores a significant gap in the existing literature.

2.3. Theoretical Basis and Research Hypothesis

Based on the neoclassical cost theory [58], which holds that EPT directly increases corporate operating expenses by internalizing pollution costs, compresses profit margins and triggers investors to reassess risk, we hypothesize the following, considering that China’s EPT policy is mainly based on “volume-based taxation” and companies cannot immediately offset this cost. In the short term, it is difficult for companies to fully absorb the tax burden through price transfer or efficiency improvement, resulting in market valuation discounts, so we hypothesize the following:

H1. 

EPT has a negative effect on Tobin’s Q.

According to the institutional pressure theory [59], highly marketized regions have stronger environmental law enforcement and need to strengthen government supervision, public participation and corporate source control to improve corporate carbon emission reduction performance [60]. However, investors are also more sensitive to the pricing of environmental risks [61], resulting in polluting companies facing the dual pressure of “strong regulation-high discount”. In contrast, in less marketized regions, policy implementation flexibility may buffer tax shocks through informal institutions (such as local protection). Thus, we hypothesize:

H2. 

The negative effect of EPT on Tobin’s Q is stronger in the eastern region (high marketization level) than in other regions.

Based on the perspective of finance, profitability and innovation economic development theory [62,63,64], we consider several moderating factors. The debt-to-asset ratio reflects capital structure and financial risk, which affects investment and market value. The operating profit margin indicates profitability and competitiveness, influencing a firm’s ability to withstand external shocks. R&D investment is linked to innovation capability and long-term potential, affecting market performance and sustainable development. Therefore, we propose:

H3a. 

The debt-to-asset ratio has a moderating effect between EPT and company value.

H3b. 

The operating profit margin has a moderating effect between EPT and company value.

H3c. 

The amount of R&D investment has a moderating effect between EPT and company value.

The traditional Porter hypothesis emphasizes that environmental regulation encourages green technology innovation, thereby improving the core competitiveness and sustainable development of companies [65]. Therefore, we propose the hypothesis:

H4. 

Technological innovation is the mediating factor between EPT and Tobin’s Q, and promoting innovation through EPT will promote the Tobin’s Q of companies.

3. Data and Methodology

The sample consists of listed companies in polluting industries that reported pollution discharge fees or EPT in their financial statements. The data are from 2018 to 2022. With reference to studies by Zhou et al. [66] and Zor [67], this study categorizes the sample into heavily polluting and non-heavily polluting industries based on the latest industry classification codes issued by the China Securities Regulatory Commission. Some missing values are interpolated using regression prediction models. The variables come from the PPMANDATA and CSMAR databases, and the invention patent data comes from the State Intellectual Property Office of China. At the same time, companies classified as (ST), (PT) and (*ST) are excluded. Finally, sample data of 872 listed companies are obtained, with a total of 9592 observations.

3.1. Variable Definition

3.1.1. Explained Variable

Tobin’s Q value is defined as the ratio of a company’s market value to its replacement cost. If a company’s Tobin’s Q value is greater than 1, then the company’s market value exceeds its replacement cost, indicating that investors are confident in the company’s prospects, which may mean that the company has better production capacity, a more competitive market position or a higher profit margin. Conversely, if a company’s Tobin’s Q value is less than 1, then the company’s market value is lower than its replacement cost, which may mean that the company is undervalued or facing operational challenges.

3.1.2. Core Explanatory Variables and Other Variables

EPT replaced the prior pollution discharge fee system. To facilitate heterogeneity analysis in later sections, we treat the pollution discharge fee data from 2012 to 2017 as a proxy for the EPT in the period before its implementation in 2018. This data is merged with post-2018 data to form a consolidated dataset spanning 2012–2022. Following the established practice of Rao et al. [68], Shen & Zhang [69], and Zhou et al. [70] in variable selection, our study builds upon this foundation by incorporating a comprehensive set of carefully chosen control variables. These variables are designed to fully capture the operational and financial characteristics of the listed companies. All variables, along with their definitions, are presented in

Table 1. Variable statistics and explanations.

Variable Types	Variable (Symbol)	Explanation and Calculation Methods (Original Unit)	Data Source
Explained variable	Tobin’s Q	Calculation formula: (market value of tradable shares + net asset value per share of non-tradable shares + book value of liabilities)/book value of assets (100 million yuan)	PPMANDATA
	OGM	Operating income minus operating costs divided by operating income (100 million yuan)
Explanatory variables	EPT	Natural logarithm transformation of annual actual environmental protection tax payment (million yuan)
	EPTP	Natural logarithm transformation of EPT payable (million yuan)
	TP	Natural logarithm transformation of taxes payable (million yuan)
	TS	Natural logarithm transformation of taxes and surcharges (million yuan)
Control variable	Age	The natural logarithm of the listing year plus 1 (1)	CSMAR
	MS	The stock price is multiplied by the natural logarithm transformation of the total number of shares to represent the market size (100 million yuan)	PPMANDATA
	NE	The natural logarithm of the number of employees in the company
	FL	Total debt to shareholders’ equity ratio (100 million yuan)	CSMAR
	ROA	Ratio of net profit to total assets (100 million yuan)
	NIA	Initial cost of intangible assets—accumulated amortization—impairment provision, then natural logarithm transformation (100 million yuan)	PPMANDATA
	FE	Natural logarithm transformation of financial expenses (million yuan)
	RTSHS	The ratio of the total number of shares held by the top 10 shareholders to the total share capital (1)
	PL	The chemical oxygen demand and ammonia nitrogen emissions in industrial wastewater, sulfur dioxide and nitrogen oxide emissions in industrial waste gas are standardized and converted into a unified pollution equivalent number, and the total is added with 1 to take the logarithm, which represents pollution level (pollution equivalent value)
Control and Moderating variable	DAR	Total liabilities to total assets ratio (million yuan)
Moderating variable	OPM	Ratio of operating profit to operating income (100 million yuan)
	R&D	The natural logarithm of the amount of all R&D investment (1)
Mediating variable	PAQ	The calculation method will be introduced in detail in the subsequent mediation effect model analysis
	QAP	The calculation method will be introduced in detail in the subsequent mediation effect model analysis

Source: developed by the authors.

, while

Table 2. Descriptive Statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
Tobin’s Q	9592	1.860	1.156	0.615	21.296
EPT	9592	14.226	1.556	7.921	20.892
Age	9592	2.460	0.589	0.000	3.434
MS	9592	23.246	1.235	20.24	28.726
NE	9592	8.307	1.217	2.890	13.253
FL	9592	1.534	3.444	−7.646	270.994
ROA	9592	0.035	0.067	−1.856	0.759
NIA	9592	19.370	1.584	11.640	26.314
FE	9592	17.970	1.625	9.011	24.049
RTSHS	9592	55.668	14.526	8.780	98.585
PL	9592	0.144	0.004	0.134	0.153
DAR	9592	0.455	0.191	0.011	1.303

Source: calculated by the authors.

provides the descriptive statistics.

3.2. Descriptive Statistics

Descriptive statistics for the key variables from the dataset of 9592 observations are presented in Table 2. Tobin’s Q has a mean of 1.860 and a standard deviation of 1.156, suggesting moderate market valuation relative to asset value with considerable cross-sectional variation. The Environmental Protection Tax (EPT) exhibits a mean of 14.226 and a standard deviation of 1.556, pointing to significant disparities in the environmental regulatory pressure faced by different companies.

While descriptive statistics provide a preliminary overview, they do not reveal distributional trends. Therefore, we plot the distribution trend of the data. Figure 5 presents the trend in Tobin’s Q of polluting listed companies over the period 2012–2022.

Figure 5 depicts the temporal distribution of Tobin’s Q for polluting companies from 2012 to 2022. The X-axis represents the year, the Y-axis indicates the sample size of polluting companies in this study, and the Z-axis denotes the values of Tobin’s Q. Overall, Tobin’s Q exhibits a fluctuating but generally upward trend, with a pronounced decline following the implementation of the 2018 EPT policy. This trajectory reflects the initial impact of the EPT policy on the Tobin’s Q of polluting companies and their subsequent adaptation process. The observed trend provides preliminary support for the documented negative impact of the 2018 EPT policy on Tobin’s Q in this study.

3.3. Variable Correlation and VIF Test

Significant correlations were found among the variables in this study. An examination of the Variance Inflation Factor (VIF) revealed that all values were below the threshold of 5, with the exception of EPT (6.219) and MS (5.514), thus establishing that multicollinearity is not a concern and validating the foundation for the ensuing empirical tests.

4. Empirical Results

4.1. Preliminary OLS Regression

The multivariate linear regression model is constructed for OLS regression, and control variables are gradually added. The regression results are shown in

Table 3. OLS regression results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)
VARIABLES	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q
EPT	−0.233 ***	−0.241 ***	−0.780 ***	−0.670 ***	−0.669 ***	−0.682 ***	−0.632 ***	−0.578 ***	−0.573 ***	−0.573 ***	−0.539 ***
	(0.007)	(0.008)	(0.013)	(0.015)	(0.015)	(0.014)	(0.014)	(0.013)	(0.014)	(0.013)	(0.014)
Age		0.057 ***	−0.063 ***	−0.070 ***	−0.068 ***	−0.016	−0.033 *	−0.068 ***	−0.096 ***	−0.022	0.005
		(0.021)	(0.019)	(0.018)	(0.018)	(0.018)	(0.017)	(0.017)	(0.018)	(0.019)	(0.020)
MS			0.808 ***	0.877 ***	0.876 ***	0.840 ***	1.020 ***	1.112 ***	1.118 ***	1.140 ***	1.136 ***
			(0.016)	(0.017)	(0.017)	(0.016)	(0.017)	(0.017)	(0.017)	(0.017)	(0.017)
NE				−0.236 ***	−0.235 ***	−0.218 ***	−0.103 ***	−0.112 ***	−0.110 ***	−0.133 ***	−0.135 ***
				(0.016)	(0.016)	(0.015)	(0.015)	(0.015)	(0.015)	(0.015)	(0.015)
FL					−0.015 ***	−0.009 ***	−0.007 ***	−0.000	−0.001	−0.001	0.001
					(0.003)	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)
ROA						3.775 ***	3.133 ***	2.529 ***	2.569 ***	2.500 ***	1.979 ***
						(0.146)	(0.141)	(0.138)	(0.138)	(0.138)	(0.151)
NIA							−0.316 ***	−0.288 ***	−0.287 ***	−0.281 ***	−0.285 ***
							(0.010)	(0.010)	(0.010)	(0.010)	(0.010)
FE								−0.190 ***	−0.190 ***	−0.195 ***	−0.183 ***
								(0.007)	(0.007)	(0.007)	(0.008)
RTSHS									−0.003 ***	−0.003 ***	−0.004 ***
									(0.001)	(0.001)	(0.001)
PL										−23.959 ***	−26.819 ***
										(2.651)	(2.663)
DAR											−0.527 ***
											(0.062)
Constant	5.177 ***	5.153 ***	−5.669 ***	−6.861 ***	−6.842 ***	−6.235 ***	−5.912 ***	−5.755 ***	−5.766 ***	−2.839 ***	−2.710 ***
	(0.103)	(0.103)	(0.239)	(0.250)	(0.249)	(0.242)	(0.231)	(0.224)	(0.223)	(0.393)	(0.392)
Observations	9592	9592	9592	9592	9592	9592	9592	9592	9592	9592	9592
R-squared	0.098	0.099	0.280	0.296	0.298	0.344	0.404	0.443	0.444	0.448	0.452

Notes: Robust standard errors in parentheses, *** p < 0.01, * p < 0.1. Source: calculated by the authors.

.

The OLS regression results in Table 3 consistently show a significant negative coefficient for the EPT across all model specifications. The statistical significance of variables like firm size (MS) and intangible assets (NIA) further confirms that the model captures key value drivers. Overall, the OLS results provide preliminary support for H1.

4.2. Model Selection

This study explores the impact of EPT on corporate value. Selecting a fixed effects model (FE) or a random effects model (RE) is a key step in panel data analysis. The choice of the two directly affects the consistency and effectiveness of the estimation results. Therefore, we conduct a Hausman test. If the test statistic is significant (p < 0.05), reject H0 and select FE; otherwise, select RE. The regression results of the individual fixed effect and random effect models are shown in

Table 4. Regression results of the fixed effect and random effect models.

	FE	RE
VARIABLES	Tobin’s Q	Tobin’s Q
EPT	−0.329 ***	−0.419 ***
	(0.016)	(0.015)
Control variables	YES	YES
Constant	−7.843 ***	−5.594 ***
	(0.478)	(0.405)
Observations	9592	9592
Number of id	872	872
R-squared	0.441

Notes: Standard errors in parentheses, *** p < 0.01. Source: calculated by the authors.

. The Hausman test results show that the Hausman test statistic is 1340.51, corresponding to a p value is 0.0000, which strongly rejecting the null hypothesis. This indicates a correlation between the individual effect and the explanatory variables, justifying the selection of the fixed effects model.

4.3. Benchmark Regression

Therefore, we use the high-dimensional fixed effects model for further empirical testing and construct the equation as follows:

$$Y_{ijt}={\alpha }_{ijt}+\beta X_{ijt}+{\mu }_i+{\gamma }_t+{\varphi }_j+{\epsilon }_{itj}$$

(1)

where $Y_{ijt}$ is the explained variable of company i in industry j in year t, ${\alpha }_{ijt}$ is a constant term, $X_{ijt}$ is the core explanatory variable EPT, ${\mu }_i$ is the individual fixed effect, ${\gamma }_t$ is the time fixed effect, ${\varphi }_j$ is the heavy pollution industry fixed effect, ${\epsilon }_{itj}$ is the random error term. Based on this model, empirical analysis is conducted. Since control variables are gradually added to the baseline regression results, but individual fixed effects, time fixed effects, and industry fixed effects of heavy pollution industries are not controlled, we adjust the categories of fixed effects and regress them separately. The benchmark regression results are shown in

Table 5. Benchmark regression results.

	(12)	(13)	(14)
VARIABLES	Tobin’s Q	Tobin’s Q	Tobin’s Q
EPT	−0.329 ***	−0.275 ***	−0.274 ***
	(0.034)	(0.035)	(0.034)
Age	−0.149 **	0.044	0.043
	(0.064)	(0.071)	(0.072)
MS	1.486 ***	1.314 ***	1.314 ***
	(0.066)	(0.076)	(0.076)
NE	−0.429 ***	−0.399 ***	−0.400 ***
	(0.046)	(0.044)	(0.044)
FL	0.0002	−0.0012	−0.0013
	(0.001)	(0.001)	(0.001)
ROA	0.272	0.450 *	0.456 **
	(0.226)	(0.230)	(0.229)
NIA	−0.336 ***	−0.306 ***	−0.305 ***
	(0.036)	(0.035)	(0.035)
FE	−0.0920 ***	−0.0841 ***	−0.0838 ***
	(0.012)	(0.012)	(0.012)
RTSHS	−0.00692 ***	−0.00520 ***	−0.00514 ***
	(0.002)	(0.002)	(0.002)
PL	−52.30 ***	3.298	3.106
	(4.550)	(5.595)	(5.589)
DAR	−0.299 *	−0.268	−0.270 *
	(0.163)	(0.165)	(0.162)
Constant	−7.843 ***	−14.22 ***	−14.19 ***
	(1.117)	(1.449)	(1.449)
ID FE	YES	YES	YES
Year FE	NO	YES	YES
Industry FE	NO	NO	YES
Observations	9592	9592	9592
R-squared	0.740	0.757	0.757

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01, ** p < 0.05, * p < 0.1. Source: calculated by the authors.

.

After controlling for the fixed effects of individual company (ID), year and classification of heavy polluting companies in benchmark regression column (14) in Table 5, the estimated coefficient of the core explanatory variable is −0.274 ***, which proves that EPT has a significant negative impact on Tobin’s Q. For every unit increase in EPT, the Tobin’s Q value of the company decreases by 0.274 units. This indicates that an increase in EPT intensity is associated with a decrease in Tobin’s Q, suggesting that the market perceives higher environmental taxes as a net cost that can dampen a firm’s market valuation. Thus, H1 is supported.

Economically, this negative correlation may reflect investor concerns about increased compliance costs and potential squeezes on profit margins in the short term. It underscores the direct financial burden that environmental regulations can impose on companies.

However, this initial finding should not be interpreted in isolation. The positive and significant coefficient on ROA highlights that profitability remains a primary driver of firm value. This suggests that if companies can transform environmental pressures into long-term efficiency gains or innovation (as suggested by the Porter Hypothesis [32]), the initial negative impact might be mitigated or even reversed. The results thus present a classic trade-off between regulatory costs and potential efficiency gains, a central debate in environmental economics. The deeper implication underscores a critical challenge encountered by pollution-intensive listed companies within China’s green transformation: the internalization of environmental externalities exerts a direct, negative impact on their market valuation. This reflects that their prevailing business models may not have established a sustainable pathway wherein profitability is reconciled with environmental costs.

4.4. Robustness Test

This study will conduct a variety of robustness tests to ensure the accuracy of the estimated results.

4.4.1. Placebo Test

To verify that the negative impact of EPT on Tobin’s Q observed in the baseline regression was not driven by unobserved confounding factors or randomness, this study employed a rigorous placebo test. The core design follows logic similar to a double-blind clinical trial: if a placebo (in this test, a randomly generated EPT variable) that should theoretically have no effect is applied to a control group (simulated false samples in this test), and a statistically significant effect is still observed, then the genuine effect found in the baseline regression may be considered a statistical fluke.

Specifically, using Stata’s 18 permute command, the values of the core explanatory variable EPT were randomly shuffled across all observations while strictly keeping all other variables unchanged. This process was repeated 1000 times, generating 1000 simulated datasets. For each dataset, the identical baseline regression model was re-estimated, including all control variables, while absorbing fixed effects for firm identity, year, and whether a firm was heavily polluting, with standard errors clustered at the firm level. After each regression, the estimated coefficient (beta), standard error (se), and degrees of freedom (df) for EPT were recorded, and the corresponding t-statistic and p-value were calculated. This constructed an empirical distribution of the estimates and their statistics under the null hypothesis (EPT truly has no effect).

For visualization and evaluation, the distribution of the simulated t-statistics was plotted and compared against the real t-statistic from the actual regression. Additionally, a scatter plot of p-values against coefficients was used to intuitively display the relationship between the coefficient estimates and their statistical significance. A reference line at p = 0.1 helped quickly identify instances of spurious significance in the simulations. The test results are shown in Figure 6 and Figure 7.

As can be seen from Figure 6, the t-value density plot shows good normality, indicating that the model is statistically robust. The p-value in Figure 7 follows a uniform distribution, and the red dotted line indicates the significance level (0.1). Most of the p-values are concentrated above the red dotted line, which contrasts with the highly significant true estimate (−0.274 ***). Therefore, the t-value and p-value distribution obtained by the placebo test in this study are reasonable, and the model has high robustness.

4.4.2. Alternative Variables

We replace the explained variable with OPM and the core explanatory variables with EPT payable, taxes payable, and taxes and surcharges. The results are shown in

Table 6. Alternative variables.

	(15)	(16)	(17)	(18)
	Replace the Explained Variable	Replace Explanatory Variables
VARIABLES	OPM	Tobin’s Q	Tobin’s Q	Tobin’s Q
EPT	−0.008 ***
	(0.002)
EPTP		−0.274 ***
		(0.034)
TP			−0.126 ***
			(0.017)
TS				−0.160 ***
				(0.028)
Control variables	YES	YES	YES	YES
Constant	−0.0780	−14.14 ***	−14.99 ***	−13.50 ***
	(0.107)	(1.449)	(1.564)	(1.463)
Id FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Industry FE	YES	YES	YES	YES
Observations	9592	9592	9103	9592
R-squared	0.896	0.757	0.755	0.753

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01. Source: calculated by the authors.

.

In column (15,16,17,18) of Table 6, after replacing the variables, the regression coefficient is still negative, which once again verifies that EPT has a negative impact on Tobin’s Q.

4.4.3. Distinguishing Sample Test

In the original dataset, pollution discharge fees collected before 2018 were used as a proxy for the EPT in the model analysis, while the actual EPT data from 2018 onward were utilized. This study divided the overall sample into two subsamples to separately examine whether the impact on Tobin’s Q remains significantly negative. The results are shown in

Table 7. Test results.

	(19)	(20)
VARIABLES	Tobin’s Q (Year < 2018)	Tobin’s Q (Year ≥ 2018)
EPT	−0.220 ***	−0.250 ***
	(0.037)	(0.048)
Control variables	YES	YES
Constant	−13.44 ***	−23.75 ***
	(1.699)	(2.718)
Observations	5232	4360
R-squared	0.803	0.859

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01. Source: calculated by the authors.

.

In column (19,20) of Table 7, the implementation of EPT (2018 onward) exerts a greater negative impact on corporate value than the pollution discharge fee period, with high statistical significance. This demonstrates that the EPT effect not only persists post-reform but has intensified. These results confirm that EPT significantly reduces Tobin’s Q, with this negative effect becoming more pronounced after the statutory implementation of EPT.

4.4.4. Endogeneity Test

In order to ensure the consistency of estimation, avoid pseudo-regression, and identify and correct data problems, this study uses the instrumental variable method (IV-2SLS) for endogeneity test. Referring to Cai et al. [71], we construct an interaction term as an instrumental variable: the product of the river density in the city where a company is located and the water pollution intensity index of its industry (RLP). The validity of this instrumental variable is based on the following logic: cities with dense river networks are more sensitive to water pollution, and this sensitivity disproportionately falls on industries that are inherently high in water pollution intensity. Consequently, this interaction term strongly predicts the actual environmental tax burden faced by companies. Meanwhile, river density is a natural geographical feature, and industry pollution intensity is an exogenous technological characteristic; their interaction term is unlikely to affect the market value of individual companies through any causal channel other than influencing the environmental tax burden. Therefore, this meets the selection criteria for instrumental variables. The regression results are shown in

Table 8. IV-2SLS regression results.

	(21)	(22)
	Phase I	Phase II
VARIABLES	EPT	Tobin’s Q
RLP	0.218 ***
	(0.040)
EPT		−1.531 ***
		(0.324)
Control variables	YES	YES
Id FE	YES	YES
Year FE	YES	YES
Industry FE	YES	YES
Obs	9592	9592
F	29.766 ***	32.41
CD Wald F	96.226
10% IV	16.38
LM	24.502 ***
AR test	42.24 ***

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01. Source: calculated by the authors.

.

In column (21) of Table 8, the regression coefficient of the instrumental variable RLP on EPT is 0.218 ***, indicating a statistically significant positive effect of RLP on EPT and confirming a strong correlation. The Kleibergen–Paap Wald F-statistic of 29.766 *** significantly exceeds the Stock–Yogo critical value (16.38 at the 10% level), while the CD Wald F-statistic of 96.226 also surpasses the critical threshold. These results strongly reject the null hypothesis of weak identification, further validating the strength of the instrumental variable. Additionally, the LM test yields a statistic of 42.24 ***, indicating the presence of endogeneity and justifying the use of the instrumental variable approach. In the second-stage regression presented in column (22), the coefficient of EPT on Tobin’s Q is −1.531 ***, which remains significantly negative at the 1% level. Therefore, this study passes the endogeneity test, proves the effectiveness of the instrumental variables, and demonstrated the robustness of the benchmark regression results.

4.5. Heterogeneity Test

Conducting heterogeneity tests is an important research tool that helps ensure the depth, accuracy and practicality of the research. To uncover differential effects and improve the accuracy and generalizability of our findings, we conduct multiple heterogeneity tests. These include tests comparing the pollution discharge fee and EPT periods, tests based on industry pollution intensity, and tests across geographical regions (the geographical regions are divided into the eastern, central, western and northeastern regions according to the division instructions in the 2021 National Bureau of Statistics of China’s “How are economic zones divided?”). The regression results of all heterogeneity tests are shown in

Table 9. Pollution discharge fees and EPT, whether the industry is heavily polluted Heterogeneity Test.

	(23)	(24)	(25)	(26)
	2013–2017 (Pollution Discharge Fees)	2018–2022 (EPT)	Non-Heavy Pollution Industries	Heavy Polluting Industries
VARIABLES	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q
Sewage charges	−0.220 ***
	(0.037)
EPT		−0.250 ***	−0.284 ***	−0.220 ***
		(0.048)	(0.038)	(0.044)
Control variables	YES	YES	YES	YES
Constant	−13.44 ***	−23.75 ***	−17.13 ***	−8.996 ***
	(1.699)	(2.718)	(1.884)	(1.738)
Id FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Industry FE	YES	YES	YES	YES
Observations	5232	4360	6647	2938
R-squared	0.803	0.859	0.761	0.766

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01. Source: calculated by the authors.

and

Table 10. Regression results of geographical region heterogeneity.

	(27)	(28)	(29)	(30)
VARIABLES	Tobin’s Q	Tobin’s Q	Tobin’s Q	Tobin’s Q
EPT	−0.326 ***	−0.253 ***	−0.0384	−0.142 ***
	(0.049)	(0.049)	(0.119)	(0.0489)
Control variables	YES	YES	YES	YES
Constant	−14.86 ***	−14.29 ***	−14.52	−11.35 ***
	(1.942)	(2.357)	(8.888)	(3.120)
Id FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Industry FE	YES	YES	YES	YES
Observations	6463	1649	288	1190
R-squared	0.752	0.781	0.775	0.797

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01. Source: calculated by the authors.

.

In column (24) of Table 9, we can see that the intensity of the post-legislation effect of EPT has increased significantly, and the absolute value of the coefficient estimate has expanded by 3.0 percentage points, reflecting that legal constraints are more deterrent than administrative charges. The mandatory nature of taxation reduces the bargaining space of companies, resulting in a more sensitive market valuation reaction and amplifying policy shocks. In column (25,26) EPT has a significant negative impact on the Tobin’s Q of both heavily polluting and non-heavily polluting industries, and the effect is stronger for non-heavily polluting industries.

The regional heterogeneity of EPT in column (27) of Table 10 shows that the estimated coefficient in the east is −0.326 ***, which is significantly negative and has the strongest impact, reflecting that companies in economically developed regions face stricter environmental supervision and market pricing pressure. In column (28), The estimated coefficient in the central region is −0.253 ***, with the second-most negative effect. This may be due to the rapid industrial transformation (such as taking over the transfer of high-pollution industries in the east), and the tax burden impact is partially offset by technological upgrading. In column (29), the estimated coefficient in the northeast is −0.0384, which may be due to policy protection or implicit subsidies to buffer the tax burden. In column (30), the coefficient in the west is −0.142 ***, with a significant negative effect but weaker than that in the east and central regions, which may benefit from the policy tilt of “Western Development”.

The inhibitory effect of EPT on corporate valuation shows a gradient characteristic of “East > Central > West > Northeast”, which is closely related to the regional marketization levels, industrial structures, and policy support. Therefore, H2 is supported.

4.6. Mechanism Test

In order to further reveal the potential mechanism in the causal chain, this study conducted a moderating effect test and a mediating effect test, respectively.

4.6.1. Moderating Effect

The moderating variables used in this study are debt-to-asset ratio, operating profit margin, and R&D investment amount. The moderating effect model is constructed as follows:

$$Y_{ijt}={\alpha }_{ijt}+\beta X_{ijt}+\beta X_{ijt}\times {Mod}_{ijt}+{\mu }_i+{\gamma }_t+{\varphi }_j+{\epsilon }_{itj}$$

(2)

where ${Mod}_{ijt}$ is the moderating variable, $X_{ijt}\times {Mod}_{ijt}$ is the interaction term between the core explanatory variable and the moderating variable, which is the focus of our attention, and the explanations of other variables are as in model (1). The regression results are shown in

Table 11. Moderating effect model regression results.

	(31)	(32)	(33)
VARIABLES	Tobin’s Q	Tobin’s Q	Tobin’s Q
EPT	−0.143 ***	−0.271 ***	−0.224 ***
	(0.048)	(0.034)	(0.035)
Xijt × Modijt	−0.277 ***	−0.015 *	−0.003 ***
	(0.099)	(0.008)	(0.001)
Control variables	YES	YES	YES
Constant	−16.27 ***	−14.27 ***	−14.52 ***
	(1.682)	(1.449)	(1.456)
Id FE	YES	YES	YES
Year FE	YES	YES	YES
Industry FE	YES	YES	YES
Observations	9592	9592	9592
R-squared	0.758	0.758	0.758

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01, * p < 0.1. Source: calculated by the authors.

.

The regression results of the moderating effect model in column (31) of Table 11 show that the estimated coefficient of EPT in the main effect is −0.143 ***, and the negative effect is weakened, but the interaction term −0.277 *** is significantly negative. For every unit increase in the debt-to-asset ratio, the marginal negative effect of EPT on Tobin’s Q is amplified by 0.3 percentage points, which verifies that highly leveraged companies are more seriously affected by EPT, and highly indebted companies face dual pressures (environmental compliance costs and interest expenses). Insufficient financial flexibility exacerbates valuation discounts, supporting H3a.

In column (32), the estimated coefficient of the interaction term between EPT and operating profit margin is −0.015 *, which is negative and marginally significant. Companies with high operating profit margins can buffer the impact of EPT and weaken the marginal negative impact. This may be because high-profit companies can make short-term environmental investments (such as equipment renewal), or because investors interpret such investments as positive signals. Therefore, H3b is supported.

In column (33), the estimated coefficient of EPT income and R&D investment amount is −0.003 ***, which is significantly negative. Innovation, fueled by R&D investment, is a core method for increasing corporate value, ease cost pressure, and slow down the downward trend of market valuation, so H3c is supported. This suggests that companies need to respond to EPT challenges through strategic reforms rather than relying on spontaneous adjustments.

4.6.2. Mediating Effect

This study introduces innovation as a mediating variable to explore the potential mechanism, and again employs R&D investment as one measure. In addition, two indicators—patent application quality and authorized patent quality—are considered to analyze the output stage of innovation. R&D investment represents the input, while patent application quality and patent grant quality represent the output. The mechanism is tested separately, and the measurement methods of patent application quality and authorized patent quality are innovated [72]. This study obtains the main classification number of the patents of listed companies. The IPC patent classification number format used in China is “department—major category—minor category—major group—minor group”, such as A01B01/00.

First of all, if only the number of main classification numbers contained in the company patents is used to measure the quality of company patents, it cannot accurately distinguish qualitative differences among patents and may lead to overestimation. For example, a company has three patents, whose main classification numbers are A01B01/01, A01B01/02, and A01B01/03, while the main classification numbers of the three patents of another company are A01B02/00, B02C03/00, and C03D04/00. Although both companies have three patents, the first company’s patents fall under only one major group (A01B01), while the second’s span three distinct major groups (A01B02, B02C03, C03D04). Obviously, the technical level used in the second company’s patent is greater than that of the first, so its patent quality is correspondingly higher. This study refers to the calculation idea of the Herfindahl Index and defines the quality of corporate patents at the large group level:

$${\mathrm{P}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{t}}_{\mathrm{i}\mathrm{t}}=1-\sum {\left({\displaystyle \frac{Z_{mit}}{Z_{it}}}\right)}^2$$

(3)

where $Z_{mit}$ is the cumulative number of invention and utility model patents applied for by company i under group m as of year t, and $Z_{it}$ is the cumulative number of patents applied for by company i under all groups as of year t. The larger the value of Patent_it, the higher the quality of the patents applied by the company, and similarly for the quality of authorized patents.

In the data processing process of calculating this indicator, we only selected invention patents and utility model patents. Therefore, we constructed the mediation effect model based on the stepwise regression method [73] as follows:

$${Med}_{ijt}={\alpha }_{ijt}+\beta X_{ijt}+{\mu }_i+{\gamma }_t+{\varphi }_j+{\epsilon }_{itj}$$

(4)

$$Y_{ijt}={\phi }_{ijt}+\psi X_{ijt}+{Med}_{ijt}+{\mu }_i+{\gamma }_t+{\varphi }_j+{\epsilon }_{itj}$$

(5)

where ${Med}_{ijt}$ is the moderating variable, and the other variables are explained as in model (1, 2). The regression results are shown in

Table 12. Regression results of the mediation effect model.

	(34)	(35)	(36)	(37)	(38)	(39)
VARIABLES	R&D	Tobin’s Q	PAQ	Tobin’s Q	QAP	Tobin’s Q
EPT	0.126 ***	−0.268 ***	0.0109 ***	−0.272 ***	0.008 ***	−0.273 ***
	(0.020)	(0.034)	(0.004)	(0.034)	(0.003)	(0.034)
R&D		−0.049 ***
		(0.014)
PAQ				−0.205 ***
				(0.079)
QAP						−0.137 *
						(0.075)
Control variables	YES	YES	YES	YES	YES	YES
Constant	9.710 ***	−13.72 ***	0.830 ***	−14.02 ***	0.387 **	−14.13 ***
	(1.220)	(1.455)	(0.170)	(1.441)	(0.166)	(1.446)
Id FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
Industry FE	YES	YES	YES	YES	YES	YES
Observations	9592	9592	9592	9592	9592	9592
R-squared	0.814	0.758	0.401	0.757	0.420	0.757

Notes: We report robust standard errors in parentheses and cluster robust standard errors at the individual level. *** p < 0.01, ** p < 0.05, * p < 0.1. Source: calculated by the authors.

.

It can be seen that the model regression results are all significant in Table 12, but in order to test whether the existence of the mediation effect is significant, a mediation effect test is needed. This study uses the modified Bootstrap method [74,75]. By repeatedly extracting samples 1000 times and calculating the indirect effect, the Bootstrap distribution of the indirect effect can be obtained, and then its confidence interval can be estimated. This can provide a more accurate estimate of the size and significance of the mediation effect. The Bootstrap results are shown in

Table 13. Bootstrap.

	Observed	Bootstrap			Normal-Based
	Coefficient	Std. Err.	z	p > |z|	[95% Conf. Interval]
M1 indirect effect	−0.034	0.006	−5.53	0.000	−0.046	−0.022
M1 direct effect	−0.268	0.025	−10.74	0.000	−0.317	−0.219
M2 indirect effect	−0.002	0.001	−2.87	0.004	−0.004	−0.001
M2 direct effect	−0.273	0.025	−10.98	0.000	−0.321	−0.224
M3 indirect effect	−0.002	0.001	−2.92	0.004	−0.005	−0.001
M3 direct effect	−0.272	0.025	−10.95	0.000	−0.321	−0.223

Source: calculated by the authors.

.

Through the Bootstrap test in Table 13, we believe that EPT significantly increases R&D investment (0.126 ***), indicating that EPT pressure prompts polluting companies to increase R&D spending. EPT has a significant positive impact on the quality of applied patents (0.0109 ***) and the quality of authorized patents (0.008 ***). It has a significant negative impact on Tobin’s Q in both the R&D investment and the stage where innovation is transformed into outputs (patents). The order of the calculated total effect is M1 (−0.3021 ***) > M2 (−0.2755 ***) > M3 (−0.2751 ***), indicating that both the innovation input and the innovation results failed to be converted into market valuation improvement, which is inconsistent with the Porter hypothesis.

Based on the theoretical mechanism of cost pressure, market expectations, and innovation lag, companies’ R&D investments and outputs in end-of-pipe treatment technologies for compliance purposes increase costs and crowd out short-term resources and profits. This represents short-term passive innovation driven by environmental regulations. Such passive innovation leads to a negative correlation with Tobin’s Q and insufficient green premiums, as the market has not yet formed a value consensus on green technologies. Due to the dominance of passive innovation, short-term market behavior, and the lack of green premiums, market valuation is ultimately undermined. Therefore, we reject H4.

5. Discussion

5.1. Verification of Hypothesis

Based on the data of Chinese listed companies, this study systematically examines the impact mechanism and boundary conditions of EPT on corporate market value. The empirical results reveal the multidimensional complexity of the effect of environmental regulation policies, which partially verifies the predictions of classical theory and highlights the special mechanism in the context of China’s transitional economy.

First, the main regression confirms the significant negative impact of EPT on corporate Tobin’s Q, which is verified by the placebo test, substitution variable test and the multidimensional robustness tests using the instrumental variable method. Therefore, H1 is verified and accepted, which supports the core assertion of neoclassical cost theory. EPT directly erodes corporate profits and market valuations by internalizing compliance costs. This result is consistent with the evidence of [21,76,77], indicating that in the early stage of policy implementation, companies generally face short-term pain dominated by “cost shock”.

Secondly, the heterogeneity test reveals the differentiated mechanism of the impact of environmental policies (EPT) on corporate valuation (Tobin’s Q), which echoes the core findings of recent international studies. First, in terms of policy tool heterogeneity, the negative effect of EPT is significantly stronger than that of pollution discharge fee, verifying the stronger mandatory effect of taxation [78]. This mandatory effect constrains the level of corporate risk-taking [79]. Second, industry heterogeneity shows that non-heavy-polluting industries are more sensitive, because heavily polluting companies may benefit from peer effects within their industry, and the market may have already anticipated their environmental costs [80]. Finally, regional heterogeneity shows a gradient effect of “East > Central > West > Northeast”, reflecting the interaction between marketization level and policy intervention. The high marketization in the east leads to more efficient pricing of environmental costs, resulting in a greater valuation loss for polluting firms, consistent with [81]. These findings all support H2, highlighting that environmental policies need to be precisely designed in combination with industry characteristics and regional development differences to avoid “one-size-fits-all” supervision.

Then, the influence mechanism of EPT on corporate valuation (Tobin’s Q) was revealed through the moderating effect test. The moderating effect of financial leverage showed that high asset-liability ratio significantly amplified the negative effect of EPT. High-debt polluting companies have increased financial distress risks and weakened market confidence due to the superposition of environmental compliance costs and interest expenses. Secondly, although the moderating effect of operating profit margin is marginally negative, high-profit polluting companies partially buffer the impact of EPT through short-term environmental investment, which may be because investors regard green investment as a signal of long-term competitiveness, which is partially consistent with the greenwashing premium [82]. Finally, the negative moderating effect of R&D investment shows that innovation has failed to effectively hedge EPT costs, but the long R&D cycle and policy uncertainty have suppressed valuations. Therefore, supports H3a, H3b, and H3c.

Finally, this study found that the innovation compensation mechanism of Porter’s hypothesis was not fully supported in this study. This is in stark similarity to the conclusion of the innovation path proposed by [83], so we reject H4. Under the regulation of China’s 2018 EPT, polluting companies significantly increase R&D investment and strive to enhance the quality of patent applications and grants to meet compliance standards and ensure operational continuity. However, in the short term, such externally pressured innovation acts as a negative mediating channel, where innovation costs dominate, thereby suppressing corporate market value. The core theoretical mechanism is as follows: First, EPT-induced innovation exhibits a defensive nature and cost attribute, as resources are heavily allocated to end-of-pipe treatments or process improvements. While such investments improve innovation quality, they directly crowd out short-term profits and are perceived by investors as financial burdens due to their inability to swiftly translate into market revenue. Second, market expectations and information asymmetry exacerbate this effect. When investors observe surging R&D expenditures, they are more likely to interpret them as indicators of severe compliance costs and operational risks rather than assurances of future competitiveness, leading to valuation downgrades. Thus, although the EPT successfully stimulates both stages of innovation input and output, the accompanying high costs, resource crowding-out, and negative market signals collectively contribute to a negative relationship between innovation activities and corporate market value, revealing the potential short-term pains in the realization of the innovation compensation effect posited by the Porter Hypothesis. In summary, the negative impact of the EPT on Tobin’s Q through innovation mediation reflects the combined effects of cost pressure, market expectations, and innovation lag. This underscores the importance of designing environmental policies that account for polluting companies’ financial capacity and innovation types to avoid excessive short-term costs. Simultaneously, polluting companies should communicate the long-term value of their innovations to improve market expectations.

5.2. Theoretical Contribution

This study develops an integrated model of “institutional pressure-resource constraints-market feedback,” revealing the multiple moderating and mediating mechanisms through which environmental regulations exert their effects. This provides a new analytical framework for environmental economics in emerging markets.

5.3. Policy Implications

This study has three implications for policy design: First, it is necessary to establish a differentiated and dynamic EPT system, implement a tax rate progressive mechanism for highly marketized regions and heavily polluting industries in the east, and set a policy buffer period for western and startup polluting companies; second, promote the transformation of innovation incentive policies, from R&D investment subsidies to green patent commercialization support, and overcome barriers in the innovation value chain; third, improve environmental information disclosure and market pricing systems, guide investors to identify long-term green value, and correct valuation bias caused by short-term cost shocks.

5.4. Limitations and Future Directions

This study has several limitations. First, although fixed-effects models were employed, endogeneity issues between EPT and corporate value may not be fully resolved. Second, while innovative attempts were made to measure EPT intensity and patent quality, their accuracy and validity still require further improvement. Third, as the conclusions are based on a sample of Chinese listed companies, their generalizability to non-listed companies or other emerging markets should be approached with caution. Additionally, the mechanism tests primarily focus on innovation pathways and do not comprehensively incorporate other potential mediating variables such as reputation and supply chain management.

In summary, the findings provide some support for the “compliance cost hypothesis.” However, under the short-term effects of the EPT policy, the study fails to validate the innovation compensation mechanism of the “Porter Hypothesis.” This hypothesis may be constrained by the current stage of market development, and its long-term dynamics warrant further investigation. Future research could explore how to optimize innovation strategies to counteract this negative impact, such as by focusing on high-value innovations or enhancing investor relations.

6. Conclusions

Based on the data of Chinese polluting companies, this study systematically reveals the impact mechanism and action boundaries of EPT on corporate market value. The main conclusions are as follows:

First, EPT has a significant negative effect on the Tobin’s Q of companies. For every 1 percentage point increase in tax burden, the company valuation decreases by 0.27–0.33 percentage points. This effect is verified by multidimensional robustness tests and instrumental variables (IV), confirming that EPT directly compresses the profit space of companies through the cost transmission mechanism and triggers investors to re-evaluate long-term environmental risks, which is consistent with the expectations of neoclassical cost theory.

Second, the policy effect shows significant heterogeneity. In terms of time dimension, EPT is a stronger constraint mechanism, which increases the valuation impairment of companies. In terms of space dimension, the tax burden in the eastern region has a greater impact on the value of companies due to its high level of marketization and strict environmental law enforcement; in terms of industry dimension, the valuation loss of high-pollution industries is weaker than that of non-heavily polluting industries, reflecting the linkage pricing logic of EPT and pollution risk.

Third, green innovation has failed to effectively hedge the impact of tax burden, and R&D investment and patent quality have even aggravated the negative impact. Empirical evidence shows that Chinese corporate innovation is concentrated on end-of-pipe treatment technologies (such as sewage treatment equipment). Its short-term cost increase and delayed commercialization lead to the “Porter Paradox”, highlighting the insufficient explanatory power of the traditional Porter hypothesis under the constraints of innovation quality and institutional context. This study provides a basis for policy optimization under the “dual carbon” goal.