Can environmental tax bring strong porter effect? Evidence from Chinese listed companies

Existing researches verify Porter hypothesis mainly through one of its core establishment paths, innovation compensation, but ignore the other one, first-mover advantages. This paper considers both these paths and further distinguishes environmental R&D and non-environmental R&D in empirical study. Based on the smooth transition principle of “charge to tax” in China, this paper, taking Chinese A-share listed companies that have disclosed environmental R&D from 2008 to 2017 as sample and predicting environmental tax by pollution charge, analyzes the relationship between environmental tax and firm performance and its transmission mechanism. First, our results show that environmental tax can improve firm performance, and this influence remains in long term. After introducing instrument variable (IV) to deal with endogeneity and conducting a series of robustness tests, we then find that the relationship between environmental tax and firm performance is a robust causality. Eventually, we apply mediating effect model to further confirm the two core paths of Porter hypothesis by manifesting that environmental tax can affect firm performance through both innovation compensation and first-mover advantages. Our contributions are revealing the micro mechanism of environmental tax on firm performance, providing evidence from China to support strong Porter hypothesis, and offering some key points regarding environmental tax reform and corporate green strategy.


Introduction
Negative externality in the exploitation and utilization of environmental resources cannot be solved by market alone. As a form of Pigouvian taxes, pollution charge is an important way to internalize the external cost of a firm. Chinese government promulgated "Interim Measures for the Collection of Pollution Charge" in 1982, which formally established the pollution charge system in China. It has had a profound influence on Chinese environmental governance in more than 30 years (Guo et al., 2019). In order to further standardize the pollution charges, "Environmental Tax Law of China" was implemented in 2018 to change pollution charges into environmental tax officially. According to the polluter pays principle (PPP), firms are the major object of the environmental tax, which may burden them with more operating cost. So, we come up with the following questions naturally: Will environmental tax affect the competitiveness of firms, and can it produce strong Porter effect? If so, what is the path? To answer these two questions is of great significance to evaluate the implementation effect of environmental tax and guide the reform to the next stage.
Although the impact of the levy of environmental tax on macroeconomic development and firms' behavior as micro-players in China is still unknown due to lack of relevant data and short execution time, what is known is that the legislation of environmental tax follows the principle of smooth transition of "charge to tax" that no significant change in the levy object, scope and calculation standard between environmental protection tax and pollution charge (Wang et al., 2019a, b). The biggest change from pollution charge to environmental tax is that the legal status of pollution charge is relatively low, which is easy to Communicated by Eyup Dogan. form levy and management obstacles (Wang et al., 2019a). The inherent legal authority of environmental tax further strengthens the tax rigidity, which will significantly improve the intensity of levy (Xu, 2015;Huang and Li, 2018). Based on the fact that there is no big difference between environmental tax and pollution charge, we can evaluate the implementation effect of Chinese environmental tax through the policy of pollution charge (Lu et al., 2019) and discuss whether environmental tax can produce strong Porter effect and how, so as to provide policy reference for the reform of environmental tax. This paper may have the following marginal contributions. First of all, the existing studies do not explain the mechanism of environmental regulation affecting firm competitiveness clearly. On the one hand, although there are relevant literature on mechanism of innovation compensation, they do not further distinguish between environmental R&D and non-environmental R&D whose impacts are obviously different. This paper separates environmental innovation from overall innovation by collecting the corporate environmental R&D data in CSR reports, focusing on whether environmental innovation can bring innovation compensation effect. On the other hand, previous researches mainly discuss about innovation compensation but ignore the effect of first-mover advantage although Porter hypothesis contains these two core paths. On the contrary, this paper analyzes both the effects of innovation compensation and first-mover advantage, providing micro evidence from China to validate strong Porter hypothesis and its mechanism. Secondly, this paper fully considers the endogenous problem resulting from reciprocal causation and omitted variables. We prove that there is a robust causal relationship between environmental tax and firm performance by introducing instrumental variables and conducting a series of robustness tests, which provides causal identification evidence for the strong Porter hypothesis. Thirdly, although a handful of literature compare different types of environmental regulation rather than only use comprehensive environmental regulation as independent variable, it is difficult to analyze the causality and mechanism of specific environmental regulation tools under this approach. For more in-depth analysis, this paper only takes the most typical market-based environmental regulation in China, environmental tax, into consideration and discusses corresponding causality relationship and influence mechanism.
This study proceeds as follows. Section 2 contains literature review. Section 3 introduces policy background and characteristic facts. Section 4 provides the methodology and describes the data. Section 5 discusses empirical procedures and major findings. Section 6 consists of research conclusions and policy recommendations.

Studies on environmental tax
The existing literature on the economic consequences of environmental tax mainly divide into macro and micro levels. At the macro level, the "double dividend" of environmental tax is a trending topic (Bashir et al., 2021a). The double dividend of environmental tax means that it can bring other social dividend while improving environmental quality (Pearce, 1991;Fullerton and Metcalf, 1997) . It has become a consensus among most scholars that environmental tax can bring environmental dividends (Bjorner and Jensen 2002;Miller and Vela 2018;Bashir et al, 2020). However, there are some different views on social dividend. For example, three representatives of them are, first, efficiency dividend: Komen and Peerlings (1999) found that energy tax in the Netherlands can improve environmental quality and reduce tax distortion, and Takeda (2007) found that carbon tax in Japan reduces distorted tax revenue. Using data from OECD countries, Bashir et al. (2021b) found that environmental tax promotes energy efficiency. Second of them are employment dividend: Carraro et al. (1996) used CGE model to simulate the effect of European carbon tax and found that it reduces the unemployment rate. Bezdek et al. (2008) found that environmental tax is conducive to the expansion of environmental protection industry, thus creating new jobs. Third one is distribution dividend: Budzinski (2002) found that the collection of environmental tax can reduce the tax burden of low-income people and make social distribution more equitable. Heerden et al. (2006) took South African families as research sample and found that environmental tax can narrow the gap between the rich and the poor. At the micro level, environmental tax can promote firm innovation (Porter and Linde, 1995) and improve corporate green investment (Huang and Lei, 2021). Bertarelli and Lodi (2019) found that environmental tax will positively affect eco-innovation propensity and, indirectly, export propensity.

Studies on Porter hypothesis
In order to test Porter hypothesis better, Jaffe and Palmer (1997) divided it into three forms. The first one is weak Porter hypothesis, which implies that strict environmental regulation policy can stimulate innovation in firms, but its influence on innovation that increases firm competitiveness is uncertain. Scholars have basically reached a consensus that the weak Porter hypothesis is tenable. Calel and Dechezleprêtre (2016) found that regulated firms have higher innovation investment than unregulated firms based on European environmental regulation policy research. Empirical studies by Manderson and Kneller (2012), Nesta et al. (2014), Song et al. (2019) and Shang et al. (2021) also found that environmental regulation and innovation are positively correlated. The second is the narrow Porter hypothesis, which indicates that flexible regulatory policies incentivize firms greater to innovate than stiff ones. López-Gamero et al. (2010) found that flexible environmental regulation is conducive to encourage firms to develop new processes and products while stiff regulatory policies are not. Jaffe et al. (2002), Brouhle et al. (2013), Ambec et al. (2013) and Sun et al. (2021) also prove that compared with direct regulation based on command and control such as environmental standards and emission limits, market-based and flexible regulation more greatly motivate firms to innovate. The third is the strong Porter hypothesis, which means that innovation in firms provoked by stronger environmental regulation can completely offset any additional costs from environment protection and improve firms' competitiveness. The conclusions of existing researches on strong Porter hypothesis are controversial, and scholars often conduct research from the perspective of productivity. Denison (1981) found that the reason for the productivity of the USA decreased by 16% during 1972-1975 is the intensified environmental regulation. Cagatay and Mihci (2006) and Hering and Poncet (2014) also get a similar conclusion that environmental regulation has an adverse effect on competitiveness. However, Berman and Bui (2001) found that the productivity growth of the petroleum smelting industry in Los Angeles, where strict air quality control has been implemented, is much higher than that in other parts of the USA. Other literature, such as Hamamoto (2006), Peuckert (2014), Wang et al. (2019a, b), found evidence that environmental regulation improves productivity as well.
From the literature above, we can see that the major controversy existing in three forms of Porter hypothesis is the strong Porter hypothesis. We infer that there are four important reasons for the inconsistency. Firstly, Porter and Van der Linde (1995) believe that strict and properly designed environmental regulation can stimulate innovation and partially or completely offset the adverse effect of compliance cost. Therefore, the prerequisite in the establishment of the strong Porter hypothesis is strict and properly designed environmental regulation. Although what is "strict and properly designed" environmental regulation is still inconclusive, studies have confirmed that there is an inflexibility in command-control environmental regulation, and inflexible regulation even inhibits the environmental innovation behavior in firms (Montero, 2002;Requate and Unold, 2003). The comprehensive environmental regulation indicators used in literature often include inflexible environmental regulation, which may make the results inaccurate for they do not comply the prerequisite of the strong Porter hypothesis. Secondly, environmental regulation indicators are often endogenous in economic research (Copeland and Taylor, 2004): Environmental regulation variables and dependent variables are reciprocal causation, or environmental regulation variables correlated with omitted variables and other unobservable factors lead to endogenous problems, which will affect the accuracy of the conclusion greatly. Thirdly, the existing literature does not make a clear distinction between environmental innovation and non-environmental innovation. Researchers often consider the overall innovation in firms, but environmental regulation promotes green-biased innovation the most (Requate and Unold, 2003;Acemoglu et al. 2012) , and the effects of the two types of innovation on firms competitiveness are obviously different, misleading the results. Fourth, most of the researches measure the firm competitiveness by their total factor productivity (TFP). In fact, the firm competitiveness is also affected by many factors in addition to TFP, including demand factors, factor endowments and corporate image. For example, the environmental protection behavior of a firm may also enhance competitiveness through providing environment-friendly products and establishing a green corporate social image. Kammerer (2009) found that some consumers are more willing to pay for green products with high prices when the products reduce pollution and have added value.
We attempt to alleviate the four problems mentioned above. Taking Chinese A-share listed companies that disclosed environmental R&D from 2008 to 2017 as samples, we focus on the environmental tax which is the most typical market-oriented environmental regulation in China. 1 We measure firm competitiveness by firm performance to test whether environmental tax can produce a strong Porter effect. Furthermore, instrumental variable is introduced to reduce the adverse effect of endogeneity, and a series of robustness tests are conducted to prove that there is a stable causal relationship between environmental tax and firm performance. In the end, this paper establishes a mediating effect model to analyze the mechanism from environmental tax to firm competitiveness according to two paths of Porter hypothesis-innovation compensation and first-mover advantage. Figure 1 shows the empirical research framework of this paper.

Policy background and characteristic facts
Pollution charge made its debut out of the polluter pays principle (PPP) proposed by the Environment Committee of the Organization for Economic Cooperation and Development (OECD) in 1972. The "Environmental Protection Law of China for Trial Implementation" proposed in 1979 marked the establishment of pollution charge policy in some pilot areas in China. China issued the "Interim Measures for the Collection of Pollution Charge" in 1982, indicating that the pollution charge policy was formally established and widely implemented throughout the country. China promulgated "Regulation on the Administration of Collection and Use of Pollutant Discharge Fees" in 2003, which improved the pollution charge system greatly. The original charge for single factor of pollution exceeding the standard was replaced by the total multi-factor charge based on the type and quantity of pollutants. Pollution charge policy has played a significant role for nearly 40 years in China, but environmental problems are still common in China (Xu and Xie, 2016). The limitations of the pollution charge are exposed gradually, mainly due to the lack of rigidity in law enforcement and excessive intervention by local governments (Lu et al., 2019). In order to improve these deficiencies, the "Environmental Tax Law of China" was implemented in 2018, presenting that the pollution charge is transformed into environmental tax officially.
According to the polluter pays principle (PPP), firms are the basic units that pay for pollution. Figure 2 shows the total amount of pollution charge and the total number of firms paid for pollution in China from 1995 to 2015. The amount of pollution has overall maintained a rising trend. It is worth noting that the amount of pollution charge has increased sharply after 2003. One reason is that the new regulation issued in 2003 requires that double charge has to be paid for the pollution amount exceeding emission standard. The number of firms paying pollution charges began to decline significantly after 2003. There are two possible reasons: First, the introduction of new regulations in 2003 has greatly increased the pollution costs of firms. Firms have to adopt cleaner and more efficient technologies to reduce pollution emission out of maximizing their own interests, and some firms refrain from paying charge due to their effective green transformation. Second, firms with weak innovation ability and low productivity will face high pollution costs in the already competitive market and could be eliminated from the competition and exit the market eventually. A possible reason for the slowing growth rate of pollution charge after 2008 is that local governments have to loosen the collection of pollution charge to secure economic growth under the impact of What is the path of environmental tax bringing strong Porter effect?  Combining observations on the amount of pollution charge and the number of paying firms, we find that the number of paying firms has been declining since 2003, while the general trend in the amount of pollution charge is still rising, which indirectly indicates that the actual intensity of levy is increasing. Figure 3 shows the details of pollution charge in several years. It can be found that pollution charge mainly levies on waste gas and water, while the charge for industrial residue and noise only accounts for a small proportion. Therefore, two emission indicators, waste gas and polluted water, are used to measure the intensity of environmental tax in our paper.

Data source and processing
The study sample consists of Chinese A-shares listed companies in the period of 2008-2017. We choose this time interval out of the following two reasons: First, the China Securities Regulatory Commission has recommended listed companies to issue corporate social responsibility (CSR) reports since 2008, so corporate environmental R&D data can be obtained from 2008. Second, 2017 is the last year for the collection of pollution charge, and from 2018 environmental tax is levied, but no corresponding data is available at present. The raw data is processed as following rules: (1)

Fig. 3 Details of pollution charge in China (China Environmental
Yearbook only disclosed the details of the pollution charge in these 3 years, but we can infer that the levy on waste gas and waste water also accounts for the most of the pollution charge in other years)

Dependent variable
Scholars mainly use return on assets (ROA) and TobinQ as proxy variables of firm performance (Morgan et al., 2002). The calculation of TobinQ value involves the internal financial data and external market data. However, the effectiveness of the Chinese capital market has been questioned by some scholars due to its incomplete construction and irrational investor (Xu et al., 2006). Therefore, we use return on assets (ROA), return on equity (ROE) and return on sales (ROS) as proxy variables of firm performance in baseline regression and TobinQ in robustness check.

Independent variables
This paper uses pollution charge data to predict and analyze environmental tax, as there is no available data on Chinese environmental tax by now. We argue that this method holds because similarities do exist between these two caused by smooth transition of "charge to tax" reform mentioned above. Lu et al. (2019) also point out that the types of pollution charge include polluted water, waste gas, industrial residue and noise, which has no substantial change after the "charge to tax" reform. It can be seen from Fig. 3 that the charge for industrial residue and noise only accounts for a small proportion, and there are also many missing values in the data of industrial residue and noise emission. Therefore, referring to Shen et al. (2017), we select SO 2 and COD in waste gas and water as pollution indicators and calculate the common pollution factor-by-factor analysis method and use the ratio of pollution charge and the common pollution factor as proxy variables of environmental tax.

Mediating variables
The China Securities Regulatory Commission issued the guidelines for environmental information disclosure of listed companies in 2008 to encourage listed companies to disclose environmental information in corporate social responsibility (CSR) report. Some listed companies have disclosed environmental R&D since then, which provides data support for this study. These data are mainly disclosed in the environmental protection and sustainable development parts of the CSR report. We manually collected the data from 2008 to 2017. In line with Wang et al. (2017), we use the natural logarithm for the initial environmental R&D data to avoid the influence of heteroscedasticity. The market share is measured by the ratio of the revenue of firms to the total revenue of their belonged industry according to the industry classification standard of the China Securities Regulatory Commission.

Control variables
Referring to Rammer et al. (2017), this research selects the control variables from three aspects of corporation characteristics, financial position and firm governance. The specific control variables are selected as following: (1) Firm size (Size), measured by natural logarithm of the total assets of the firms.
(2) Financial leverage (Lev), measured by the ratio of total liabilities to total assets. (3) Management expenses (Cost), measured by the ratio of management expenses to total revenue. (4) Equity concentration (Top1), measured by the shareholding ratio of the largest shareholder of the firm.
(5) Growth of firm (Growth), measured by the increase rate of main business revenue. We also control regional variables to reduce the adverse effects of omitted variables: (6) Regional economic development level (GDP), measured by regional real GDP growth rate. (7) Regional market environment (Market), measured by regional Marketization Index.

Model
To investigate the impact of environmental tax on firm performance, we run the following baseline regression 2 : where 2 Thanks for comments from an anonymous reviewer. Since the environmental tax may also have a U-shaped or inverted U-shaped relationship with firm performance rather than a linear relationship, we also add the square term of Tax, Tax_sq, to the model for testing, as shown in Appendix Table 7. It can be seen from the regression results of adding the square term that the regression coefficients of Tax and Tax_sq in all models are not significant at the same time, and the coefficient of determination (R-squared) is low, indicating that there is no U-shaped or inverted U-shaped relationship. 3 In fact, there may be a reciprocal causation between environmental tax and firm performance. Therefore, we have added the environmental tax variable lagged one period to reduce the negative impact of reciprocal causation. But this approach can not completely deal with endogenous issues, and we will further use IV method to address endogenous issues.
observe to some extent. Considering the macro impact in different years, such as the financial crisis or the change of credit policy that will affect the firm performance, we also introduce the time dummy variable Year. ε i,t is the disturbance term. Considering that the disturbance term may have heteroscedasticity, we use the modified Wald test to test it under the fixed effect, and the results show that heteroscedasticity exists. To avoid the influence of heteroscedasticity, we cluster the standard errors on firm level in all regressions. The above test results are shown in Table 2.

Baseline regression
The baseline regression results are represented in Table 3. From the results of columns 1-3 in Table 3, the regression coefficients of Tax are significantly positive, indicating that there is a positive correlation between environmental tax and firm performance. In order to test the long-term effect of environmental tax and reduce the impact of reciprocal causation, we also test the environmental tax lagged one period in the baseline regression. From the results of columns 4-6 in Table 3, the regression coefficient of Tax is still significantly positive, which indicates that the positive promotion  Hamamoto (2006) and Wang et al. (2019a, b). The former one found that environmental policies have improved the productivity of manufacturing industry in Japan, and the latter one applied data from OECD countries to prove that environmental policies have increased green productivity in the industrial sectors of these countries. Their researches are based on data from industry level, while our research from micro level at enterprises shows environmental tax promoting competitiveness. However, it is worth noticing that this paper only takes the firms that are going concern into this regression because those low-tech firms that are unable to create effective green innovation could be gradually eliminated by the market due to the increase of operating costs. So, the conclusion may only hold true for the firms that continue to survive on the marktet.

Endogenous problems
Many literatures point out that there exist endogenous issues among environmental tax (Copeland and Taylor, 2004;Rubashkina et al., 2015). Endogenous problem refers to the correlation between explanatory variables and disturbance term, i.e. Cov(Tax, ) ≠ 0 in this paper. The possible sources of endogeneity include the followings, and we take them into consideration: (1) Reverse causality, that is, the environmental protection tax will affect the firm performance, but the firm performance in turn may also affect the environmental tax. For example, one government may implement a stricter Table 4 The results of IV-2SLS (1) *** p < 0.01, ** p < 0.05, * p < 0. environmental tax policy for specific firms .
(2) Omitted variable. It refers to those variables related to the independent variable environmental tax that exist in the disturbance term. These variables are not included in the control variables because they are unobservable or difficult to estimate. For example, industrial policy, official turnover, regional factor endowment and other factors are related to the environmental protection tax possibly and exist in the disturbance term. However, it is difficult for us to accurately quantify and fully include them into the model. (3) Sample selection bias. We use the data from listed companies due to data availability, and listed companies are often the better representatives of all companies, so our sample selection process is not random. (4) Errors in the original data will also lead to endogenous problems. Although this paper has tested environmental tax variables lagged one period, the endogeneity may still exist because of omitted variables and other factors. The endogenous problem of omitted variable will cause our independent variable environmental tax to be related to the disturbance term, i.e. Cov(Tax, ) ≠ 0 , one of the important conditions for the establishment of traditional OLS is Cov(Tax, ) ≠ 0 ; otherwise, the estimated amount obtained will be inconsistent, and there will be endogeneity bias. In this case, the consistent estimator can be obtained by introducing the instrumental variable (IV) and using Two Stage Least Square (2SLS) method (Wooldridge, 2015). Therefore, this paper further uses IV-2SLS method to conduct empirical tests on environmental tax and firm performance.
According to the estimation method of 2SLS, we first take the IV Energy and the control variable as dependent variables and tax as the independent variable for regression (Eq. 2). This step is also called the first stage in 2SLS.

Model (3) is the fitted value of Tax obtained by regression of Model 2.
Then, the fitted value T ax is used to replace the Tax in model (1) for regression, which is also called the second stage in 2SLS.
The selection of IV should satisfy two conditions. One is that IV should be highly correlated with explanatory variables. The other is that the IV is not related to the (2) Tax it = 0 + 1 Energy it + 2 Size it + 3 Lev it + 4 Cost it + 5 Top1 it disturbance term. There are two reasons why the standard coal of regional energy consumption is a feasible instrumental variable. Firstly, this IV meets the first condition, that is, it is highly related to environmental tax, because each region formulates environmental tax policy according to its own energy dependence correspondingly. Secondly, this IV meets the second condition that the regional standard coal of energy consumption will not affect firm performance directly. Table 4 shows the regression results of IV-2SLS. Firstly, we use DWH test to test the endogeneity of Tax.
The statistical values of Durbin score and Wu-Hausman F in columns 2-3 of Table 4 are both statistically significant, rejecting the null hypothesis that the variables are exogenous and indicating that variable Tax does have endogeneity. Column 1 of Table 4 reports the first stage regression results, showing that there is a significant negative correlation between the standard coal of energy consumption and environmental tax, which is in line with our expectation that the more a region relies on high energy consumption industries, the lower the intensity of emission levy will be set. The regression results in the first stage prove that the IV is highly correlated with independent variables again. The Kleibergen-Paap rk LM statistics, which are employed to examine whether the selected IV and the endogenous variables are relevant, are all significant at the level of 0.01, indicating that the under-identification hypothesis is rejected. In addition, the Kleibergen-Paap rk Wald F statistics exceed the critical value of the Stock-Yogo test at 10% level, which indicates that the weak instrument variable problem can be eliminated. The regression coefficients of Tax in columns 2-4 of Table 4 are significantly positive, showing that the relationship between environmental tax and firm performance is causality after overcoming the endogenous problem. Our approach to reducing endogenous impact is similar to that of research on related environmental issues. In order to overcome the mutual causality between haze pollution and economic development, Chen and Chen (2018) adopted air flow coefficient and environmental governance level of local government as the IV of haze pollution to study the impact of haze pollution on economic development and then used 2SLS to prove the causal relationship between haze pollution and economic development. Similarly, when studying the impact of environmental policy on the productivity of European manufacturing sectors, Rubashkina et al. (2015) used the average share of environmental policy intensity for eight adverse sectors of the same country excluding the current sector as the IV of environmental policy and also used 2SLS for regression analysis to overcome the negative influence from endogeneity.

Robustness check
In order to further check the reliability of the conclusion, we carry out robustness check from three aspects. The first is the replacement of dependent variable. In order to consider the performance of firms in the capital market, we use TobinQ value as the dependent variable to do the regression again, and the results are shown in column 1 of Table 5. The second one is the replacement of the independent variable. We use the ratio of pollution charge and industrial value added (IVA) as Tax2 for the proxy variable of environmental tax, and the results are shown in columns 2-4 of Table 5. The third one is the replacement of regression model. Considering that missing variables may exist in the industry and region level, we control the fixed effect of industry and region rather than of firm in new regression model, and the results are shown in columns 5-7 of Table 5. In a word, we can see that the regression results in Table 5 are consistent with the main conclusions above, further proving that the conclusions of this paper are robust.

Path analysis
We have already known a robust causal relationship between environmental tax and firm performance after applying the IV-2SLS method and a series of robustness tests. Then, another important question is that what is the path of environmental tax affecting firm performance?
Based on Porter hypothesis, we speculate that firms offset the adverse effect of compliance cost and further improve their performance through the effect of innovation compensation and first-mover advantage. Brandt et al. (2012) and Tombe and Winter (2015) have confirmed our other speculation that environmental tax plays a role in eliminating backward production capacity and optimizing allocation of resources. Therefore, the environmental R&D and market share of firms may play a partial mediating role in the process of environmental tax affecting firm performance. More specifically, corporate environmental R&D can bring innovation compensation and improve firm performance by improving production technology, optimizing production process, increasing energy efficiency and utilizing recyclable wastes; Firms who pay more attention to consumers' green preference can provide environment-friendly products and further occupy new markets, establishing a good corporate image, improving firm performance through firstmover advantage. Figure 4 shows the mechanism of Porter hypothesis.
According to Baron and Kenny (1986), the specific steps to test mediating effect are as follows: (1) Establish the regression between independent variable (Tax) and dependent variable (Performance), and test whether the regression coefficient of the independent variable (Tax) is significant. If not, there is no stable relationship between the two, and the mediating effect is impossible. If it is significant, we proceed to the  (2) Establish the regression of independent variable (Tax) to mediating variables (ERD and MS) in turn. If the coefficients of independent variable (Tax) are not significant, there is no mediating effect. If they are significant, go to the third step.
(3) Establish the regression of independent variable (Tax) and mediating variables (ERD and MS) to dependent variable (Performance). If the coefficients of independent variable (Tax) and mediating variables (ERD and MS) are all significant, there is a partial mediating effect. If the coefficients of the mediating variables (ERD and MS) are significant while the coefficient of the independent variable (Tax) is not significant, there is a complete mediating effect. (4) If at least one of the mediating variable coefficients in the regressions above is not significant, Sobel (1982) test is required. Referring to Baron and Kenny (1986), this paper runs the following mediating effect regression model: where ERD i,t represents corporate environmental R&D and MS i,t represents firm market share. The remaining variables are the same as the baseline model.
We have completed the first step in the mediating effect test above (see the regression results of columns 1-3 in Table 3), so here we start from the second step. There is a significant positive correlation (column 1 in Table 6) between environmental tax and firm environmental R&D, meaning that environmental tax can stimulate firms to increase investment in environmental innovation, which also proves the weak Porter hypothesis. The regression results of the column 2 in Table 6 show that the environmental tax increases the market share of firm. The reason for this result may be that firms can give full play to the first-mover advantage to expand market or seize the new market through providing environment-friendly products, meeting the green preference of some consumers and breaking the original market boundaries. Then, we include environmental tax, environmental R&D and market share into the model at the same time, and the results of columns 3-5 in Table 6 show that the regression coefficients of the three variables are significantly positive, which indicates that environmental R&D and market share play the partial mediating effect. (5) In summary, the conclusions prove that environmental tax can improve firm performance and produce strong Porter effect through two paths, innovation compensation and first-mover advantage. This conclusion proves the existing literature in two aspects: On the one hand, Ouyang et al. (2020) pointed out that the deepening environmental regulation forces the industry to reduce the pollution prevention cost by improving technological innovation capacity, thus creating a "compensation effect". Du and Li (2019) believe that environmental regulation has a "following cost effect" and "innovation compensation effect" in the energy industry, and the "innovation compensation effect" is dominant. On the other hand, Nehrt (1998) proposed that strict environmental regulation can improve the competitiveness of companies in the international market through first-mover advantages, and Wirth et al. (2013) pointed out that firm learning and first-mover advantages are two sources of potential competitive advantage in listed companies.

Conclusion and policy implication
Based on the data of Chinese A-share listed companies from 2008 to 2017, this paper manually collects the environmental R&D data disclosed in corporate social responsibility (CSR) reports, and forecasts and analyzes the impact of environmental tax on firm performance and its transmission path according to the practice of pollution charge in China. According to a series of empirical research results, we believe that the levy of environmental tax can promote firm performance and achieve the strong Porter effect, and this promotion effect is still effective in a long run. After introducing instrumental variables to deal with endogeneity and carrying out a series of robustness tests, we prove that the relationship between environmental tax and firm performance is a robust causality. However, it is worth noting that this paper only considers the firms that going concern, while the firms that are eliminated or shut down are not considered, so the conclusions in this paper may only be effective for the firms that continue to survive on the marktet. Our further study finds that environmental tax can improve firm performance by bringing the effect of innovation compensation and first-mover advantage. Specifically, environmental tax promotes firms to increase environmental R&D investment, bringing innovation compensation effect; environmental tax encourages firms to develop new environment-friendly products and establish green corporate image, bringing first-mover advantage effect. And these two effects play a partial mediating role in the process of environmental tax affecting firm performance. The conclusions of this paper predict the implementation effect of Chinese environmental tax and provide micro evidence from China to verify the strong Porter hypothesis. This paper also focuses on the analysis of the two core paths of Porter hypothesis and interprets the ways to realize the "win-win" situation within environmental protection and firm competitiveness. Based on our findings, we provide the following policy implications: First, our regression results show that there is a positive relationship between environmental tax intensity and firm performance, and we inference this positive relationship is a causality by our further study with IV introduced . Therefore, in order to provoke a strong Porter effect, we suggest governments especially environmental administration department should further push forward the strength of environmental tax (e.g. expanding the scope of environmental tax) and complete the rigidity of law enforcement (e.g. increasing the levy rate on major pollutants) to encourage firms to invest more in environmental R&D and further improve their performance. It is worth noticing that some current literature on environmental management of Chinese local governments has pointed out that state-owned firms and larger local tax payers are often given preferential treatment in environmental law enforcement (Xi 2017; Huang and Lei, 2021). And this unfair phenomenon will make firms that pay attention to environmental protection suffer from greater social costs and even furtherly lead to "bad money drives out good". Bashir et al. (2021c) point out that economic freedom bears a positive impact, and interaction with resource endowment is negative. In a word, we call for that increasing the intensity of environmental tax appropriately and fairly ensuring the implementation and enforcement of environmental tax are two key points for Chinese environmental tax reform in the next step.
Second, by discovering environmental tax enhancing firms' performance through the two mechanism of innovation compensation and first-mover advantage, we argue that although the levy of environmental tax will increase operating cost for firms, it also ignites firms to take green strategy proactively and thus improves their competitiveness. According to our conclusions, we suggest firms should implement forward-looking green strategies because it can benefit firms from the following two ways. One is that the firms that carry out environmental innovation actively can reduce pollution emissions and improve resource utility and productivity so as to take advantage of innovation compensation effect. The other one is that, led by green strategies, the firms that fully consider the consumers' green consumption preference and provide eco-friendly products will break the original market boundaries and establish a positive responsible social image and will further stimulate a firstmover advantage effect to increase market share and enhance the long-term competitive advantage. In other words, when environmental regulations are getting stricter and stricter, firms should realize that complying environmental regulation passively only increases operating costs and decreases the possibility of sustainable development in a long run.
There are some limitations in our research as well, which implies some future research topics. Firstly, some firms reduce their emissions by improving their own technological process flow, which is an endogenous growth. And some firms directly purchase cleaning equipment or technology from outside, but recording this in their own environmental protection R&D. So, it will be interesting to further distinguish these two different environmental R&D and analyze their impacts on the sustainable development of firms respectively. Secondly, market-based environmental regulations mainly include environmental tax, emissions trading, subsidies and deposit return, and we only discussed one of them. Future studies could conduct some comparison among these different kinds of regulations. Lastly, the economic data, corporate financial data, CSR report as well as pollutant emission data in our research could be more detailed and updated.
Funding We are grateful for the financial support from the Western Project of the National Social Foundation of China (19XJL006) and the Chongqing Graduate Research and Innovation Project (CYS20052) and (CYB20054).
Data availability All data included in this study are available upon request by contact with the corresponding author.

Declarations
Ethical approval and consent to participate Not applicable.

Competing interests
The authors declare no competing interests.