Does the carbon emission trading scheme foster the development of enterprises across various industries? An empirical study based on micro data from China

Abstract To achieve the goal of carbon reduction, China has been piloting the Carbon Emission Trading System (CETS) since 2013. As the economy faces a downward trend, it is significant to explore the impact of CETS on business development. There is still debate in academia about whether this policy can boost the level of business development. This paper, based on all A-share data of listed companies in China from 2009 to 2018, uses the Difference in Differences (DID) model to verify the impact of CETS on the input of capital and labor factors and the level of technology in enterprises and discusses the industry heterogeneity of this impact in detail. Placebo tests, propensity score matching, and triple differences ensure the robustness of the conclusions. In further research, this paper decomposes the policy effect of CETS. It regresses the impact of carbon quota prices and carbon market trading scale on business development. The final conclusion is that CETS has a positive impact on the input of capital and labor in enterprises and a negative impact on the level of technology. After distinguishing industries, this conclusion shows differences according to different characteristics of high emissions and low emissions. In addition, the increase in carbon quota prices hinders business development, while the scale of carbon market trading shows an inverted “U” relationship with business development. The article provides meaningful policy references for China and countries in the early stages of CETS construction.


Introduction
Human activities have led to a continuous increase in global greenhouse gas emissions [1].In 2022, the total global carbon emissions exceeded 36.8 billion tons, with China's emissions accounting for a staggering 31.06%. 1 In response to this challenge, China has proposed the "Carbon peaking and Carbon neutrality" goal. 2 Given that the potential for emission reduction in the technological field is shrinking, it is particularly important to improve the carbon emission reduction efficiency of enterprises from the institutional field.Among them, the Carbon Emission Trading Scheme (CETS) is an important way to reduce carbon emissions.It restricts the total amount of regional carbon emissions by allowing companies to trade carbon dioxide emission allowance.While this method stimulates enterprises to reduce carbon emissions, it also poses new challenges to their development.Therefore, a deep analysis of the impact of the carbon emission trading mechanism on enterprise development is of great significance for achieving China's "Dual Carbon" goal and provides a reference for other countries to carry out CETS.
The development of China's CETS can be divided into three stages: The first stage (2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) was mainly involved in international Clean Development Mechanism projects. 3The second stage (2013-2020) carried out CETS pilots in eight provinces and cities including Beijing, Shanghai, Tianjin, Chongqing, Hubei, Guangdong, Shenzhen, and Fujian.The third stage (from June 2021 to the present) established a national carbon emission trading market, first incorporating the power industry, paving the way for other high-emission industries.Since 2013, various pilot carbon emission trading markets have started operations one after another, with the covered carbon emissions only less than the European Union Emission Trading Scheme (EU ETS).In comparison, there are significant differences and similarities between China's CETS and EU ETS.Firstly, the participants in the trade are different.China's CETS has relatively fewer participants, and the industries are relatively concentrated in high-emission industries.Secondly, the richness of trading varieties is different.The trading varieties of China's CETS are relatively fewer, currently mainly focusing on carbon dioxide emission rights.Lastly, the trading methods are different.China's CETS mainly relies on government regulation, with a lower degree of marketization.The development of China's CETS should follow the national conditions based on the successful experience of EU ETS, control the scale of carbon market trading in total, gradually add trading varieties, and increase the proportion of carbon allowance auctions.
As the primary participants in the CETS, enterprises are profoundly affected by this mechanism.Consequently, a significant amount of research has been conducted on the relationship between CETS and enterprise development.From the perspective of the impact mechanism, CETS influences enterprise development by altering their level of technological innovation.A series of studies have analyzed the impact of CETS on enterprises' green innovation capabilities, financial performance, and profitability [2-4].However, there is still debate in academia about the existence and heterogeneity of this impact under different industry and enterprise development characteristics.For instance, research by Zhang and Liu [5] shows that the carbon emission trading mechanism has a heterogeneous impact on the financial performance of enterprises in different industries, a conclusion also reflected in the study by Zhang and Wu [6].We believe that this difference is due to the inconsistent selection of industry samples and development indicators.Another series of studies has further analyzed the impact of the carbon emission trading mechanism on enterprise development from the perspective of carbon allowance prices [7,8].Researchers have found that the emergence of carbon prices directly changes the cost structure of enterprises.However, there is little research on other characteristics of the carbon trading market, such as the vastly different trading volumes in different pilot markets.This provides a good inspiration for our research.
Thus, we pose the following research questions: Can carbon emissions trading mechanisms foster business development?How do these mechanisms manifest differently among companies across various industries?Beyond the price of carbon allowances, are there other factors that influence business development, and if so, what specific forms does this influence take?
Firstly, we select data from Chinese A-share listed companies spanning 2009-2018.We break down enterprise development into two dimensions: traditional factor inputs and technological level.Using the Difference-in-Differences (DiD) method, we explore the impact of CETS on enterprise development, addressing the ongoing academic debate on this issue.Secondly, we refine our conclusions at the industry level.Lastly, we decompose the policy effects of CETS using two indicators: carbon allowance prices and carbon market trading volume.This further investigation into the significant components of CETS on enterprise development enriches our existing conclusions and provides more insightful policy recommendations.
The primary contributions and innovations of this paper are as follows.First, in measuring enterprise development, we considered not only the input of production factors but also the level of technology, thus avoiding the issue of single indicator selection.Second, we further measured the impact of CETS on enterprise development from two dimensions: carbon allowance price and carbon market trading volume.This approach enriches the conclusions of most studies that only use carbon allowance price as the main variable.Third, we found an inverted "U" relationship between the scale of carbon market trading and enterprise development, providing an important reference for subsequent research in this field.Finally, considering that CETS will cover more trading entities and involve more industries in the future, our data covers all industries of A-shares listed in China, rather than focusing only on highemission industries like most studies, fully considering the interrelation of industry development.
The remainder of this paper is structured as follows.Section Literature review reviews the literature in the relevant field and analyzes the theoretical mechanisms of the research topic.Section Research Methods introduces the empirical model and presents the empirical data.Section Empirical research and analysis of results showcases the empirical results of the study.Finally, Section Results and discussion provides an analysis and discussion of the results.

Carbon emissions trading scheme
The externalities of carbon emissions cannot be self-corrected through market behavior and require elimination through government regulation or market adjustments.Coase [9] was the first to propose the use of market mechanisms and the definition of property rights to address this issue.Building on this, Dales [8] introduced the concept of emission trading by incorporating property rights into the field of pollution regulation.Additionally, Montgomery [10] compared the cost differences of various environmental regulatory measures, demonstrating that emission trading has the lowest cost under conditions of perfect competition.In 2002, the Netherlands and the World Bank were the first to implement an emission trading mechanism.Subsequently, the European Union Emissions Trading Scheme (EU ETS) was gradually implemented in different stages for countries at different levels of development, serving as a template for other countries [11].
The Carbon Emission Trading Scheme (CETS) shares the same principles as pollution emission trading and other greenhouse gas emission trading mechanisms.Theoretically, it is a beneficial policy that promotes emission reduction and enterprise development.However, achieving a perfect competition market in reality is challenging due to issues such as asymmetric market information and significant differences in enterprise scale, which could lead to oligopoly or monopoly.The impact of this market-oriented regulatory measure on enterprise development has become a hot topic among scholars.
In general, research on CETS focuses on its characteristics.Firstly, carbon allowances, as the most direct product of CETS, are limited in quantity.Within a compliance period, the government or regulatory authority first determines the total carbon emissions of all regulated units in a specific area and then sets the total carbon allowances [12,13].A reasonable setting of total carbon allowances is a prerequisite for the effective operation of CETS.In practice, the EU primarily determines the total carbon emissions of member countries in the first phase using a bottom-up approach [14], ensuring the successful implementation of the policy.
Secondly, the distribution method of carbon allowances is diverse.To avoid price biases in allowances due to issues like information asymmetry, different carbon emission markets have experimented with various allowance distribution methods, mainly including "grandfathering", baseline method, and auction method.The auction method has been implemented in the EU ETS and several other countries and regions, and is considered to be the future trend for carbon allowance distribution [15,16].
Thirdly, the price of carbon allowances is volatile.The flexible fluctuation of allowance prices, a benefit of the market mechanism, drives the price toward equilibrium [17].Ellerman, Marcantonini [18] introduced the pricing mechanism of the EU ETS.The government sets a floor price for the allowances, but the actual price is determined by market supply and demand, as the distribution and trading of allowances are decided by market participants [19].Fell, Hintermann [20] pointed out that although the price of carbon allowances fluctuates under the free market mechanism, the range of price fluctuation is a significant factor affecting market stability.
Finally, the policy effects of CETS also show heterogeneity due to different scales of the carbon trading market.The effective operation of CETS is based on the flexible flow of carbon allowances among enterprises, requiring participants to conduct transactions smoothly in the market.This depends on the trading scale of the market, i.e. the number of market participants and the trading volume [21].In the EU ETS, the total allowance volume has increased from 210 million tons in 2005 to 407 million tons in 2020.These data indicate that the trading scale of the EU emission trading market is gradually expanding, and the liquidity and policy efficiency of carbon allowances are improving [22].Overall, the inception of the Carbon Emissions Trading Scheme is rooted in the global call for ecological environmental protection.The primary attributes of CETS encompass the total volume of carbon allowances, their allocation methods, prices, and the agility of the carbon trading market.We recognize that CETS, as a mainstream environmental regulatory tool, will affect a company's carbon emissions and economic benefits by altering their production costs, ultimately influencing their development.Hence, it is deemed necessary for us to conduct a qualitative study on the impact of CETS on company development.

Development of enterprises
The development of a company, according to existing research, can be divided into three factors.
The first factor is the resources that a company possesses, which are various production elements, primarily including labor and capital.The second factor is technological aspects, such as a company's production efficiency.This factor determines a company's adaptability to environmental changes and actual production capacity.The third factor is consumer value (market share and customer satisfaction) and the company's own interests (profitability and progress) [23,24].
When the price of a product is exogenously given (in a perfect competition market), profit is a function of production.That is to say, under the market mechanism, a company's profit and its production technology and factor inputs are different manifestations of the same issue.Classical economic theory holds that a company's production mainly comes from labor factors, capital factors, and the input of land and energy, with each factor's output efficiency varying and interrelated.Among them, labor and capital as traditional production factors are indispensable [25,26].Mature companies tend to replace labor with machines, which can improve productivity.Even if the total input cost of the two methods is the same, we believe that automated industrial enterprises have a growth advantage over labor-intensive enterprises [27,28].With the improvement of information technology level in enterprise production, there is a global trend of labor factor share shifting to capital factor share.The decline in the price of investment goods, the shift in labor share, and the improvement in technology level are all profoundly affecting the operating profits of enterprises [29,30].After the wide application of CETS, the price of production factors changes accordingly.Factors are tilting from traditional resources (such as labor and land, etc.) to capital and technology [31].
The traditional input of production factors in enterprises can be represented by many indicators, and how to characterize the level of technology is a current hotspot.In the Solow model, production technology refers more to the efficiency of various non-production factor inputs, such as the progress of the management level is also included [32].In order to improve production efficiency, enterprises must achieve innovative development at all levels of operation.In existing research, most scholars use total factor productivity to represent the above meaning.Higher total factor productivity implies more scientific management and advanced manufacturing technology.Such enterprises are more likely to receive investment and policy support and have strong growth potential [33][34][35][36].

Mechanism of the influence
The role of CETS as a positive driver for enterprise development has been a long-standing debate in academia.Theoretically, market-driven environmental regulation represented by CETS can not only save funds but also promote fairness and environmental justice.This process ensures that enterprises maximize profits while reducing their carbon footprint.The result of market equilibrium is that the marginal emission reduction costs of all enterprises are equal, and the entire society is Pareto optimal.However, regardless of the governance measures adopted, the reduction in carbon dioxide emissions must be achieved by reducing the consumption of traditional energy and improving the efficiency of traditional energy use.In order not to reduce production and lose the original market share, enterprises must improve their production technology.However, the development of new technologies will increase production costs and reduce enterprise efficiency, at least in the short term [37].
The research of Jaffe and Palmer [38] points out that in addition to the changes in the flow of production factors, environmental regulations will squeeze corporate investment.Enterprises with other mature production models will become inefficient due to compliance with new policies.Recent studies have found similar results.Lin and Huang [39] believe that the emission reduction effect of China's CETS comes more from government intervention rather than market mechanisms, which brings more efficiency losses.The research of Zhang and Vigne [40] shows that environmental policies have damaged the total factor productivity, sales ability, and profitability of enterprises, but the negative impact will gradually decrease over time.The research of Wang and Zhang [41] is more detailed, they believe that CETS has damaged the profits of enterprises and reduced market share.Among them, the oil and chemical industries are particularly prominent.
Another perspective, represented by the Porter Hypothesis, suggests that moderate environmental regulation can aid business development by enhancing their innovative capabilities.This is because environmental regulation increases social welfare, and businesses with lower costs receive a "compensation" from policy in the context of improved welfare, leading to an increase in total factor productivity [42].This view is also supported by macro-level research.In a study by Qi et al. [43], they used Chinese provincial panel data and built a DID model, suggesting that CETS reduced carbon emissions without causing economic growth loss.Similarly, Zhang and Zhang [44] added variables such as regional industrial output and the proportion of secondary industry to the DID model, arguing that CETS could reduce carbon emissions while promoting the development of a low-carbon economy.Yang et al. [45] further proposed that this optimization of energy consumption structure was achieved by reducing coal consumption.
On this basis, micro-level research provides theoretical support for our article.For instance, companies with high emissions will face losses and extinction under the market mechanism, while emerging companies that adapt to regulatory measures will have more development opportunities [46][47][48][49].Furthermore, the impact of environmental regulations cannot be generalized, as the systems in different pilot areas are not identical.Compared to other pilot cities, Shenzhen has more auction quotas.Moreover, CETS shows different impacts on businesses of different sizes, different ownerships, and in different regions and industries.For example, large state-owned enterprises in China, with their strong financing capabilities and low financing costs, can quickly upgrade technology to meet environmental regulatory measures [50].
From the perspective of the total factor productivity of businesses, Wu and Wang [7] believe that there is a persistent and significant positive causal relationship between the carbon price and total factor productivity of businesses, which is consistent with Porter's hypothesis.Different from total factor productivity, the stock return rate of listed companies can also reflect the development ability of businesses from a certain perspective.Wen et al. [2] pointed out that CETS has a positive driving effect on stock returns, and the carbon premium in stock returns increases one year after the policy is implemented, but this finding is only based on data from the Shenzhen area and excludes all enterprises in the pilot areas.However, when examining this impact from the perspective of the overall market index, the conclusion is no longer robust.The supplementary research by Wen et al. [51] shows that there is a significant long-term and short-term negative correlation between CETS and the overall stock market.Further research by Guo and Feng [52] shows that the spillover effect of CETS on stock returns is not significant, and it shows an "M" type fluctuation before and after compliance, which means that this fluctuation may be caused by different environmental constraints in the pilot areas.This nonlinear relationship also provides a reference for our further research.Renewable energy power producers, nature reserves, and other enterprises that generate carbon sinks or profit from new energy, obviously benefit from the positive promotion of corporate profits when facing production constraints of environmental regulations, which is equivalent to obtaining additional output of their original products [53,54].
In summary, there is a dispute over the ability of CETS to promote technological improvements in enterprises, but the existence of this impact can be theoretically confirmed.The theoretical mechanism framework of this study is shown in Figure 1.Firstly, to solve the negative externality caused by excessive carbon emissions, CETS provides liquidity for carbon quotas, allowing enterprises to offset external diseconomies by buying and selling quotas under clear property rights.The price of allowances and the scale of the corresponding market can depict the impact of the policy from different angles.According to classical economic theory, price is the main factor affecting supply and demand relationships, while trade scale reflects the real credibility of commodity prices.Assuming that the number of traders is not large, the possibility of control is very high, which means that the trade itself does not reflect value, or that there is a large error between it and the real value.The more people there are, the closer the price is to reality, the more people reach a consensus, the less likely it is to overturn the price in the future [55].
Under the constraint of compliance, the shortterm production cost of enterprises increases, and economic benefits are affected due to the influence of price and scale.Through theoretical analysis, this study believes that maintaining the continuous progress of total factor productivity and expanding the input of enterprise factors is the ultimate way to achieve technological progress and promote enterprise development.On the one hand, technological improvement and efficient management are manifestations of improved enterprise productivity; on the other hand, the increase in factor input is a manifestation of enterprise scale expansion.These two aspects describe the path for enterprises to meet regulatory requirements and move toward Pareto optimality under the background of CETS implementation from different dimensions of enterprise development.It should be clarified that in the theoretical analysis, enterprises produce a single homogeneous product and operate in a completely competitive market.However, enterprises of different industries, different scales, and different stages of development may be affected by policies to different degrees.Their abilities to respond to regulatory measures, obtain information, production methods, financing capabilities, etc. are all different.Therefore, in the face of complex reality, traditional theory cannot simply prove the policy effect of CETS, and more empirical research is needed to determine whether different industries have a consistent performance on the above theoretical impact path.

Research methods
After conducting an in-depth study of related literature and theoretical analysis, we have identified the main topics of this research: Can the Carbon Emission Trading Scheme (CETS) promote enterprise development?How does this effect manifest in enterprises across different industries?Besides the price of carbon quotas, what other factors (such as the scale of carbon market transactions) affect enterprise development, and how do these impacts specifically manifest?Evidently, our research primarily focuses on the measurement and decomposition of the policy effects of CETS.Therefore, we have chosen the widely used Difference in Differences (DID) model as the basic regression method.The DID model estimates policy effects by comparing the average changes in outcomes before and after treatment in the treatment and control groups.To address potential estimation biases in the DID model, we also employed Propensity Score Matching (PSM) and Triple Differences (TDID) for robustness checks.By combining these additional methods, our goal is to ensure the reliability and validity of the basic research results.
In addition, our research also employs the fixed effects regression method.This method is particularly suitable for panel data, i.e. data where we observe the same entity multiple times over time.Fixed effects regression allows us to control for unobserved individual characteristics that may be time-invariant and skew our results.Our research uses panel data, with enterprises as entities.Panel data has several advantages, including controlling for individual heterogeneity, tracking dynamic changes, providing more variability, reducing collinearity between variables, and providing more degrees of freedom.These characteristics of panel data are very consistent with the methods we have chosen, making them an appropriate choice for our research.In the following sections, we will delve into the specifics of our model settings, variable selection, and data sources to provide a comprehensive overview of our research methods.

Models setting
Based on the previous sections, we first build the DID model as follows.
In this model, BD it stands for the development indicator of the i company in the t year.Post t and Treat i are dummy variables that represent the time of the CETS implementation and the experimental group, respectively (when the time is post-2013, Post ¼ 1, otherwise Post ¼ 0; when the company is part of the CETS experimental group, Treat ¼ 1, otherwise Treat ¼ 0).X it represents a set of control variables.k signifies individual fixed effects, l stands for timefixed effects, and e it is the random error term.
In our further investigation into the effects of CETS, we aim to analyze the impact of the price of carbon allowances and carbon market trading volume on enterprise development.To achieve this, we utilize a two-way fixed effects regression model.In this model, the independent variables are the carbon allowance's price and the carbon market trading volume, while the dependent variable is enterprise development.It is important to note that fixed effects analysis of panel data is commonly employed in similar research scenarios [56,57].
Upon examining real data, we have observed significant variation and volatility in the carbon market trading volume across different pilot areas in China.We have also discovered that the trading volume is influenced not only by the emission control intentions of corporations but also by the government's overall emission control limits.Consequently, when constructing our model, we assume a nonlinear relationship between enterprise development and the carbon market trading volume.
Hence, we proceed to establish the afore-mentioned model in the following manner.
In the model, Price it is the variable representing the price at which a given enterprise (i) either buys or sells carbon allowances in a specific year (t).On the other hand, Volume it is the variable that denotes the average daily total volume of trades in the carbon emissions market where the given enterprise (i) operates in a specific year (t), effectively illustrating the scale of carbon market trading.Furthermore, Volume 2 is a variable that signifies the square of the carbon market trading volume.This particular variable is utilized to examine the potential nonlinear relationship between the scale of carbon market trading and the development of the enterprise.As for the other variables in this model, their settings remain consistent with those in model (1), and thus, will not be elaborated upon further in this context.

Dependent variables
To gain a comprehensive understanding of enterprise development, this study focuses on two crucial aspects: traditional factor inputs and technological level.Traditional factor inputs encompass capital and labor, which are measured by a company's fixed asset investment and average annual number of employees, respectively.These indicators have been selected due to their practical significance and their role in the model.They offer a measurable approach to assess a company's capital and labor inputs, thereby reflecting the company's scale and production capacity.
Furthermore, the study also takes into account the technological level of the company, which is represented by the company's total factor productivity (TFP).This indicator considers intangible production factors like technological progress and management efficiency.TFP serves as an indicator of a company's ability to generate more output with the same level of input, indicating an enhancement in technological level and management efficiency.We employ the semi-parametric estimation method proposed by Levinsohn and Petrin [58] to calculate a firm's total factor productivity, which is the ratio of total output to all inputs. 4This method effectively circumvents the data loss that other methods might incur and does away with the assumption that investment is strictly >0 [59].

Independent variables
Firstly, our initial regression model (1) employs the independent variable Post � Treat, which represents the Difference-in-Differences (DID) variable.Secondly, for further investigation by model (2), we have selected carbon quota prices (Price) and carbon market trading volume (Volume) as additional independent variables.The price directly reflects the Carbon Emission Trading Scheme (CETS).Additionally, Volume is included to ensure the validity of the Coase theorem and the assumption of zero transaction costs, thereby representing the residual effects of the CETS policy.

Control variables
To mitigate the potential impact of omitted variables on our regression results, we've incorporated a series of control variables that are pertinent to enterprise development.As enterprises are fundamentally profit-oriented, we've prioritized variables that represent enterprise profitability.The profitability of an enterprise is essentially the valueadded process of its production factors.The higher the return on investment, the stronger the profitability of the enterprise's total assets; conversely, the lower the return on investment, the weaker the profitability.Therefore, we've chosen the Return on Total Assets (ROA) as our first control variable.
Secondly, to realize capital appreciation, an enterprise must first possess capital.The ability to obtain capital is a crucial aspect of enterprise development.The adequacy of internal capital is an important indicator in this regard.Enterprises with sufficient internal capital can often obtain external capital at a lower financing cost and are not subject to financing constraints when adding new production factors.Conversely, the degree of dependence on external funds during the development of an enterprise is also an important indicator.Enterprises with a low debt-to-asset ratio can often obtain external funds at a lower cost.Therefore, we've chosen Cash Flow and the Debtto-Asset Ratio as indicators to characterize the ability of enterprises to obtain capital from the perspectives of endogenous funds and external financing capabilities.These are our second and third control variables.
Thirdly, after obtaining capital, scientific management is a necessary condition for realizing capital appreciation.The level of financial work handling by listed companies directly impacts investors' evaluation of the company's operational capabilities.Additionally, organizational and operational capabilities are important factors that influence production efficiency within a company.Therefore, we've chosen the ratio of Financial Expenses to Debt Balance and the ratio of Management Input to Total Operating Income to represent financial and management capabilities.These are our fourth and fifth control variables.
Lastly, the growth stage and size of the enterprise are factors that many related studies consider.Larger enterprises often have better financing and management capabilities and are more likely to receive government support.Relatively mature enterprises are more likely to achieve a high market share within the industry.Therefore, we've chosen the Total Asset Level and Enterprise Age to represent the size and growth stage of the enterprise [43,56,57].
We have organized a schematic table of all the above variables as follows (Table 1).

Source of data
We conducted a research study focusing on Ashare listed companies in China from 2009 to 2019. 5 To ensure the accuracy of our findings, we excluded data from the years after 2020 to eliminate any abnormal data caused by the pandemic.Additionally, we removed companies that were listed after 2013, as China's CETS was only implemented in 2013.To maintain the authenticity and reliability of our data, we excluded samples with the status of ST and � ST.We also performed a 1% tail treatment on the research variables to eliminate any outliers.After all these adjustments, our full sample consists of 2132 listed companies.In Table 2, we present the comprehensive data of the  Our sample encompasses 19,139 valid observations.The variation in the observation period across different companies is due to their respective years of market entry, meaning not all companies have the same length of the observation period.We have gathered data from eight carbon exchanges located in Beijing, Shenzhen, Shanghai, Guangdong, Fujian, Chongqing, Tianjin, and Hubei, spanning the years 2013-2019.It's worth noting that Chongqing, Tianjin, and Hubei started disclosing data in 2014, while Fujian began in 2017.All the data were procured from the WIND and CSMAR databases.The descriptive statistics related to the collected data are presented in the following paragraph (Table 3).

Basic research-policy effects of CETS
To ensure the validity and reliability of the Difference-in-Differences (DID) analysis, it is crucial to conduct a parallel trend test, with the dynamic effect parallel trend test being the prevalent method [60,61].For our study, we selected 79 pilot enterprises from the CETS pilot areas as the treatment group and 586 non-pilot enterprises from the same areas as the control group.The grey box in Figure 2 represents the 90% confidence interval.It can be observed that the regression coefficient of the interaction term between time and pilot enterprises before 2013 is not significantly different from zero.This suggests that there was no significant difference between the experimental group and the control group before the policy was implemented, indicating that the parallel trend test had been successfully passed.The detailed test results are as follows.
We use model (1) to verify the policy effects of CETS.The empirical results are shown in Table 4.
The results show that CETS has had a significant impact on enterprises' capital investment, labor input, and technological level, with varying directions of influence.Specifically, CETS has had a positive effect on the input of traditional production factors in enterprises.This is because enterprises need to invest more capital to purchase environmental protection equipment, and they also need to invest more personnel to operate and maintain these devices.However, CETS has had a negative impact on the technological level of enterprises.This is due to the fact that China's CETS is still in its early stages of development, the system is not yet perfect, and the price of carbon allowances is relatively low, which has resulted in a lack of motivation for technological innovation in enterprises.
The above results are for all industries in the pilot areas, and we believe that CETS will have heterogeneous impacts across industries.Existing research  often focuses on high-emission industries [5,31], as the Chinese government's document "Carbon Emission Trading Management Measures (Trial)" states that the national CETS will cover steel, power, aviation, chemical, construction, petrochemical, nonferrous metals, and paper industries.However, in the pilot phase of CETS, many companies in the list of controlled emission companies published by various places belong to non-high emission industries, such as discretionary consumer services in the retail and real estate categories.We believe that industries are interconnected, and the evaluation of a policy's effects should not only focus on its key control areas but also consider its impact on other industries.Based on this idea and the availability of data, we finally selected seven major industries in the pilot areas for heterogeneity analysis. 6The regression results are shown in Table 5.
Overall, the implementation of CETS has had an impact on major industries.Firstly, it has led to a significant increase in capital input in the transportation, manufacturing, and real estate industries; while it has caused a significant decrease in capital input in the information and technology services, energy production, leasing and business services, and mining industries.Secondly, CETS has increased labor input in the transportation, manufacturing, real estate, and wholesale and retail industries; while it has decreased labor input in the energy production, leasing and business services, and mining industries.Lastly, CETS has improved the technological level in the transportation and manufacturing industries; while it has decreased the technological level in the real estate, energy production, leasing and business services, and mining industries.
For the transportation, manufacturing, and real estate industries, the implementation of CETS has increased both capital and labor input, and also improved their technological level.This might be because these industries need more capital and labor input to improve production technology and processes to meet emission reduction requirements under the CETS policy.
For the information and technology services, energy production, leasing and business services, and mining industries, the implementation of CETS has led to a decrease in both capital and labor input, and also a decrease in their technological level.This might be due to various reasons (such as financial constraints, technological limitations, etc.) that prevent these industries from introducing more advanced production technologies, thus leading to a decrease in technological level.
For the wholesale and retail industry, the implementation of CETS has increased labor input.This might be because this industry needs more labor to improve production processes to meet emission reduction requirements under the CETS policy.

Robustness test
To enhance the credibility of our conclusions, we employ three methods for robustness checks: placebo tests, Propensity Score Matching-Difference  in Differences (PSM-DID), and triple differences.We draw on the research methods of Qi et al. [43], and Ren et al. [59], using placebo tests and PSM-DID to verify our original conclusions.Additionally, we take into account that while CETS was being implemented, China was also piloting other economic policies, such as VAT reform and supply-side reform.The development of enterprises would also be influenced by these policies.To ensure the robustness of our conclusions, we conduct a triple differences robustness analysis based on the double differences.

Placebo test
The placebo test is a method used to verify the validity of the research design and the robustness of the results.By altering the treatment time or treatment group, the placebo test can help us examine whether the research results depend on the choice of treatment time and group.We changed the policy time to 2011 and swapped the treatment and control groups.In this fictitious policy time point and policy grouping, we hope to see that the results of the DID regression are no longer significant (Table 6).
In the placebo test, we found that the variables related to enterprise development became insignificant.This result was significant in the basic regression shown in Table 4, and more than half of the industries were significantly affected by CETS as shown in the industry heterogeneity analysis in Table 5.This result indicates that our basic regression results are not dependent on the choice of time and group, and the results are robust.

Propensity Score Matching-Difference in Differences test
Next, we further apply PSM-DID for robustness testing.The Propensity Score Matching-Difference in Differences (PSM-DID) method is a powerful tool for robustness testing, combining the advantages of Propensity Score Matching and Difference in Differences.It first uses PSM to balance the observed characteristics between the treatment group and the control group, and then uses DID to control for possible fixed effects between the treatment group and the control group.The main goal of this method is to obtain an unbiased estimate of the treatment effect by simulating the conditions of a random experiment, which can effectively reduce selection bias and improve the accuracy of the estimation results.
In this section, we select the full sample, retaining 79 companies in the experimental group, while expanding the control group to 2053 companies.Since the number of control groups in the sample is much larger than the experimental group, to ensure the integrity of the sample data, this paper selects the control variables in the benchmark regression, applies the k-nearest neighbor matching method, and sets k ¼ 20 (i.e.matching 1 in the experimental group with 20 in the control group).The control variables in the base regression and the affiliation of the enterprise's prefecture-level city are selected as covariates.
After propensity score matching is completed, the sample data loses 40 control group firms and retains 98% of the data.The results of further double-differenced regressions are shown in Table 7, and the regression results for each industry are generally consistent with the previous results, suggesting that our conclusions are robust.

Triple difference
Theoretically, the advent of CETS has the greatest impact on the manufacturing industry, as this sector emits more carbon dioxide.To eliminate the influence of other policies on other industries in the sample during the same period, this section  conducts a triple difference robustness verification.This practice is referenced from the research of Ren et al. [59], and the model is constructed as follows based on this method.
Compared to the double difference, we have added a "group" dimension here, limiting the industries participating in the policy to the key emission control industries stipulated by the Chinese government, to reduce the impact of other environmental regulatory policies.We record the CETS key impact enterprises in the transportation, manufacturing, and energy production industries, which have the most significant regression results, and set a new variable "DDD".The regression results using Equation (3) are as follows (Table 8).
As can be seen, after controlling for the dimension of the industries mainly affected by CETS, CETS significantly drives the input of capital and labor factors in the main affected industries and accelerates the technological upgrading of these industries.The results of the triple difference test are consistent with the benchmark regression results, and our conclusions can be considered robust.To further summarize the impact of CETS on enterprise development, we will decompose the policy effects for further study in the next section.

Further research-the impact of carbon allowance prices and the volume of carbon trading market
By utilizing the DID model to estimate the effects of the CETS policy, we believe that CETS has a significant impact on the development of businesses, but this is only a general impact.In the theoretical analysis, we have already mentioned the four main characteristics of CETS, with the carbon quota price being the most direct manifestation of CETS, and the scale of the carbon market trading is also one of the main factors.How they affect the development of businesses is a question we are concerned about.Only by refining the policy effects of CETS can we propose more targeted policy improvement suggestions.In this section, we use Equation (2), taking the business development indicators as dependent variables, and performing individual fixed effect regression. 7The results are shown in Table 9.
The results show that the price of carbon quotas has a negative impact on the capital, labor input, and technological level of enterprises, while the scale of the carbon trading market has a positive impact.These results provide important insights for understanding the impact of CETS on enterprise development.The effective operation of CETS is based on the flexible flow of carbon allowances among enterprises, which requires participants to be able to smoothly trade in the market.This depends on the scale of the market, that is, the number of market participants and the volume of transactions [21].However, during the pilot phase, there are huge differences in different carbon trading markets, and the scale of carbon market transactions also shows strong fluctuations.The research of Guo and Feng [52] verified the "M" type impact of CETS on enterprise development.Combined with previous theoretical analysis and our conjecture, "the huge difference in the scale of carbon market transactions may have a nonlinear impact on enterprise development".We believe that adding the square term of Volume will make the conclusion more comprehensive (Table 10).
Upon incorporating the square of Volume into the regression, some changes in the results are observed.Firstly, the impact of carbon quota prices on the capital input of enterprises has shifted from significantly negative to insignificant, while its impact on labor input and technological level remains significantly negative.Secondly, the impact of the scale of the carbon trading market on the capital, labor input, and technological level of enterprises has all turned significantly positive.Lastly, the square of Volume has a significantly negative impact on the capital, labor input, and technological level of enterprises.These results suggest that there indeed exists a nonlinear relationship between the scale of the carbon trading market and the development of enterprises.When the market scale is small, increasing the market scale can significantly enhance the capital, labor input, and technological level of enterprises.However, when the market scale reaches a certain level, further increasing the market scale may lead to a decline in the capital, labor input, and technological level of enterprises.This might be due to the fact that an overly large market scale could increase market instability, thereby negatively affecting the development of enterprises.
In addition, the shift in the impact of carbon quota prices on the capital input of enterprises from significantly negative to insignificant might be because when the market scale is larger, enterprises have more opportunities to optimize the use of their carbon quotas through market transactions, thereby alleviating the pressure on enterprise capital input due to rising carbon quota prices.A simple illustration of this inverted "U" relationship is as follows (Figure 3).

Results and discussion
To promote decarbonization, the mechanism of Carbon Emissions Trading Scheme (CETS) is being widely used by various countries.The European Union Emissions Trading System (EU ETS) is the earliest and most mature carbon trading system in the world.China launched the CETS pilot in 2013 and established a national carbon trading market in 2021.
Firstly, we used the data of Chinese A-share listed companies from 2009 to 2018 to construct a Difference-in-Differences (DID) model to verify the policy effect of CETS on enterprise development.We also explored the industry heterogeneity of the impact of CETS on enterprise development.
In further research, we decomposed CETS into carbon quota prices and carbon market trading volume, and discussed the impact of CETS on enterprise development at different levels.This provides a reference for improving the national carbon trading market.

Review and interpretation of the results
Based on our empirical analysis, we have drawn the following research conclusions: Firstly, the implementation of CETS has had a significant positive impact on the capital and labor inputs of enterprises.This indicates that the introduction of the policy has encouraged enterprises to increase their investment in human and material capital.The underlying reason is that enterprises need to adapt to the new regulatory environment, improve production processes, and meet the emission reduction targets set by CETS.This involves investing in low-carbon production equipment, employee training, and other emission reduction strategy needs.However, the policy has had a significant negative impact on the technological level of enterprises.This may seem counterintuitive at first glance, as one might expect the introduction of CETS to encourage enterprises to adopt cleaner and more advanced technologies.However, this negative impact can be attributed to several factors.For example, the cost of adopting new technologies may be too high for some enterprises, especially in the short term.In addition, the availability and accessibility of clean technologies may be limited, and there may be significant adoption barriers, such as a lack of technical knowledge, market uncertainty, and regulatory complexity.These findings highlight the complex and multifaceted impact of CETS on enterprise development, emphasizing the need for careful consideration and strategic planning when implementing such policies.
Secondly, in the study of industry heterogeneity, we found that the impact of CETS varies significantly among different industries.For high-emission industries, such as transportation, manufacturing, and real estate, the implementation of CETS has significantly increased the capital and labor inputs of these industries, while also improving their technological level.This is because these industries need to invest more capital and labor to improve production technologies and processes to meet emission reduction requirements when dealing with CETS policies.This result indicates that the impact of CETS policies on high-emission industries mainly manifests as encouraging these industries to upgrade technology and improve production processes to adapt to stricter carbon emission restrictions.
For low-emission industries, such as information and technology services, energy production (only thermal power generation in the energy production industry belongs to high emissions, other new energy companies are low emissions), leasing and business services, and mining, the implementation of CETS has led to a significant decrease in the capital and labor inputs of these industries, and their technological level has also declined.This is because these industries, due to various reasons (such as financing constraints, technological barriers, etc.), cannot introduce more advanced production technologies when dealing with CETS policies, leading to a decline in technological level.This result indicates that the impact of CETS policies on low-emission industries mainly manifests as pressure on these industries' production inputs and technological levels, requiring further policy support and technological guidance.The above conclusions also fully reflect the different characteristics between industries.For example, high-emission industries are often resource-oriented, and the cost of technological upgrading for these enterprises is relatively low, while the cost of technological upgrading for low-emission industries is relatively high.This makes the performance of CETS different among these industries.
Thirdly, in further research, we analyzed the impact on enterprises from two dimensions: carbon quota price and carbon market trading volume.When considering the carbon quota price and carbon market trading volume, we found that the carbon quota price has a negative impact on the capital, labor input, and technological level of enterprises.This is because the rise in the carbon quota price increases the production costs of enterprises, thereby compressing the space for enterprise investment and technological upgrading.The impact of the carbon market trading volume on enterprises is positive.This is because the expansion of market size improves the liquidity of carbon quotas, allowing enterprises to adjust their production strategies more flexibly, thereby promoting enterprise development.
Then, when we further consider the nonlinear effect of the carbon market trading volume, we find that the impact of the carbon quota price on the capital input of enterprises has changed from significantly negative to insignificant.This may be because when the market size is larger, enterprises have more opportunities to optimize the use of their carbon quotas through market transactions, thereby alleviating the pressure of the rise in carbon quota prices on enterprise capital input.The impact of the carbon market trading volume on enterprises remains positive, but the impact of its square term is negative.This indicates that there is a nonlinear relationship between the impact of the carbon market trading volume on enterprise development.That is, the expansion of the market size within a certain range can promote enterprise development, but when the market size is too large, it may bring about market instability, thereby having a negative impact on enterprise development.
Finally, the interesting conclusion we found is that the carbon market trading volume and enterprise development present an inverted "U" relationship.In the early stage of the formation of the carbon market, the expansion of the trading volume means that the market activity is continuously improving, and the number of participating entities is constantly increasing.Based on the expectation of the future market and the recognition of the lag in technological investment, enterprises will increase their investments in the early stage of the implementation of CETS, which will bring beneficial effects to enterprise development.However, when the market trading volume reaches an extreme value, as the market size continues to expand, the investment motivation of enterprises weakens.The change in corporate investment behavior will inhibit enterprise development.Therefore, in the process of promoting the national carbon trading market, the market access threshold should be reasonably set to ensure the orderly operation of the national carbon trading market.
In our research conclusion, the impact of CETS on enterprise development is similar to the views of scholars such as Zou and Zhong [31] and differs from the views of scholars such as Wu and Wang [7].However, in terms of the existence of the policy effect of CETS, this article agrees with the views of other scholars, and we also believe that CETS has significantly affected enterprise development.The shortcomings of our research are as follows.First, our data only includes listed companies, which are larger in scale, and we have not considered small and micro enterprises, and there is still room for improvement in sample selection.Second, due to the availability of data, we did not consider the impact of the total carbon quota accounting method in the market, the proportion of carbon quota auctions, and the intensity of supervision.Third, the carbon market trading volume has a nonlinear impact on enterprise development and has a maximum value.Although this is an interesting result, this article does not delve into the cause of this phenomenon, which may be caused by the interaction between the carbon quota price and the carbon market trading volume.We will address this in future research.

Comparison with related studies
In our research, the impact of the carbon quota price and the impact of the carbon market trading volume showed opposite directions in the first few years of the policy.This is because enterprises are first affected by price changes, while market size represents the degree of market activity and has a lag.According to Coase's theorem, the larger the market size, the wider the policy, the more industries covered, and the market tends to be a "perfectly competitive market".At this time, relevant enterprises are more inclined to develop advanced technology.Without distinguishing industry differences, CETS indeed reduced the overall technology level of enterprises in the pilot areas, and the industry has not been distinguished for the time being [40].
Zhang et al.
[3] conducted a study on the eight main industries covered by CETS, and they believe that CETS has no significant impact on corporate technological innovation overall.In this paper, we found that CETS has a clear positive impact on the development of enterprises in the manufacturing, transportation, and energy production industries, and overall, CETS has a negative impact on the technology level of the entire industry.The difference in this conclusion may be multifaceted.We chose total factor productivity to represent the level of technology, not the innovation indicators of enterprises.In addition, within the pilot sample, the coverage of the industry is more than the eight main industries, and studying the industry within the policy alone seems unable to explain the correlation between industries.Our conclusion can also be supported by similar studies.For example, Huang and Chen [34] pointed out that the improvement of the green total factor productivity of enterprises by CETS must rely on the spatial siphon effect to achieve.The study by Liu et al. [36] also shows that in the agricultural field, the pull of CETS on total factor productivity presents an inverted "U" shape, and most regions are still in the stage of negative impact.
Furthermore, in studies on the impact on other aspects of enterprises, Zhang and Liu [5] pointed out that the impact of CETS on corporate financial performance is lagging.This view is similar to the nonlinear relationship we found between enterprise development and carbon quota prices and carbon market trading volume.However, the research conclusion of Wu and Wang [7] believes that the impact of CETS on enterprises is continuous.This difference is caused by different research methods and also differences in sample selection.We found that in their study, the explanatory variables are the cross products of carbon prices, policy dummy variables, and time point dummy variables, that is, different carbon prices are used to adjust policy effects.This paper believes that carbon prices are affected differently by the trading volume of different pilot markets and cannot alone represent the policy effect of CETS.In our empirical research results on the scale of the carbon market, they often contradict the results related to carbon prices.
The activity level of a regional carbon market can be reflected by its trading volume, and a more active market is more likely to have a scientific management mode and advanced environmental protection concepts.Combining the empirical conclusions of the extreme points, we believe that the total factor productivity of most enterprises is still negatively affected by the trading volume and has not reached the extreme point.Only when the trading volume crosses the inflection point will this situation be reversed.This means that as the market scale expands to a certain extent, more and more enterprises are affected by policies, and enterprises are more inclined to develop advanced technologies, rather than passively accepting the economic losses brought about by environmental regulations.After the development of technology, more advanced enterprises will increase equipment investment and expand production scale.At the same time, the management level will also improve with the popularization of policies, and labor will increase with the expansion of enterprise scale.This finding is different from the linear effect found in most literature, and this difference can be attributed to the fact that most previous studies only examined the impact of quota prices [39,51] and ignored the activity level of the carbon market.
In summary, our research has both similarities and innovations compared to existing literature.In terms of industry differentiation, we believe it should be comprehensive.In terms of the decomposition of CETS, we believe it should be multi-layered.But this is just the beginning of our research on this issue.There will always be debates in the academic world, and this is exactly the motivation for our continued research.

Suggestions and outlook
At the end of the article, we hope to propose targeted policy suggestions through the comparison of CETS and EU ETS, as well as our empirical conclusions.The CETS and the EU ETS are the two largest carbon emissions trading markets in the world, and they have both similarities and differences.Firstly, the EU ETS is the earliest established carbon emissions trading market in the world, which has been in operation since 2005 and has gone through multiple trading phases, with relatively mature market rules.On the other hand, China's CETS started as a pilot in 2013, and it wasn't until 2021 that a national carbon trading market was officially established, which is in the stage of development and improvement.Secondly, the EU ETS covers most of the important industries in all EU member countries, including power, heat, petrochemical, etc. Thirdly, China's CETS currently mainly covers the power industry but plans to gradually expand to include multiple key industries such as steel, chemical, and building materials.The EU ETS adopts a combination of auction and free allocation in the distribution of emission rights, while China's CETS currently mainly adopts the method of free allocation, but also plans to gradually introduce the auction mechanism.Fourthly, the carbon price of the EU ETS is determined by the market supply and demand relationship, while the carbon price of China's CETS is also determined by the market, but the government sometimes intervenes to maintain price stability.To improve China's CETS, it is necessary to refer to the successful experience of the EU ETS.Combining the research conclusions of this article, we propose the following policy suggestions for China and similar countries that are in the early stage of establishing CETS 1. Gradually expanding the coverage of the carbon trading market: Currently, China's CETS mainly covers the power sector.However, to reduce carbon emissions more effectively, it should be gradually expanded to include several key industries, including iron and steel, chemicals, and building materials.2. Introducing an auction mechanism: At present, China's CETS mainly adopts a free allocation method, but this may lead to inefficient allocation of carbon emission rights.An auction mechanism should be gradually introduced to better utilize the market mechanism to allocate resources.
3. Establishment of a stable carbon price mechanism: The stability of carbon prices is crucial to the investment decisions of enterprises.The Government should minimize its intervention in carbon prices and allow market supply and demand to dominate price formation.4. Provide technical support and financial assistance: In view of the negative impact of the CETS policy on the technological level of enterprises, the Government should provide technical support and financial assistance to help enterprises upgrade their technology, especially for those industries that face financial and technological constraints in responding to the CETS policy.5. Enhance the policy design of industry differentiation: As there are significant differences in the impacts of CETS on different industries, the policy design should take full account of industry differentiation.For high-emission industries, technological upgrading and improvement of production processes can be encouraged through the provision of more technological and financial support; for lowemission industries, they can be assisted through the provision of more policy support and technological guidance to help these industries to For low-emission industries, more policy support and technical guidance can be provided to help these industries cope with the pressure brought by the CETS policy.6. Optimize carbon market design: In view of the impact of carbon quota price and carbon market trading scale on the development of enterprises, the government should optimize the design of the carbon market, for example, it can ensure the stable operation of the carbon market by setting a reasonable price of carbon quota; at the same time, it should control the trading scale of the carbon market to avoid the instability of the market caused by the excessive market scale.7. Strengthening international cooperation: China should actively participate in the construction of the global carbon market and trade carbon emission rights with other countries and regions to realize the optimal allocation of global carbon emissions.
While our study provides some valuable insights, there are still many questions that deserve further exploration.For example, how does CETS affect firms' innovative behavior and competitiveness, and does the implementation of CETS lead to the geographic relocation of firms?These questions need to be further explored in our future research.
Finally, we hope that our study will provide valuable references for policy makers to better understand and respond to the challenges of CETS to achieve China's carbon neutrality goals.At the same time, we also hope that our study will inspire more scholars to conduct in-depth research on this important topic, to push our understanding and application of CETS to new heights.

Notes
1. Data compiled from "Carbon Monitor", URL: https://www.carbonmonitor.org.cn/.2. "Carbon peaking" refers to the process of carbon dioxide emissions reaching a historical high and then entering a continuous decline."Carbon neutrality" means that enterprises or individuals offset their own carbon dioxide emissions within a certain period of time by planting trees, saving energy and reducing emissions.China has promised to peak its carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060.3. The cooperation mechanism established by Article 12 of the Kyoto Protocol states that developed countries cooperate with developing countries at the project level by providing funding and technology, and that greenhouse gas emission reductions achieved through the projects can be used by developed country Parties to meet their commitments under Article 3 of the Kyoto Protocol.4. Specifically, during the calculation, the capital input variable is represented by the net value of fixed assets, the labor input is represented by employee compensation, and the intermediate input is the sum of operating costs, sales expenses, management expenses, and financial expenses, subtracting depreciation, amortization, and employee compensation.5. To eliminate anomalous data during the epidemic, we will exclude years after 2020 from the study sample.6.The industry classification guidelines are based on the industry classification published by the China Securities Regulatory Commission.7. Since the carbon price and the scale of the carbon market trading are only recorded after the policy is implemented, the number of years for data collection is reduced.Therefore, in this section, we no longer consider the fixed effects of time.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Figure 1 .
Figure 1.Mechanisms of influence at the theoretical level.

Figure 3 .
Figure 3. Extreme values of enterprise development under the influence of volume.

Table 2 . Full sample-descriptive statistics table.
sample of companies, including the indicators of total factor productivity.This table provides descriptive statistics that give an overview of the corporate data within the full sample. full

Table 7 . Robustness tests-propensity score matching-difference in differences.
VariablesTransport and postal services Info and Tech services Manufacturing Real estate Wholesale and retail Energy production Note: Robust t-statistics in parentheses ��� p < 0.01, �� p < 0.05, � p < 0.1.