Abstract
Introduction:
To advance low‐carbon transformation and sustainable growth, low‐carbon city pilot policy (LCP) was first implemented in China in 2010, and the scope of pilot cities has been gradually expanded in the subsequent years. Manufacturing enterprises, as microscopic subjects of the national economy, are leading force in building a new engine of green growth.
Methods:
Taking LCP as a quasi‐natural experiment, this study leverages a multi‐temporal difference‐in‐difference model to investigate how this policy affects green total factor productivity (GTFP) in the manufacturing sector and the corresponding potential mechanisms, where the research is grounded in panel data of Chinese manufacturing enterprises listed on the A‐share markets in Shanghai and Shenzhen (2005–2024).
Results:
It is found in this study that LCP exerts a positive effect on manufacturing enterprises' GTFP, with the conclusion holds steady following multiple robustness checks. State-owned and highly polluting manufacturing enterprises experience a more pronounced boost in their GTFP. Through mechanism test, it is demonstrated that the policy drives the improvement of GTFP chiefly via advancing green innovation and increasing green output.
Discussion:
These findings offer empirical evidence for the effectiveness of LCP in promoting green growth among manufacturing enterprises, with implications for extending similar policies to other regions and enterprise types.
1 Introduction
With global warming, countries all across the world now agree that low-carbon transition is necessary. Since 2006, China ranks as the global top carbon dioxide producer. As a practitioner for green growth, China has outlined the dual carbon goals, which demonstrating its commitment as a major country to actively respond to climate change. A over 34% fall in China’s carbon emission intensity from 2013–2023, yielding distinct outcomes in pollution and carbon reduction. Focusing on “carbon reduction and carbon abatement”, the National Development and Reform Commission (NDRC) initiated the low-carbon city pilot policy (LCP) in 2010, and the pilot scope was then progressively expanded in 2012 and 2017. The Recommendations for Formulating the 15th Five-Year Plan put forward that manufacturing sector’s digital and intelligent transformation should be advanced, the development of intelligent, green and service-oriented manufacturing boosted, and the transformation of industrial models and enterprise organizational forms accelerated. Manufacturing enterprises, as a major player in carbon emission reduction, are not only the key main body to promote green manufacturing, but also an important driving force for construction of low-carbon cities. Enhancing the GTFP of manufacturing enterprises, reducing carbon emissions, and forming the coupling synergy of environment and economy are of great significance for realizing China’s “dual-carbon” goal and green and sustainable high-quality development.
As an important environmental regulatory instrument, LCP has attracted extensive scholarly attention. Existing studies have assessed its implementation effects from multiple perspectives, including pollution reduction, economic development, technological innovation, employment, and social welfare. In terms of environmental performance, LCP significantly reduces carbon emissions, and the environmental benefits generated by the policy exceed the government’s investment costs in environmental governance (Li et al., 2024). In terms of economic outcomes, LCP has been found to help attract foreign capital, accelerate the upgrading of industrial structures, and support the high-quality development of enterprises (Pan et al., 2024; Zheng et al., 2021; Dong et al., 2024). With respect to technological innovation, the policy can encourage green innovation at the city level and may also produce spillover effects in surrounding areas (Pan and Zhao, 2024; Peng et al., 2025). Beyond these economic and innovation-related effects, recent studies have also explored its welfare implications, showing that LCP may contribute to employment expansion, improvements in residents’ wellbeing, and the promotion of common prosperity (Fu et al., 2024; Wang et al., 2024).
A second group of studies relevant to this paper concerns TFP and GTFP. TFP measures production efficiency by capturing the amount of output produced from a given set of input factors. Prior research indicates that moderate environmental regulation can stimulate patent applications and raise firm-level TFP (Martínez-Zarzoso et al., 2019). Chen et al. (2021), using approaches such as PSM-DID, also show that LCP contributes to improvements in enterprise TFP. Compared with traditional TFP, GTFP places greater emphasis on environmental constraints and resource efficiency, reflecting productivity performance under the requirements of green transformation and sustainable development. More recently, researchers have paid increasing attention to the relationship between LCP and GTFP. Nevertheless, existing evidence is mainly based on city- or industry-level analyses (Yuan et al., 2025; Wang et al., 2023), suggesting that LCP can promote urban GTFP primarily by reducing energy consumption and encouraging technological innovation.
The above literature provides a useful foundation for this study, but several issues remain insufficiently addressed. First, existing research has mainly evaluated the effects of LCP from a macro perspective, with emphasis on pollution reduction, economic performance, and technological innovation. However, less is known about how firms respond to this policy at the micro level, particularly manufacturing firms, which constitute an important source of carbon emissions. Second, studies on GTFP are still largely concentrated at the city or industry level, and firm-level evidence on the relationship between LCP and GTFP remains relatively scarce. This makes it difficult to explain how city-level policy shocks are transformed into changes in green production efficiency through firms’ behavioral adjustments. Third, although the overall effectiveness of LCP has been widely discussed, its transmission channels at the enterprise level and its differentiated impacts across firms with different characteristics have not been fully explored. Therefore, whether and how LCP improves the GTFP of manufacturing firms still requires further empirical examination. Against this background, this study investigates the effect of LCP on manufacturing firms’ GTFP and explores the mechanisms behind this effect from a micro-level perspective.
The following elements represent this paper’s potential marginal contributions: first, it offers fresh research viewpoints from the micro viewpoint. Previous study has not thoroughly examined firms, instead, it has concentrated on the city and industry. This paper investigates LCP’s influence on the GTFP of manufacturing firms by using firm-level data, providing key micro evidence to assess the policy’s actual effect. Second, it utilizes a multi-temporal DID model. Considering that the policy was initially piloted, then promoted, and implemented in batches across cities, it is not feasible to designate a single specific year as the node for its concrete implementation. In view of the flexibility in the policy’s implementation and the dynamic interventions received by different cities, it uses a multi-temporal DID model. Thirdly, it explores the role mechanism of LCP on the improvement of GTFP in manufacturing industry from green innovation and green output. Understanding the impact and role mechanism can provide theoretical support and practical guidance for formulating more targeted and effective policies in the future. It can help policymakers to adjust and optimize the policy content, improve policy implementation methods, and better promote the green transformation and development of the manufacturing industry. The research framework of this study is shown in Figure 1.
FIGURE 1
2 Theoretical analysis and research hypotheses
2.1 LCP and GTFP of enterprises
The LCP is intended to implement an economic development mode with low energy consumption, low pollution, low emission, while building a resource-conserving, environment-oriented society marked by low-carbon living philosophies and practices. Porter and Linde (1995) think proper environmental regulation can spur enterprises to engage in environmental protection technology, and green technology innovation will curbs pollution and eases cost pressures from environmental regulation through productivity improvement, which is Porter’s hypothesis. As an effective means of solving environmental problems, environmental regulation can show great benefits in reducing environmental pollution, encouraging industrial modernization and technological innovation (Zhou et al., 2024), and after implementing environmental objectives and constraints, the government can raise the regional GTFP by facilitating technological creativity and upgrading economies of scale (Tian and Feng, 2022). The essence of GTFP improvement is to optimize financial gains and minimize environmental pollution with minimum resource input. As one of the means of comprehensive environmental regulation, LCP functions through two approaches: incentives and constraints. A wide spectrum of targeted incentive policies will be enacted by government across pilot cities, including financial subsidies, tax exemptions, lowering the threshold of financing, etc., to alleviate the strain on enterprises’ operations and production, and to guide enterprises in conducting green transformation, adopt clean, green, and high-efficiency production methods, reduce carbon emissions, and thus enhance the GTFP. Meanwhile, government will also take punitive measures, such as: interviews, fines, etc. To force enterprises to carry out pollution control, specifically, enterprises will give up the production of high-energy-consuming products, and shift to producing clean products, in addition, enterprises will increase the investment in pollution control, and the GTFP is subsequently increased when double clean control is used during the production process and at the end of production (Jiang et al., 2024). Therefore, from two perspectives, under the dual role of incentives and penalties, enterprises will focus more on reducing pollution, strengthen the comprehensive utilization of resources, and improve GTFP.
Based on this analysis, the paper proposes:
Hypothesis 1LCP can improve manufacturing enterprises’ GTFP.
2.2 Green innovation and GTFP of firms
According to Schumpeter’s theory of innovation and endogenous growth, innovation is an endogenous factor that boosts business productivity (Aghion and Howitt, 1992). As the LCP is put into practice, it brought about productivity improvement, energy saving and emission abatement, and green restructuring of industrial structure, which in turn reduces the overall carbon emission intensity of the city. Enterprises may effectively lower environmental hazards and other detrimental effects of resource consumption by using green innovation. Green innovation serves as an effective approach for firms to reduce environmental risks as well as other negative impacts of resource use (Fang and Li, 2024), and through innovative improvement of products, processes, or management, enterprises gradually realize green transformation and environmental sustainability (Lian et al., 2022). Green innovation can be split into two stages, the green R&D and green achievement conversion. During green R&D stage, the government’s environmental regulations provided a clear direction for enterprises to engage in green R&D, with the positive subsidies and incentives continue to increase, in order to obtain concessions and subsidies and focus on their own sustainable development, etc., enterprises will continue to introduce new technologies and new products, backward technology upgrading and reforming, R&D, of green technologies in keeping with the minimal carbon concept, used for the whole process of production and operation, so that the use of resources can be maximized, and thus improve the enterprise’s green transformation and environmental sustainability, and then improve the enterprise’s GTFP (Shi and Li, 2019), enterprises pioneering the implementation of green R&D activities tend to protect intellectual property rights through patent applications (Ambec et al., 2013), thus increasing the output of green patents and other outputs, which enhances green R&D’s effectiveness. In the green achievement conversion stage, the enterprise’s pollution investment may successfully encourage the change and improve the enterprise’s green process, attracting the participation of social capital, which brings funds and advanced management experience, prompting the enterprise to focus on the innovation process to enhance the efficiency of the green achievement conversion, and realize the green innovation (Xiao et al., 2022). To sum up, manufacturing enterprises will step up their R&D as part of LCP, carry out green innovation while upgrading existing production processes, improve green R&D and achievement conversion efficiency, thus enhance GTFP and acknowledge how the economy and ecology may grow together.
Based on this analysis, the paper proposes:
Hypothesis 2LCP can improve manufacturing enterprises’ GTFP by green innovation.
2.3 Green output and GTFP of enterprises
The green output implies the realization of factor allocation optimization and green technological innovation (Ma and Sun, 2024). First of all, enterprises are the core force in creating social and economic wealth, and also the requester of natural resources (Zhang and Zhou, 2024), in order to reduce the carbon emissions of the production process, manufacturing enterprises will improve the efficiency of the utilization of internal funds, increase the environmental protection investment used for the introduction of the end-of-pipe disposal technology and the source treatment technology, and realize the greening, cleansing, and high efficiency of production through the input of clean energy to improve the factor By investing in clean energy to realize green, clean and efficient production, the comprehensive utilization rate of factors will be improved, thus promoting the simultaneous enhancement of GTFP (Pan et al., 2022; Wang et al., 2022). Secondly, government’s environmental regulation can make enterprises adjust their resource strategies and adopt green production behaviors by “radiating” their “green will” (Zhang et al., 2018), in which green technological innovation provides technical support for green production, and provides timely and comprehensive information on green production. Technical support, in order to timely and comprehensively grasp the technological frontier and green products and other key information, enterprise management will continue to learn new technologies and new forms of business (Hao et al., 2023), according to the guidelines for reducing carbon emissions and the demands of clean technology R&D, for the purpose of keeping speeding up their own production’s cleanliness and the greening of their goods (Han et al., 2023). The surge in green invention patent applications is a sign of businesses’ green technological innovation. Through green invention patents, enterprises upgrade their existing equipment, for example, increase the input and use of intelligent manufacturing equipment, digital testing and input equipment, and various software, etc., improve their production processes, launch differentiated green products, and improve their GTFP.
Based on this analysis, the paper proposes:
Hypothesis 3LCP can improve manufacturing enterprises’ GTFP by green output.
3 Variable selection and model setting
3.1 Sample selection and data sources
The study’s research sample consists of A-share listed manufacturing companies in China’s Shanghai and Shenzhen markets between 2005 and 2024. The following methods are used to process the data to alleviate anomalous samples’ influence: ① Excluding manufacturing companies that are listed and have “ST” or “*ST” in their abbreviated name; ② Excluding the manufacturing companies on the list that have incomplete data; ③ Using linear interpolation to compensate for each company’s missing data. Following the aforementioned processing, 550 listed manufacturing companies with a total of 11,000 observations make up the balanced panel data that this article ultimately receives. The CSMAR Cathay provided the enterprise-level data used in this study, while the Chinese City Statistical Yearbook provided the majority of the city-level data. Local statistical yearbooks and statistical bulletins are combined to supplement the missing values of particular samples.
3.2 Variable selection
3.2.1 Explained variable
GTFP is based on traditional TFP and takes into account energy use and emission levels, which can objectively assess the quality and green level of firms’ sustainable growth (Wang and Wang, 2021). To measure GTFP in this research, we take the academic article of Wu et al. (2022) as a reference and utilize SBM model for measurement. The SBM (Slacked-Based Measure) model fully considers the adverse effect of neglecting slack variables on the process of efficiency measurement in the previous models, and at the same time avoiding the effect of inefficiency measurement arising from the same proportionate growth or contraction of output inputs in traditional DEA models. Traditional DEA models often use distance functions or curve measures to measure the efficiency of non-expected outputs, but the SBM model can effectively avoid this problem.
Based on the model constructed on the basis of Tone (2001), non-expected outputs are added to the SBM, as shown in Equation 1:In the equation, ; ; ; ; . , and refer to slack variables, which characterize input indicators, expected outputs and unexpected outputs. is the weight of the decision-making module. When , which indicates that the inputs and outputs of the decision-making units are fully effective. When , which indicates the loss of efficiency of decision-making units.
The indicators used to calculate GTFP are shown in Table 1.
TABLE 1
| Serial | Index | Data source |
|---|---|---|
| Investment index | Capital input | CSMAR |
| Energy input | CSMAR | |
| Labor input | CSMAR | |
| Expect output | Operating income | CSMAR |
| Unexpected output | emissions | Chinese city statistical yearbook |
| Dust emission | ||
| Wastewater discharge |
Input and output indicators used in the GTFP measurement.
3.2.2 Core explanatory variable
We use LCP as a quasi-natural experiment for empirical analysis in this study. To show LCP’s treatment effect of LCP, we use the Group × Post interaction term formed by the city-type dummy variable for enterprises’ located cities and the time dummy variable for LCP policy rollout. Specifically, we set the time dummy variable Post to 0 and 1, which correspond to the pre-implementation and post-implementation phases of LCP, and we assign 1 to the Group variable for low-carbon pilot cities as the experimental group and 0 to non-low-carbon pilot cities as the control group in this study. The relevant time dummy variables for low-carbon pilot cities are not consistently consistent across cities due to LCP's pilot-first-then-promote implementation model.
3.2.3 Control variables
We control for firm-level and city-level variables to prevent estimation distortion stemming from unobserved covariates in this research, with firm-level control vaiables including firm size (SIZE), firm age (AGE), debt to asset ratio (LEV), rate of return on total assets (AMC), capital investment (CI), inventory ratio (INV), and operating net cash flow (CAF). Meanwhile, the industrial development level (IGDP) of the enterprise's located city is selected as the city-level control variable. The definitions of all variables are presented in Table 2.
TABLE 2
| Variable | Variable name | Symbol | Definition |
|---|---|---|---|
| Explained variable | Green total factor productivity of enterprises | GTFP | Measure by the SBM model based on the input-output basis |
| Core explanatory variables | Policy-time interaction term | LCP | Dummy variable, a value of 1 is designated for the variable when the sample enterprise lies in a policy-affected region and the corresponding year is the policy impact period, 0 being the assigned value in other cases |
| Control variable | Firm size | SIZE | Ln (total employees at year-end) |
| Firm age | AGE | Ln (current year-establishment year+1) | |
| Debt to asset ratio | LEV | Total liabilities/total assets | |
| Rate of return on total assets | AMC | Net fixed assets/total assets | |
| Capital investment | CI | Ln (net fixed assets) | |
| Inventory ratio | INV | Inventory/total assets | |
| Operating net cash flow | CAF | Operating net cash flow/total assets | |
| Level of regional industrial development | IGDP | Ln (gross industrial output value) |
Variable definition.
3.3 Descriptive statistics
Table 3 summarizes the descriptive statistical findings for the primary variables in this study. For GTFP, the maximum is 1.000 and the minimum is 0.004, with a standard deviation stands at 0.276. This finding reflects significant heterogeneity in GTFP among the sample manufacturing enterprises, showing that their levels of GTFP are not consistent. For the policy variable LCP, its mean of 0.392 indicates that approximately 39% of sample enterprises are subject to LCP due to their location in low-carbon pilot cities.
TABLE 3
| Variable | Obs | Mean | S.D | Min | Max |
|---|---|---|---|---|---|
| GTFP | 11000 | 0.280 | 0.276 | 0.004 | 1.000 |
| LCP | 11000 | 0.392 | 0.488 | 0.000 | 1.000 |
| SIZE | 11000 | 8.069 | 1.255 | 0.000 | 12.290 |
| AGE | 11000 | 2.890 | 0.369 | 1.386 | 3.807 |
| LEV | 11000 | 0.516 | 0.506 | 0.000 | 28.548 |
| AMC | 11000 | 0.265 | 0.162 | 0.000 | 0.902 |
| CI | 11000 | 11.487 | 1.593 | −0.016 | 16.937 |
| INV | 11000 | 0.164 | 0.116 | 0.000 | 0.877 |
| CAF | 11000 | 0.049 | 0.080 | −0.658 | 0.920 |
| IGDP | 11000 | 16.592 | 1.149 | 12.210 | 18.636 |
Descriptive statistics of the variables.
3.4 Empirical model
NDRC promulgated LCP in 2010 and progressively extended its scope in 2012 and 2017. Considering that the LCP is implemented in each city in batches, for the objective of evaluating the pilot policy’s impact on manufacturing firms’ GTFP empirically, the following multi-temporal DID regression model is constructed as shown in Equation 2:
GTFP indicates green total factor productivity of manufacturing firms. LCP indicates low carbon city pilot policy. Control_Var is control variables. CompanyFE is individual fixed effent. Year fixed effect is known as YearFE. is a phrase for randomized disruption.
4 Results and discussion
4.1 Benchmark regression result analysis
This study first performs a benchmark regression on the multi-temporal DID model to investigate the effect of the LCP on the GTFP of manufacturing enterprises. The results are displayed in Table 4. Column (1) does not include control variables, individual and year fixed effects are not controlled, and the main explanatory variable LCP’s estimated coefficient is 0.018 (p<0.01). The estimation outcomes are uniformly significant at the 1% level when control variables are included sequentially in columns (2) and (3) based on adjusting for individual and year fixed effects. The estimation result indicates that LCP significantly affects the GTFP of manufacturing firms, thereby supporting Hypothesis 1.
TABLE 4
| Variable | (1) GTFP | (2) GTFP | (3) GTFP |
|---|---|---|---|
| LCP | 0.018*** (0.007) | 0.025*** (0.007) | 0.025*** (0.007) |
| SIZE | | −0.064*** (0.003) | −0.049*** (0.004) |
| AGE | | 0.004 (0.029) | −0.020 (0.029) |
| LEV | | 0.025*** (0.004) | 0.025*** (0.004) |
| AMC | | −0.272*** (0.022) | −0.194*** (0.026) |
| CI | | | −0.020*** (0.004) |
| INV | | | 0.078*** (0.027) |
| CAF | | | 0.064** (0.027) |
| IGDP | | | 0.091*** (0.008) |
| _cons | 0.273*** (0.003) | 0.834*** (0.087) | −0.538*** (0.154) |
| Year*Company fixed effect | No | Yes | Yes |
| N | 11000 | 11000 | 11000 |
| adj_R2 | 0.047 | 0.112 | 0.128 |
Benchmark regression.
***, **, and * represent significance levels of 1%, 5%, and 10% respectively. The standard error is in parentheses. Similarly here in after.
Moreover, the estimated coefficients of the control variables vary across firm-level and city-level characteristics. Specifically, LEV, INV, and CAF are significantly positive, indicating that moderate debt financing, stronger production organization capacity, and sufficient internal cash flow may help firms improve green production efficiency. By contrast, SIZE, AMC, and CI are significantly negative, suggesting that larger firm size, stronger short-term profitability, and higher capital investment do not necessarily translate into higher GTFP, which may be related to organizational inertia, profit-oriented production expansion, or extensive capital input in some manufacturing firms. The coefficient of AGE is negative but statistically insignificant, possibly because firm age does not directly determine green production efficiency; rather, its effect may depend on firms’ technology renewal, management capability, and green transformation incentives. In addition, IGDP is significantly positive, implying that a higher level of urban industrial development can provide better industrial supporting conditions and external economies for improving firms’ green efficiency.
4.2 Robustness test
4.2.1 Parallel trend test
Since policy shocks emerge at different time points for different low-carbon pilot towns, it’s not feasible to simply designate a time dummy variable for a certain year as the benchmark of LCP implementation. For each low-carbon pilot city, a distinct dummy variable representing relative temporal value of policy rollout must be created. Equation 3 is designed to perform the parallel trend test in this investigation, and its particular version is given as follows:
The observations from the n years prior to each city’s designation as a pilot city, the pilot designation year, and the n years after this designation make up the time dummy variables. All dummy variables have a value of 0 for cities that are not identified as pilot ones. Since the policy was initially put into effect for the first low-carbon pilot city in 2010, and since the research’s sample observation window is set from 2005 to 2024, some pilot towns do not have sample data available for the −5 period or the years before. In order to prevent multicollinearity, it is required to erase this particular time dummy variable and mix the data of other cities in the period before to −5 with that of the −5 period. Therefore, the benchmark periods for the stage prior to policy implementation can only be time dummy variables of the policy periods −1, −2, −3, and −4 at most. According to the results of the parallel trend test, as shown in Figure 2, the calculated coefficients for all pre-policy eras oscillate about zero,with their confidence intervals all encompassing zero. It means that low-carbon pilot program fulfills the parallel trend hypothesis because there were no discernible trending differences between treatment group and control group before the policy was put into place. Meanwhile, throughout the given period, the predicted coefficients for all post-policy eras progressively turn significantly positive, and their confidence intervals stop containing zero. This empirical result shows that LCP’s adoption has a long-lasting, statistically notable treatment effect on manufacturing companies’ GTFP.
FIGURE 2
4.2.2 Placebo test
To test that the improvement of enterprises’ GTFP is brought about by the establishment of LCP, 123 sample listed businesses are chosen at random to serve as experimental group in this study, with remaining businesses serving as control group, repeats it 500 times, and regresses it once more using time dummy variables as the primary explanatory variables and its interaction term.
Figure 3 provides an intuitive display of the distribution characteristics of the predicted coefficients derived from the placebo test. The regression coefficients obtained from the placebo test show an obvious agglomeration pattern centered around zero, as seen in this figure. This is a normal and expected outcome of an effective placebo test. This distribution feature confirms that the main research findings are not influenced by erroneous correlations and successfully eliminates the impact of random variables and unobservable omitted variables on the empirical outcomes of this investigation. Overall, the regression model’s omitted variable issue does not significantly skew the empirical estimation outcomes of this study. This further allows us to conclude that the baseline regression results have good statistical robustness and dependable empirical explanatory power based on the division of experimental group and control group depending on whether manufacturing firms are located within the pilot areas of LCP.
FIGURE 3
4.2.3 Sample data screening
To prevent outlier-induced distortions to baseline regression estimates, GTFP was subjected to two-sided tailing at the 1% quantile and 99% quantile by replacing values less than the 1% and greater than the 99% quantile with values at that point. This common data treatment helps alleviate the effect of outliers discarding valuable observations, thereby enhancing the reliability of the estimation. The model was re-regressed, and the regression results in column (1) of Table 5 show that the coefficients and signs of the core explanatory variables remain largely consistent with those in the baseline regression, with no substantial changes in statistical significance. This suggests that our main findings are robust to the treatment of extreme values and are not driven by outlier observations.
TABLE 5
| Variable | (1) GTFP_w | (2) GTFP_DDF | (3) GTFP_KCA | (4) GTFP_CET | (5) GTFP_ERKU |
|---|---|---|---|---|---|
| LCP | 0.026* (0.014) | 0.048*** (0.013) | 0.026* (0.014) | 0.026* (0.014) | 0.026* (0.014) |
| KCApost | | | −0.002 (0.018) | | |
| CETpost | | | | −0.064 (0.042) | |
| ERKUpost | | | | | −0.003 (0.011) |
| SIZE | −0.049*** (0.011) | −0.023*** (0.009) | −0.049*** (0.011) | −0.049*** (0.011) | −0.049*** (0.011) |
| AGE | −0.020 (0.066) | 0.025 (0.061) | −0.021 (0.066) | −0.002 (0.003) | −0.002 (0.003) |
| LEV | 0.025*** (0.009) | −0.001 (0.010) | 0.025*** (0.009) | 0.025*** (0.009) | 0.025*** (0.009) |
| AMC | −0.194*** (0.055) | −0.254** (0.048) | −0.194*** (0.055) | −0.196*** (0.055) | −0.194*** (0.055) |
| CI | −0.020* (0.012) | 0.006 (0.010) | −0.020* (0.012) | −0.020* (0.012) | −0.020* (0.012) |
| INV | 0.078 (0.058) | 0.050 (0.048) | 0.078 (0.058) | 0.075 (0.058) | 0.078 (0.058) |
| CAF | 0.064 (0.040) | 0.084** (0.035) | 0.064 (0.040) | 0.064 (0.040) | 0.064 (0.040) |
| IGDP | 0.091*** (0.027) | 0.038 (0.023) | 0.091*** (0.028) | 0.091*** (0.027) | 0.091*** (0.028) |
| _cons | −0.489 (0.446) | 0.063 (0.374) | −0.487 (0.454) | −0.505 (0.395) | −0.508 (0.394) |
| Year*Company fixed effect | Yes | Yes | Yes | Yes | Yes |
| N | 11000 | 11000 | 11000 | 11000 | 11000 |
| adj_R2 | 0.128 | 0.067 | 0.128 | 0.128 | 0.128 |
Robustness test.
4.2.4 Using SBM-DDF model
To avoid the randomness of regression results arising from measuring GTFP with a specific method, we subsequently adopt alternative measurement approaches to re-run the regressions. In the baseline analysis, corporate GTFP is measured using the SBM model. To ensure that the estimation results are not biased due to the choice of measurement method, we replace it with the SBM-DDF model to remeasure corporate GTFP and repeat previous regression analysis based on the new measured data. These results are presented in Column (2) of Table 5 The regression results after replacing the GTFP measurement method are highly consistent with baseline regression, thich confirms the primary results of our study exhibit statistical robustness and not dependent on the specific choice of GTFP measurement method.
4.2.5 Excluding other policy effects
Within the sample range selected in this paper, the GTFP of manufacturing enterprises may be affected by other relevant policies in addition to LCP. This paper finds that the special emission limits for air pollutants, the carbon emission trading pilot program, and the list of key environmental supervision units among listed companies are closely related to the research topic of this study. To further rule out the confounding effects of other relevant policies, this study adds three policy indicators, namely, KCApost, CETpost, and ERKUpost, to the benchmark regression specification. For each policy, the indicator takes the value of 1 when a firm, or the city in which it operates, is covered by the corresponding pilot or implementation scheme in a given year, and 0 otherwise. The estimation results are presented in Columns (3)–(5) of Table 5. After these policy variables are introduced step by step, the estimated coefficient of LCP continues to be positive and statistically significant at the 10% level. This suggests that the baseline conclusion remains stable and is not materially driven by the influence of other concurrent policies.
4.2.6 Adding city fixed effects
Given the combination of city-level policy variation and firm-level observations in this study, enterprises’ GTFP may also be influenced by unobserved city-specific characteristics. Therefore, following Zhao and Zhang (2025), city fixed effects are introduced into the benchmark specification as an additional robustness check, so as to account for time-invariant heterogeneity across cities and strengthen the credibility of the empirical results. Table 6 presents the corresponding estimates. Column (1) shows the original benchmark result, whereas Columns (2) and (3) provide the estimates obtained after incorporating city fixed effects. The coefficient of LCP remains positive and statistically significant at the 1% level, irrespective of the inclusion of control variables. These results suggest that the positive impact of LCP on firms’ GTFP persists after accounting for unobserved city-level differences, further supporting the robustness of the core conclusion.
TABLE 6
| Variable | (1) GTFP | (2) GTFP | (3) GTFP |
|---|---|---|---|
| LCP | 0.025*** (0.007) | 0.023*** (0.007) | 0.032*** (0.007) |
| _cons | −0.538*** (0.154) | 0.271*** (0.003) | −1.243*** (0.178) |
| Controls | Yes | No | Yes |
| Year FE | Yes | Yes | Yes |
| Company FE | Yes | Yes | Yes |
| City FE | No | Yes | Yes |
| N | 11000 | 11000 | 11000 |
| adj_R2 | 0.128 | 0.001 | 0.097 |
Robustness test: Adding city fixed effects.
4.2.7 Heterogeneous treatment effects
To evaluate whether variation in treatment timing may bias the staggered DID estimates, this study applies the Goodman-Bacon decomposition to the benchmark DID result, and the corresponding results are presented in Table 7. The decomposition shows that comparisons involving later-treated and earlier-treated units make only a minor contribution to the aggregate estimate. This implies that the benchmark result is primarily identified from more reliable comparison groups, rather than being driven by negative-weight problems or distorted estimates. More specifically, treated-versus-never-treated comparisons contribute 90.4% of the total weight, indicating that they constitute the dominant source of identification. In contrast, comparisons in which earlier-treated units are compared with later-treated units represent only 7.1% of the total weight, suggesting that their role in shaping the final estimate is negligible. Overall, although some treatment-timing heterogeneity is present, its effect on the staggered DID estimation is limited, and the main empirical conclusion remains robust.
TABLE 7
| DD comparison | Weight | Avg DD Est |
|---|---|---|
| Earlier T vs. Later C | 0.071 | 0.049 |
| Later T vs. Earlier C | 0.025 | −0.452 |
| T vs. Never treated | 0.904 | 0.407 |
Robustness test: Goodman-Bacon decomposition results.
4.3 Mechanism testing
4.3.1 Green innovation
Green innovation acts as a vital driving force for improving GTFP. Drawing on Xiao et al. (2022), this paper measures green innovation from two stages, namely, green R&D and green achievement conversion, and adopts green R&D efficiency (GRD) and green achievement conversion efficiency (GCON) to measure corporate green innovation performance. In the green R&D stage, enterprises’ initial inputs are measured by the number of R&D personnel and R&D expenditures. Meanwhile, intermediate outputs are captured by three patent-related indicators: patent applications, valid invention patents, and green invention patents. Efficiency is subsequently evaluated using the SBM model. At the green achievement conversion stage, the outputs generated in the preceding stage are taken as core inputs. Moreover, new product sales revenue and a comprehensive environmental index are selected as final outputs, while R&D spending and the expenses related to technology introduction, assimilation, and absorption are included as supplementary inputs. The SBM model is further employed to calculate the conversion efficiency.
Table 8 presents the estimated results. As shown in Columns (1) and (2), the coefficients of LCP are significantly positive for both GRD and GCON, suggesting that LCP enhances manufacturing firms’ efficiency in green R&D and in transforming green innovation outcomes. This means that, under the combined influence of policy pressure and policy incentives, firms tend to intensify green technological innovation while also improving their capacity to embed green innovation outputs into practical production activities. The mediation analysis further reveals that green innovation serves as a key transmission channel through which LCP promotes the GTFP of manufacturing enterprises. Specifically, LCP strengthens firms’ motivation to undertake low-carbon transition and guides them to devote more innovation resources to energy saving, emissions reduction, cleaner production, and the development of green products. The improvement in GRD indicates that enterprises can generate more green technological achievements with relatively limited R&D inputs, thereby promoting the upgrading of existing production technologies and process flows. On the other hand, the improvement in GCON suggests that enterprises are better able to apply green R&D outcomes to equipment renewal, process improvement, pollution control, and product optimization, thereby transforming green technological advantages into actual productivity gains. Therefore, by promoting green R&D and green achievement conversion, LCP facilitates technological upgrading in manufacturing enterprises and ultimately improves their GTFP.
TABLE 8
| Variable | (1) GRD | (2) GCON | (3) ENIR | (4) GRP |
|---|---|---|---|---|
| LCP | 0.055*** (0.007) | 0.054*** (0.007) | 0.060*** (0.022) | 0.028* (0.015) |
| SIZE | −0.020*** (0.006) | −0.020*** (0.006) | −0.001 (0.013) | 0.033*** (0.009) |
| AGE | 0.477*** (0.020) | 0.471*** (0.018) | 0.137*** (0.040) | −0.042** (0.019) |
| LEV | −0.001 (0.009) | −0.006 (0.006) | 0.037** (0.017) | 0.074*** (0.012) |
| AMC | −0.117*** (0.033) | −0.078** (0.035) | 0.532*** (0.074) | −0.853*** (0.051) |
| CI | 0.022*** (0.005) | 0.021*** (0.006) | −0.225*** (0.012) | 0.133*** (0.008) |
| INV | −0.102*** (0.032) | −0.078** (0.031) | −0.847*** (0.080) | −0.266*** (0.056) |
| CAF | 0.060** (0.029) | 0.100*** (0.027) | −0.862*** (0.112) | −0.168** (0.078) |
| IGDP | −0.035*** (0.009) | −0.039*** (0.010) | 0.015* (0.009) | 0.015* (0.006) |
| _cons | −0.419*** (0.122) | −0.319** (0.143) | −1.969 (1.500) | −2.511*** (0.395) |
| Year*Industry fixed effect | YES | YES | YES | YES |
| N | 11000 | 11000 | 11000 | 11000 |
| adj_R2 | 0.462 | 0.429 | 0.163 | 0.134 |
Mechanism testing.
4.3.2 Green output
Green output involves the reorganization of production factors and the optimization and iteration of products. Liu et al. (2021) analyzed how environmental regulation affects labor employment by distinguishing between “middle governance” and “end governance.” Based on this logic, this study evaluates the greening of output from two dimensions: clean production, which corresponds to intermediate-stage restructuring, and green supply, which reflects end-stage product upgrading. Specifically, LCP strengthens low-carbon constraints and may push firms to adjust existing production patterns, increase investment in cleaner production and pollution control, and expand the supply of green products. These adjustments help shift firms’ output structure toward a more environmentally friendly pattern. First, under low-carbon transition regulation, manufacturing enterprises tend to make environmental protection investment in the production process, which is used for production equipment replacement, pollutant treatment, and other related activities. Therefore, the higher the environmental protection investment cost, the higher the cleanliness level of production. This paper adopts the proportion of environmental protection investment expenses to operating income (ENIR) to measure the cleanliness of the output structure. Second, as a technology-intensive component of green innovation, the number of green invention patents (GRP) can accurately reflect the level of green supply. Therefore, this paper adopts the number of GRP applications to measure green supply.
Table 8 presents the estimated results. The coefficient in Column (3) indicates that LCP has a significant positive effect on ENIR, indicating that the policy promotes the improvement of cleaner production. Column (4) shows that LCP significantly increases the number of GRP applications, thereby improving the level of green supply. Therefore, LCP can promote green adjustments on both the production side and the supply side of manufacturing enterprises. Further analysis indicates that the green output mechanism mainly reflects the role of LCP in optimizing internal factor allocation and output structure. On the one hand, the increase in ENIR shows that, under policy constraints, enterprises allocate more resources to environmental governance and clean production. This enables capital, equipment, and technological resources in the production process to be reallocated toward low-pollution and high-efficiency activities, thus reducing resource misallocation and ineffective consumption while enhancing green production performance. Meanwhile, the increase in GRP indicates that enterprises can promote product optimization and supply upgrading through green technology accumulation, shifting the output structure from traditional high-pollution products toward cleaner and low-carbon products. Therefore, LCP not only promotes internal factor reorganization through clean production but also advances product structure optimization through green supply, ultimately improving manufacturing firms’ GTFP.
4.4 Heterogeneity analysis
4.4.1 Firms ownership heterogeneity analysis
The sample data is separated into state-owned and non-state-owned enterprises based on the kind of company ownership. The model is then re-regressed to investigate the diverse effects of the LCP on various business categories. Table 9 displays the findings without control variables in columns (1) and (2) and the results with control variables in columns (3) and (4). From the results, the core explanatory variables of state-owned enterprises are significantly positive and the coefficients of non-state-owned enterprises are insignificant after adding control variables. It implies that the LCP has a greater impact on state-owned businesses’ GTFP improvement. This paper further employs the Chow test to examine whether the core coefficients differ between the grouped regressions for state-owned and non-state-owned enterprises. The results show a significant difference in the policy effect between the two types of enterprises, indicating that the heterogeneity conclusion based on ownership structure is highly robust.
TABLE 9
| Variable | (1) state-owned enterprises GTFP | (2) non-state-owned enterprises GTFP | (3) state-owned enterprises GTFP | (4) non-state-owned enterprises GTFP |
|---|---|---|---|---|
| LCP | 0.054*** (0.008) | 0.021* (0.011) | 0.027*** (0.008) | 0.006 (0.012) |
| SIZE | | | −0.052*** (0.005) | −0.039*** (0.007) |
| AGE | | | −0.028 (0.028) | −0.048 (0.041) |
| LEV | | | 0.066*** (0.009) | 0.014** (0.005) |
| AMC | | | −0.328*** (0.030) | −0.129*** (0.046) |
| CI | | | 0.007 (0.005) | −0.036*** (0.007) |
| INV | | | 0.230*** (0.035) | −0.063 (0.041) |
| CAF | | | 0.101*** (0.034) | 0.064 (0.044) |
| IGDP | | | 0.082*** (0.007) | 0.028*** (0.009) |
| _cons | 0.250*** (0.014) | 0.296*** (0.022) | −0.605*** (0.122) | 0.679*** (0.175) |
| Year*Company fixed effect | YES | YES | YES | YES |
| N | 7032 | 3968 | 7032 | 3968 |
| adj_R2 | 0.028 | 0.036 | 0.139 | 0.168 |
| Chow test | 3.28 | 3.92 | ||
| P-value | 0.000 | 0.000 | ||
Heterogeneity analysis.
As a critical environmental regulation policy, the LCP program strengthens regulatory standards and imposes penalties for non-compliance, thereby compelling firms to adjust their production and operational practices to meet stricter environmental requirements. Due to their distinct institutional attributes and closer ties to the government, state-owned enterprises (SOEs) enjoy easier access to financial support from governmental bodies and state-affiliated financial institutions compared to non-state-owned enterprises (non-SOEs). Such support often takes the form of preferential loans, tax incentives, and direct fiscal subsidies, which substantially lower the cost of green technology adoption and facility upgrading. Furthermore, SOEs exhibit higher sensitivity to policy directives and possess stronger organizational capacity for policy implementation, given their role as key agents in fulfilling national strategic objectives. This alignment with policy goals, combined with their resource advantages, enhances SOEs’ motivation and ability to pursue green and low-carbon transformation under environmental regulatory pressures. Consequently, these factors explain why the LCP exerts a more pronounced positive effect on the GTFP of state-owned manufacturing enterprises, while its impact on non-SOEs remains statistically insignificant.
4.4.2 Industry types heterogeneous analysis
In 2010, China’s Ministry of Ecology and Environment released the Guidelines for Environmental Information Disclosure of Listed Companies, which provided an official basis for identifying heavily polluting industries. According to this standard, heavily polluting industries cover 16 categories, namely, thermal power, iron and steel, cement, electrolytic aluminum, coal, metallurgy, chemicals, petrochemicals, building materials, paper-making, brewing, pharmaceuticals, fermentation, textiles, leather-making, and mining. Referring to Dai et al. (2022) and Wang and Zhong (2024), the sample firms are divided into heavily polluting and lightly polluting manufacturing firms according to this official classification. A pollution-intensity dummy variable, Heavily, is then constructed, which takes the value of 1 if a firm operates in a heavily polluting industry and 0 otherwise. Based on this classification, grouped regressions are first conducted, and the interaction term LCP × Heavily is then added to test whether the policy effect varies with industry pollution intensity. Table 10 reports the corresponding estimates. Columns (1)–(4) show the results of the grouped regressions. The estimates indicate that LCP significantly improves the GTFP of heavily polluting firms regardless of the inclusion of control variables, whereas the estimated effect for lightly polluting firms is positive but statistically insignificant. In addition, the Chow test is used to formally examine whether the key coefficients differ across the two groups. The test results confirm a significant inter-group difference, suggesting that the heterogeneity pattern based on pollution intensity is reliable.
TABLE 10
| Variable | (1) heavily polluting | (2) lightly polluting | (3) heavily polluting | (4) lightly polluting | (5) GTFP |
|---|---|---|---|---|---|
| LCP | 0.079*** (0.009) | 0.002 (0.010) | 0.036*** (0.009) | 0.007 (0.009) | 0.006 (0.016) |
| LCP × Heavily | | | | | 0.048** (0.019) |
| Size | | | −0.061*** (0.060) | −0.046*** (0.005) | −0.048*** (0.011) |
| Age | | | −0.003 (0.018) | 0.034 (0.032) | −0.002 (0.003) |
| LEV | | | 0.043*** (0.008) | 0.033*** (0.006) | 0.024*** (0.009) |
| AMC | | | −0.365*** (0.032) | −0.292*** (0.036) | −0.195*** (0.055) |
| CI | | | 0.047*** (0.005) | −0.018*** (0.005) | −0.020* (0.012) |
| INV | | | 0.201*** (0.040) | 0.082** (0.033) | 0.077 (0.058) |
| CAF | | | 0.129** (0.051) | 0.054 (0.036) | 0.063 (0.040) |
| IGDP | | | 0.033*** (0.004) | 0.063*** (0.008) | 0.093*** (0.028) |
| _cons | 0.241*** (0.016) | 0.281*** (0.016) | −0.231*** (0.076) | −0.202 (0.144) | −0.537 (0.397) |
| Year*Company fixed effect | Yes | Yes | Yes | Yes | Yes |
| N | 5022 | 5978 | 5022 | 5978 | 11000 |
| adj_R2 | 0.030 | 0.032 | 0.103 | 0.165 | 0.130 |
| Chow test | 3.29 | 3.34 | | ||
| P-value | 0.000 | 0.000 | | ||
Heterogeneity analysis.
Column (5) further includes the interaction term. The coefficient of LCP itself is insignificant, while the coefficient of LCP × Heavily is significantly positive. This means that the marginal effect of LCP is much stronger for heavily polluting firms than for lightly polluting firms. The linear combination test further shows that the total policy effect for heavily polluting firms is significantly positive, whereas that for lightly polluting firms remains insignificant. These results provide additional evidence for the existence of heterogeneous policy effects across industries with different pollution intensities.
From a theoretical perspective, carbon reduction policies may impose two opposite forces on heavily polluting firms: a compliance-cost effect and a Porter effect. Stricter environmental regulation may raise firms’ compliance burden and negatively affect their short-term performance. However, it may also enhance productivity by improving resource allocation and stimulating technological innovation as well as green transformation. The empirical evidence shows a significantly positive net effect of LCP on heavily polluting firms, implying that the Porter effect is more prominent in this setting. As the main contributors to pollutant emissions, heavily polluting industries are subject to stronger regulatory pressure and stricter emission-reduction requirements under the policy framework. Such pressure pushes firms to speed up technological upgrading and shift toward cleaner production. Meanwhile, these firms may also receive more targeted government support, including green credit, fiscal subsidies, and technical guidance, which further strengthens their incentives for green innovation and low-carbon transformation. In contrast, lightly polluting industries face weaker regulatory pressure and lower transition requirements, which may explain why the impact of LCP on their GTFP is not statistically significant.
5 Conclusion and recommendations
This study builds a multi-temporal DID model to examine the impact of LCP on the GTFP of manufacturing enterprises and its mechanism of action, using data from listed manufacturing enterprises between 2005 and 2024. It is discovered that the LCP significantly enhances the GTFP of manufacturing enterprises, and the conclusion still holds after a series of robustness tests. This conclusion varies significantly across the nature of ownership as well as across industries, with the LCP’s promotion of GTFP being more pronounced for state-owned enterprises and enterprises in heavily polluting industries, while the effect is not significant for non-state-owned enterprises and those in lightly polluting industries. Mechanism analysis shows that LCP promote GTFP through innovation greening and output greening. On one hand, the implementation of LCP policies significantly improves green technology R&D efficiency and green technological achievements’ transformation rate. On the other hand, it also markedly improves the cleanliness level of production outputs and the level of green supply.
Based on the above empirical results, this paper proposes the following policy implications.
First, LCP should be steadily advanced, and the pace of policy expansion should be optimized. The baseline regression results show that LCP can significantly improve manufacturing enterprises’ GTFP, indicating that the policy has a positive green development effect. Therefore, on the basis of summarizing existing pilot experience, the government should gradually expand the coverage of the policy. However, in the process of policy promotion, the pilot model should not be mechanically replicated. Instead, policy intensity should be reasonably determined according to differences in industrial structure, resource endowment, and green development foundation across regions. For regions with a solid foundation for green transformation, low-carbon development standards may be appropriately raised. For regions facing greater pressure from industrial transformation, supporting measures such as fiscal assistance, financial services, and technical support should be strengthened to reduce enterprises’ short-term transition pressure.
Second, the policy support system should be improved around green innovation and green output, so as to strengthen the transmission effect through which LCP improves enterprises’ GTFP. The mechanism test results show that LCP mainly promotes the GTFP of manufacturing enterprises through green innovation and green output. Therefore, the government should reduce the cost of green technology R&D through green R&D subsidies, loan interest subsidies, green patent rewards, and additional deductions for R&D expenses. At the same time, green technology trading platforms, industry-university-research cooperation mechanisms, and intellectual property protection systems should be further improved to enhance the efficiency of green achievement conversion. In terms of green output, the government should support enterprises in upgrading clean production equipment, carrying out energy-saving and carbon-reduction technological transformation, and developing green products. In addition, green product demand can be expanded through green certification, government green procurement, and green supply chain management.
Third, differentiated governance should be implemented according to ownership and industry heterogeneity, while balancing green transformation and economic efficiency. The heterogeneity test results indicate that LCP has a more pronounced promoting effect on the GTFP of state-owned enterprises and heavily polluting enterprises, while its effects on non-state-owned enterprises and lightly polluting enterprises are not significant. Therefore, for state-owned enterprises, indicators such as GTFP, reductions in energy consumption per unit of output, green patent conversion, and the level of clean production should be incorporated into the performance evaluation system. This would strengthen their demonstration role in green transformation while avoiding resource misallocation caused by undifferentiated subsidies. For private enterprises, policy tools such as green technological transformation subsidies, loan interest subsidies, tax reductions and exemptions, green guarantees, and green credit should be used to relieve financing constraints and reduce transition cost pressures. For heavily polluting industries, carbon emission constraints, pollution discharge permits, and environmental information disclosure should be further strengthened, while fiscal subsidies and green credit should be directed mainly toward clean production equipment renewal and energy-saving and carbon-reduction technological transformation projects. For lightly polluting industries, greater reliance should be placed on green certification, green supply chain management, and market-oriented incentive mechanisms to guide enterprises in improving green supply capacity and to avoid efficiency losses caused by excessive regulation.
Statements
Data availability statement
The datasets presented in this article are not readily available because the firm-level data used in this study are mainly obtained from the China Stock Market and Accounting Research Database (CSMAR). As CSMAR is a subscription-based commercial database, the original data are subject to license agreements and copyright restrictions. Therefore, the raw data cannot be directly disclosed or publicly shared by the authors. However, the data can be accessed through CSMAR by institutions or individuals with authorized subscriptions. Requests to access the datasets should be directed to China Stock Market and Accounting Research Database (CSMAR), https://data.csmar.com.
Author contributions
XZ: Formal Analysis, Methodology, Supervision, Validation, Writing – review and editing. XD: Data curation, Methodology, Visualization, Writing – original draft. ZJ: Data curation, Methodology, Visualization, Writing – original draft. XG: Conceptualization, Data curation, Methodology, Supervision, Validation, Writing – review and editing. LZ: Data curation, Formal Analysis, Methodology, Validation, Investigation, Writing – original draft, Writing – review and editing.
Funding
The author(s) declared that financial support was received for this work and/or its publication. This study is funded by the Youth Project of the Education Department of Liaoning Province in 2025, with the project name being Integrated Construction of Shenyang Modern Urban Agglomeration - Based on the Perspective of Industrial Chain and Co. chain (Project Number: LJ112510140028).
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenvs.2026.1842523/full#supplementary-material
References
1
AghionP.HowittP. (1992). A model of growth through creative destruction. Econometrica60 (2), 323–351. 10.2307/2951599
2
AmbecS.CoheA. M.ElgieS.LanoieP. (2013). The porter hypothesis at 20: can environmental regulation enhance innovation and competitiveness?Rev. Environmental Economics Policy7 (1), 2–22. 10.1093/reep/res016
3
ChenH.GuoW.FengX.WeiW.LiuH.FengY.et al (2021). The impact of low-carbon city pilot policy on the total factor productivity of listed enterprises in China. Resour. Conservation Recycl.169, 105457. 10.1016/j.resconrec.2021.105457
4
DaiQ.HuangH.ZhangX.SuY.LiuC.LiQ. (2022). Mediation effect of corporate tax burden and the relationship between environmental regulation and firm performance. Int. J. Environ. Res. Public Health19 (22), 14987. 10.3390/ijerph192214987
5
DongY.SunZ.LiW.WangL. (2024). How the low-carbon city pilot policy reduces enterprise carbon emissions and paves the way for high-quality development: evidence from a quasi-natural experiment in China. Environ. Dev. Sustain., 1–25. 10.1007/s10668-024-05724-w
6
FangL.LiZ. (2024). Corporate digitalization and green innovation: evidence from textual analysis of firm annual reports and corporate green patent data in China. Bus. Strategy Environ.33 (5), 3936–3964. 10.1002/bse.3677
7
FuL.ZhaoH.MaF.ChenJ. (2024). Estimating heterogeneous effects of China’s low-carbon pilot city policy on urban employment. J. Clean. Prod.434, 139882. 10.1016/j.jclepro.2023.139882
8
HanX. F.ChenL. T.SongW. F.LiB.ChenX. (2023). Dynamic adjustment mechanism of carbon emission on green technology innovation-driven industrial upgrading. China Population-Resources Environ.33 (12), 60–70. 10.12062/cpre.20230316
9
HaoX.LiY.RenS.HaoY. (2023). The role of digitalization on green economic growth: does industrial structure optimization and green innovation matter?J. Environmental Management.325, 116504. 10.1016/j.jenvman.2022.116504
10
JiangL. D.ZhaoX. J.ZhangG. F. (2024). Impact of pilot low-carbon city policies on the development of green trade in China. China Population-Resources Environ.34 (09), 32–44. 10.12062/cpre.20240730
11
LiZ.LaiA.CaoY.WangQ. (2024). Porter effect vs cost effect: the impact of China's low carbon city pilot on carbon emissions and economic performance. J. Environ. Manag.360, 121015. 10.1016/j.jenvman.2024.121015
12
LianG.XuA.ZhuY. (2022). Substantive green innovation or symbolic green innovation? The impact of ER on enterprise green innovation based on the dual moderating effects. J. Innovation and Knowl.7 (3), 100203. 10.1016/j.jik.2022.100203
13
LiuM.TanR.ZhangB. (2021). The costs of “blue sky”: environmental regulation, technology upgrading, and labor demand in China. J. Dev. Econ.150, 102610. 10.1016/s0377-2217(99)000407-5
14
MaS. X.SunY. L. (2024). Low-carbon transition and digital transformation of enterprises: evidence from low-carbon city pilot policy. Industrial Econ. Res.131 (04), 43–57. 10.13269/j.cnki.ier.2024.04.003
15
Martínez-ZarzosoI.Bengochea-MoranchoA.Morales-LageR. (2019). Does environmental policy stringency foster innovation and productivity in OECD countries?Energy Policy134, 110982. 10.1016/j.enpol.2019.110982
16
PanQ.ZhaoS. (2024). The impact of low-carbon city pilot policy on urban green technology innovation: based on government and public perspectives. Plos One19 (7), e0306425. 10.1371/journal.pone.0306425
17
PanA.ZhangW.ShiX.DaiL. (2022). Climate policy and low-carbon innovation: evidence from low-carbon city pilots in China. Energy Econ.112, 106129. 10.1016/j.eneco.2022.106129
18
PanS.LuX.ChaiY.HuangD.CaiY. (2024). Low carbon city and FDI inflows: evidence from China. Environ. Sci. Pollut. Res.31 (7), 10198–10212. 10.1007/s11356-023-27902-5
19
PengY.WangZ.ZhangY.WangW. (2025). Facilitating or hindering? The impact of low-carbon pilot policies on socio-ecological resilience in resource-based cities. Land14 (1), 147. 10.3390/land14010147
20
PorterM. E.LindeC. V. D. (1995). Green and competitive: ending the stalemate. Harv. Bus. Rev.73 (5), 120–134. Available online at: https://www.hbs.edu/faculty/Pages/item.aspx?num=5512.
21
ShiX.LiL. (2019). Green total factor productivity and its decomposition of Chinese manufacturing based on the MML index: 2003–2015. J. Clean. Prod.222, 998–1008. 10.1016/j.jclepro.2019.03.080
22
TianY.FengC. (2022). The internal-structural effects of different types of environmental regulations on China's green total-factor productivity. Energy Econ.113, 106246. 10.1016/j.eneco.2022.106246
23
ToneK. A. (2001). Slacks-based measure of efficiency data envelopment analysis. Eur. J. Operational Res.130 (3), 498–509. 10.1016/s0377-2217(99)00407-5
24
WangX.WangQ. (2021). Research on the impact of green finance on the upgrading of China's regional industrial structure from the perspective of sustainable development. Resour. Policy74, 102436. 10.1016/j.resourpol.2021.102436
25
WangX.ZhongX. (2024). Digital transformation and green innovation: firm-level evidence from China. Front. Environ. Sci.12, 1389255. 10.3389/fenvs.2024.1389255
26
WangL. F.ZhaoY.WangL. J. (2022). Carbon reduction effects and mechanisms in low-carbon pilot cities. Geography Res.41 (7), 1898–1912. 10.11821/dlyj020210773
27
WangK. L.LiJ.XuR. Y.PangS. Q.MiaoZ.SunH. P. (2023). The impact of low-carbon city pilot policy on green total-factor productivity in China’s cities. Environ. Sci. Pollut. Res.30 (9), 24299–24318. 10.1007/s11356-022-23934-5
28
WangW.LiD.ZhouS.HanZ. (2024). Towards sustainable development: assessing the effects of low-carbon city pilot policy on residents’ welfare. Environ. Dev. Sustain.28, 1–34. 10.1007/s10668-024-05373-z
29
WuJ.XiaQ.LiZ. (2022). Green innovation and enterprise green total factor productivity at a micro level: a perspective of technical distance. J. Clean. Prod.344, 131070. 10.1016/j.jclepro.2022.131070
30
XiaoR. Q.ChenX. T.QianL. (2022). Heterogeneous environmental regulation, government support and firms' green innovation efficiency based on a two-stage value chain perspective. Finance Trade Res.33 (09), 79–93. 10.19337/j.cnki.34-1093/f.2022.09.007
31
YuanG.LiuJ.WangY. (2025). Low-carbon city pilot policies, government attention, and green total factor productivity. Finance Res. Lett.77, 107043. 10.1016/j.frl.2025.107043
32
ZhangR.ZhouY. (2024). Green finance, carbon emission intensity, and corporate ESG performance: an empirical study based on micro data of listed companies. Soc. Sci.3, 126–140. 10.13644/j.cnki.cn31-1112.2024.03.014
33
ZhangY.WangJ.XueY.YangJ. (2018). Impact of environmental regulations on green technological innovative behavior: an empirical study in China. J. Clean. Prod.188, 763–773. 10.1016/j.jclepro.2018.04.013
34
ZhaoC.ZhangL. (2025). Can government-guided funds promote innovation output in strategic emerging enterprises? Evidence from China. PLoS One20 (10), e0334826. 10.1371/journal.pone.0334826
35
ZhengJ.ShaoX.LiuW.KongJ.ZuoG. (2021). The impact of the pilot program on industrial structure upgrading in low-carbon cities. J. Clean. Prod.290, 125868. 10.1016/j.jclepro.2021.125868
36
ZhouQ.ZhongS.ShiT.ZhangX. (2024). “Environmental regulation and haze pollution: neighbor-companion or neighbor-beggar?” in Handbook on Climate Change and Environmental Governance in China. Cheltenham, United Kingdom: Edward Elgar Publishing, 158–177. 10.4337/9781035316359.00018
Summary
Keywords
green innovation (GI), green total factor productivity, low-carbon city pilot policy, manufacturing companies, multi-temporal difference-in-difference model
Citation
Zhang X, Ding X, Jia Z, Guo X and Zhang L (2026) Low carbon transition and green total factor productivity of manufacturing firms: empirical evidence from low-carbon city pilot policy. Front. Environ. Sci. 14:1842523. doi: 10.3389/fenvs.2026.1842523
Received
01 April 2026
Revised
19 May 2026
Accepted
23 June 2026
Published
31 July 2026
Volume
14 - 2026
Edited by
Mobeen Ur Rehman, Keele University, United Kingdom
Reviewed by
Songyuan Liu, State Grid Economic and Technological Research Institute Co., Ltd., China
Xia Mao, Hainan University, China
Updates
Copyright
© 2026 Zhang, Ding, Jia, Guo and Zhang.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Xiaoling Guo, gxlgxl2008@126.com; Lin Zhang, lin_zhang1995@126.com
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.