Firstly, according to the China Environmental Protection Statistical Yearbook in 2011, following the research of Chai et al. (2022), we calculate the variation coefficient of each industrial pollutant, which is the ratio of the standard deviation of each industrial pollutant emission and its mean value ( C o V n , n = industrial waste gas emission, industrial waste water emission, industrial solid waste emission). The weight of each industrial pollutant can be expressed as Equation 1: w n = C o V n ∑ C o V n (1)

We conduct dimensionally standardized treatment on each industrial pollutant and multiply the corresponding weight ( w n ) of different industrial pollutants to obtain the dimension value as Equation 2: d n = w n A n − m n M n − m n (2) where M n and m n represent the maximum and minimum values of three industrial pollutant emissions in different industries in the same year, respectively. For the calculation of the final composite index, the standard Euclidean distance between any point X within N-dimensional Cartesian space and the lowest point (L) and highest point (H) of the pollution level, which is described as Equation 3: L = d W a s t e w a t e r 2 + d W a s t e   g a s 2 + d S o l i d   w a s t e 2 w W a s t e w a t e r 2 + w W a s t e   g a s 2 + w S o l i d   w a s t e 2 1 2 (3) H = 1 − w E x h a u s t   g a s − d E x h a u s t   g a s 2 − w W a s t e w a t e r − d W a s t e w a t e r 2 − w S o l i d   w a s t e − d S o l i d   w a s t e 2 w W a s t e w a t e r 2 + w E x h a u s t   g a s 2 + w S o l i d   w a s t e 2 1 2

Taking the arithmetic mean value of L and H to quantify the pollution extent of different industries. The manufacturing industry is classified according to its arithmetic mean value. Industries with a pollution degree higher than the arithmetic mean value are defined as heavily-polluting industries, while those with a pollution degree lower than the arithmetic mean value are tagged as non-heavy polluting industries. Founded on this, we designate the heavily-polluting sector marked as the treatment group and the non-heavy polluting sector marked as the control group, as depicted in Table 1.

Secondly, we select the panel datasets between 2010 and 2019 for Chinese manufacturing firms listed publicly as the initial samples, then process them as follows to obtain all samples in this paper: (1) Excluded all ST and *ST samples; (2) Excluded samples marked by serious data gaps; (3) Excluded the initial samples with an asset-liability ratio greater than 1; (4) Winsorized the first and last 1% of all continuous variables and finally obtained a total of 10,841 observations for 1,255 enterprises. The dataset consisted of 360 heavily-polluting companies and 895 non-heavily-polluting ones, sourced from the CSMAR database.

We adopt the two-way fixed effects model to investigate the effect of GCP on the labor productivity of Chinese heavily-polluting enterprises. It provides more consistent parameter estimates than random effects models (Bell et al., 2019); it can make better use of time-series information from panel data than difference-in-difference models (Fredriksson and Oliveira, 2019); and it does not need to rely on strong assumptions about instrumental variable assumptions compared to instrumental variable approaches (Milner et al., 2018). Two-way fixed effects models not only effectively control for the invariant properties of both the firm and time dimensions, thereby reducing omitted variable bias and endogeneity problems, but also fully exploit the time-series and cross-sectional information in panel data to improve the explanatory power and predictive accuracy of the model (Fredriksson and Oliveira, 2019; Gauthier, 2021). The regression equation is as Equation 4: L a b p r o i t = α + β 1 T r e a t e d i × T i m e t + γ C o n t r o l i t + δ i + λ t + ε i t (4) where L a b p r o i t is the labor productivity of the ith enterprise in t year. T r e a t e d i × T i m e t is the virtual variable. If the ith enterprise belongs to the heavily-polluting industry, the value of T r e a t e d i is 1; in other cases, the value of T r e a t e d i is 0. T i m e t indicates the implementation year of the GCP, the value of T i m e t before 2012 is 0; the value of T i m e t in 2012 and after 2012 is 1. β 1 represents the effect after the GCP implemented.   C o n t r o l i t as a control variable at the enterprise level. δ i and λ t stand for the fixed effects of company and time. ε i t is the random disturbance term that follows a normal distribution.

It can be seen from Table 6, before adding control variables, the coefficient of column (1) is 0.317, statistically significant with a p-value of less than 0.05. This result implies that GCP can significantly boost the labor productivity of heavily-polluting companies and also verify the existence of the “strong” Porter Hypothesis in heavily-polluting industry. This aligns with the results reported by Cui et al. (2022), thereby confirming H1. After including the inclusion of control variables, the coefficient for column (2) remains at 0.321, with a 95% confidence level. Thus, the paper’s base regression outcome proves study.

According to the results in Table 6, the total influence of GCP on the firm’s labor productivity is 0.321. After controlling the effect of commercial credit, the direct influence of GCP on the firm’s labor productivity is 0.291 (see in column (2) in Table 8). Depending on the research of Baron and Kenny (1986) and Wen et al. (2004), the commercial credit mediating influence accounts for 0.029 (0.291*2.447), about 9.03% (0.029/0.321) of the total effect. It shows that commercial credit plays a partial mediating role between GCP and the labor productivity of heavily-polluting businesses, and the H2a was verified. The above findings indicate that GCP helps to exert the environmental governance effect of financial resources by guiding the judicious distribution of bank resources among enterprises, and thus enhancing the labor productivity of heavily-polluting enterprises, which verifies H2b.

After controlling the effect of technological innovation, the direct influence of GCP on the firm’s labor productivity is 0.271 (see in column (4) in Table 8). Based on the research of Baron and Kenny (1986) and Wen et al. (2004), the impact of technological innovation as a mediator is 0.011 (0.271*0.04), about 3.43% (0.011/0.321) of the total effect. It verifies that technological innovation serves a partial mediating function in the influence of GCP on companies’ labor productivity, and the H3a is verified. Those findings indicate that GCP, as an environmental policy tool, can stimulate the internal motivation of enterprise technological innovation, thus increasing the labor productivity of Chinese heavily-polluting companies, which retested the “Porter effect” and the H3b is verified.

Given China’s institutional background, state-owned enterprises (SOE) and private enterprises face de facto unequal competition. Compared with private enterprises, SOE give more emphasis to achieving their social and environmental responsibilities. Furthermore, SOE generally have better access to financial support from banks and financial institutions compared to private enterprises. According to the result of column (1) and (2) in Table 9, there are significant different impacts of GCP on the labor productivity of SOE and POE (see column (1) and column (2)). Specifically, GCP positively impacts labor productivity in SOE, showing significance at the 5% level. While GCP positively affects the SOE’s labor productivity, it is not significant. It has been verified that GCP’s influence on labor productivity of heavily-polluting manufacturing companies is more obvious in SOE.

Varying significantly in operating efficiency, internal control, and funding capacity, enterprises across different sizes showcase stark disparities. “ Constrained by GCP, small enterprises have to halt green production activities due to financial limitations, leading to a vicious circle of “financing difficulties-green transformation difficulties”. This study defines enterprises above the industry median size as large companies (LC) and those below as small companies (SC). Table 9 columns (3) and (4) depict the varied effects of GCP on firms’ labor productivity across scales. The regression coefficient in column (3) is positively significant at the 5 percent level, in contrast to the negative and non-significant regression coefficient in column (4). This finding implies that GCP has a greater effect on labor productivity in heavily polluting manufacturing companies when compared to smaller businesses.

In contrast to central China and western China, the economic development level and financial market environment of Eastern China are relatively better, and the government and enterprises focus more on environmental protection. Therefore, enterprises in eastern China are less likely to be restricted by green credit policy because their environmental performance does not meet the standards. Based upon the province (city) where the firm is registered, we examined the heterogeneous impact of GCP on labor productivity from eastern China, central China, and western China. The results of column (5) to (7) in Table 9 show that GCP has a significant regional heterogeneity in its promotion of the enterprise’s labor productivity. Specifically, the coefficient for the fifth column is 0.420, indicating significance at the 5% level. The sixth and seventh columns’ coefficients, which are 0.247 and −0.059, respectively, are not significant. It indicates that GCP can only promote the firm’s labor productivity in Eastern China.

This research adopts panel datasets between 2010 and 2019 for Chinese manufacturing firms listed publicly. Its objective was to explore how GCP affects the labor productivity of heavily-polluting enterprises in China, along with the micro-mechanisms involved. Then, the analytical results indicate that: (1) The implementation of GCP has raised the labor productivity of Chinese heavily-polluting enterprises, which means GCP can foster a win-win scenario where environmental conservation and economic expansion coexist harmoniously. (2) By introducing GCP, the financial hurdles for heavily-polluting enterprises in China intensify, thus stimulating those enterprises to actively seek external commercial credit and obtain more financial support. (3) The implementation of GCP can stimulate innovation activities among Chinese heavily-polluting firms, and has produced the “Porter effect”. (4) The heterogeneity test shows that GCP’s stronger influence on state-owned and large heavily-polluting enterprises in eastern China.

Following the outcomes and conclusions of the empirical study, this dissertation put forward policy recommendations as follows:

It is essential for government departments to boost their monitoring of commercial banks, so that they can efficiently screen enterprises for environmental risks. For example, regularly conduct compliance reviews of banks’ environmental risk management to ensure they adhere to GCP. At the same time, provide training and technical support to banks to help them enhance their capabilities in environmental risk assessment and green project evaluation.

Banks implement differentiated credit policies for enterprises. For enterprises that perform poorly environmentally, banks should strictly control the credit lines. At the same time, they should reduce credit constraints on enterprises with good environmental performance. Such measures can force enterprises with poor environmental performance to undertake green technological innovation.

For enterprises that possess strong commercial credit financing ability, the financing constraint influence of GCP is weakened, and enterprises still lack the motivation to improve their environmental performance. Therefore, the government needs to enhance the environmental protection policies for these enterprises, establish a mandatory environmental information disclosure system, and effectively promote their green transformation.