The dependent variable is corporate GTFP (GTFP). Unlike patent-based measures of green innovation, GTFP evaluates input-output efficiency under environmental constraints. Following Li and Wang, this study uses the Malmquist-Luenberger productivity index integrated with a slacks-based measure directional distance function to assess corporate GTFP^[64]. We develop a systematic framework for measuring corporate GTFP using the following input-output indicators:

(1) Input factors: Labor is measured by the total number of employees at the firm level; capital input is measured by the net value of fixed assets; and energy input is measured by allocating city-level industrial electricity consumption based on each firm’s share of total urban employment.

(2) Desirable outputs: A firm’s expected output is measured by operating revenue, which reflects its economic performance.

(3) Non-desirable outputs: These include environmental pollutants such as sulfur dioxide, wastewater, and particulate matter (soot). Because firms do not directly report these emissions, they are estimated by multiplying city-level total emissions of each pollutant by the firm’s proportion of total urban employment. This approach assumes that pollutant generation is correlated with employment scale, providing a reasonable proxy for a firm’s environmental burden.

The independent variable is G-Finance. Because there is no standardized metric for green finance evaluation, some studies use proxy indicators such as green credit. However, this approach does not fully capture the overall development of green finance. This study constructs a provincial-level G-Finance index by integrating the main components of China’s green finance system while considering data availability. The selected instruments - green credit, green bonds, green stocks, and green funds - are weighted using the entropy weight method to ensure a systematic and reliable evaluation^[21].

Green credit represents bank-dominated indirect financing, whereas the other components represent market-dominated direct financing. Specifically, green credit is measured as the ratio of non-high-energy-consuming industry interest expenditures to total industrial interest expenditures in each province^[65]. Green bonds are measured as the ratio of each province’s green bond issuance to total bond issuance^[30]. Green stocks are measured as the proportion of the total market capitalization of environmentally friendly enterprises in each province to the total market capitalization of A-shares. Finally, green funds are measured as the ratio of the total market capitalization of green funds in each province to the total market capitalization of all funds.

This study examines two mechanisms: green innovation and carbon emissions. Green patent research and development is a widely used and effective measure of green innovation^[66]. This study uses annual green patent applications to represent green innovation (Patent). Corporate carbon emissions (Carbon) are calculated following the methodology of Wang et al.^[67]. Firms are classified based on their disclosure practices. The first category includes firms that directly disclose total carbon emissions in their reports. The second category includes firms that do not directly disclose annual carbon emissions^[68]. For these firms, total carbon emissions are estimated based on fossil energy, electricity, and thermal energy consumption^[69,70]. Following the IPCC (2006) framework and the National Development and Reform Commission (NDRC) guidelines, total carbon dioxide emissions (E) for each firm are calculated as follows:

(1) $$ E=A D \times E F $$

where AD represents resource consumption, and EF represents emission factors. For Scope 1 emissions, AD is derived from fossil fuel consumption multiplied by their average low calorific values provided in the NDRC guidelines. For Scope 2 emissions, AD includes purchased electricity and heat. EF are calibrated using official parameters. For fossil fuels, we adopt NDRC default emission factors. For purchased electricity, we apply regional average grid emission factors from the National Center for Climate Change Strategy and International Cooperation. For heat consumption, we use a standardized factor of 0.11CO₂/GJ. Finally, this study applies a natural logarithm transformation to total carbon emissions.

This study incorporates several firm-level control variables to account for additional economic factors affecting corporate GTFP, ensuring analytical comprehensiveness and clarity. Specifically, the control variables include firm age (Age), firm size (Size), revenue growth rate (Growth), return on total assets (ROA), return on invested capital (ROIC), and core business revenue growth rate (Income)^[64,71]. To enhance regression robustness, the model includes firm-level (Firm) and year (Year) fixed effects. Table 1 presents the variable definitions.

The baseline analysis does not fully account for potential endogeneity, relying on direct regression estimates and concluding that green finance significantly promotes corporate GTFP. However, potential endogeneity and omitted variable bias may undermine the causal interpretation of the relationship between green finance initiatives and corporate GTFP. To address these concerns, this study adopts a difference-in-differences (DID) approach. In 2017, the State Council designated Zhejiang, Jiangxi, Guangdong, Guizhou, and Xinjiang as pilot regions for green finance reform and innovations. This initiative aims to explore new mechanisms for green finance development, expand diversified financing channels for green industries, and enhance corporate green productivity. As a centrally coordinated policy, it generates a relatively exogenous shock at the regional level, allowing the use of a DID model to identify the impact of green finance development on corporate GTFP. These pilot zones provide a credible quasi-natural experiment for empirical identification, leading to the following model:

(7) $$ G T F P_{i, t}=\beta_{0}+\beta_{1} { Treat }_{k} \times { Post }_{t}+\beta_{2} { control }_{i, t}+\sum { Year }+\sum { Firm }+\varepsilon_{i, t} $$

where i denotes listed companies, k denotes the province in which the company is located, and t represents the year (t = 2010, 2011, ..., 2021). GTFP_i,t represents GTFP of enterprises. For Treat_k, a value of 1 is assigned to firms located in provinces with pilot zones for green finance reform and innovations, and 0 otherwise. For Post_t, a value of 1 is assigned if the observation year falls within or after the establishment of the pilot zones, and 0 otherwise. The interaction term DID_k,t = Treat_k × Post_t captures the policy effect.

Columns (1-3) of Table 4 report the corresponding results. The coefficient of the interaction term is positive and statistically significant at the 1% level (coefficient = 0.0989, t-statistic = 18.53 in Column 1), confirming that green finance reforms causally improve corporate GTFP. During China’s energy transition, enterprises often face significant “carbon transition risks”. Firms in pilot regions leverage green finance to signal environmental compliance and future growth potential to capital markets. This reduces information asymmetry for external investors and generates a significant “green premium”, enabling firms to obtain long-term capital at lower financing costs for long-horizon, high-risk green technological transformation projects. Regardless of whether control variables, time fixed effects, or firm fixed effects are included, the DID coefficient remains consistently positive and statistically significant. Consistent with the baseline results, this indicates that the establishment of pilot zones for green finance reform and innovations significantly enhances corporate GTFP in the treated regions.

Although Equation (1) controls for time and firm fixed effects, firms may also experience structural changes related to industry affiliation or registered location. Because these variations cannot be fully captured by only time and firm fixed effects, this study further incorporates time, firm, province, and industry fixed effects. This specification helps mitigate the influence of unobserved regional and sectoral heterogeneity on the estimation results. Columns (4-6) of Table 4 show that G-Finance maintains a positive and statistically significant coefficient (0.0141, t = 2.57), even under these more stringent specifications. These findings demonstrate that green finance significantly enhances corporate GTFP, consistent with the results in Table 3 and further supporting Hypothesis 1.

To mitigate potential selection bias in the green finance indicator system, we adopt established methodologies to construct alternative measures of green finance development^[72]. Specifically, we apply the entropy-weighting method to construct a composite index based on four components: green credit, green insurance, green bonds, and green funds. This newly constructed green finance index replaces the original explanatory variable for robustness testing. Columns (1-3) of Table 5 indicate that green finance continues to significantly enhance corporate GTFP under this alternative specification. These results are consistent with those reported in Table 3, further supporting Hypothesis 1.

Considering potential regional disparities in green finance effectiveness due to uneven economic development, we exclude China’s centrally administered municipalities (i.e., Beijing, Tianjin, Shanghai, and Chongqing) from the baseline sample. The robustness results in Columns (4-6) of Table 5 show that the estimated coefficients remain significantly positive. These findings confirm that Hypothesis 1 holds even after excluding these municipalities.

To mitigate the influence of extreme values, we apply 1% bidirectional winsorization to all continuous variables. Columns (1-3) of Table 6 indicate that green finance consistently improves corporate GTFP after winsorization.

Following Guo and Zhang^[36], we conduct a robustness test by altering the sample period. In 2016, China became the world’s largest issuer of green bonds. Therefore, we select the period from 2016 to 2021 as the new sample period for the regression analysis. As shown in Columns (4-6) of Table 6, the contribution of green finance to GTFP remains robust even with a reduced sample size.

Compared with non-state-owned enterprises, state-owned enterprises face more severe agency problems. Managers in state-owned enterprises are often more risk-averse and short-term oriented, leading them to avoid green technology innovation. In contrast, non-state-owned enterprises often lack strong government ties and fiscal support. To improve profitability and GTFP, they are therefore more incentivized to invest in green technological innovation to gain greater policy recognition and support, thereby enhancing their competitive advantage. Given that ownership structure may influence the effectiveness of green finance, this study classifies enterprises into non-state-owned and state-owned categories.

Table 8 presents the regression results by ownership type: Columns (1-6) report results for non-state-owned enterprises, and Columns (7-12) report results for state-owned enterprises. Columns (1), (2), (7), and (8) show that overall green finance development significantly promotes corporate GTFP. Furthermore, Columns (3-6) and (9-12) reveal that direct and indirect financing exert heterogeneous effects on corporate GTFP depending on ownership structure.

Specifically, the results show that: (1) market-dominated direct financing significantly promotes GTFP across all types of enterprises, highlighting the inclusive nature of market-based financing mechanisms; and (2) the effect of bank-dominated indirect financing on GTFP differs substantially between state-owned and non-state-owned enterprises. It significantly improves GTFP for state-owned enterprises but remains statistically insignificant for non-state-owned enterprises.

These differences can be explained as follows. First, as the ultimate owner of state-owned enterprises, the government enables these firms to operate under soft budget constraints. By contrast, non-state-owned enterprises often face substantial financing constraints and limited access to external capital. As a result, external financial support is particularly important for industrial upgrading and cost reduction in non-state-owned enterprises, thereby strengthening the positive impact of green finance on GTFP. Second, state-owned enterprises benefit from stronger policy support and lower operational risk. Banks typically impose higher creditworthiness requirements for green lending, which favors state-owned enterprises or large firms with lower perceived risk. This makes it difficult for non-state-owned enterprises to obtain sufficient green credit support. Consequently, bank-dominated indirect financing is less effective in enhancing the GTFP of non-state-owned enterprises.

Distinguishing between the traditional and renewable energy sectors is crucial for understanding how green finance influences corporate GTFP. High-polluting and energy-intensive production methods prevalent in the traditional energy sector pose significant challenges to sustainable development. In contrast, the renewable energy sector relies on more environmentally friendly production processes. Increased investment in the renewable energy sector is instrumental in achieving China’s low-carbon emission reduction goals, attracting both investor interest and substantial government support. This study categorizes the energy sector into two groups: traditional energy and renewable energy. Following Yu et al., we classify five industries (i.e., B06, B07, C25, D44, and D45) as the traditional energy sector, with all remaining industries classified as the renewable energy sector^[63]. The results of the grouped regression analyses are presented in Table 9.

Columns (1-6) of Table 9 represent the traditional energy sector, whereas Columns (7-12) cover the renewable energy sector. First, the results in Columns (1), (2), (7), and (8) indicate that green finance development significantly enhances corporate GTFP in both sectors, with a stronger effect observed in renewable energy firms. Second, Columns (3-6) and (9-12) show that direct and indirect financing affect GTFP of firms differently across the two sectors. Two main findings emerge: (1) market-dominated direct financing significantly enhances GTFP growth in both traditional and renewable energy firms, demonstrating strong industry adaptability within a “market synergy” framework; and (2) bank-dominated indirect financing exhibits substantial sectoral heterogeneity. Indirect financing significantly improves the GTFP of renewable energy firms but has no statistically significant effect on traditional energy firms.

This discrepancy can be attributed to China’s strong policy support for the renewable energy sector, which promotes green transition and sustainable development. The renewable energy industry benefits from its clean production characteristics, enabling it to obtain greater green finance support and more efficient allocation of financial resources toward structured green transformation initiatives. In contrast, indirect financing mechanisms, such as green credit, primarily target lower-risk projects, creating financing constraints for traditional high-pollution enterprises and limiting their ability to improve GTFP.

Table 10 presents the mediation analysis results for corporate green innovation. Columns (2) and (3) report results for total green innovation, Columns (4) and (5) focus on green innovation quality, and Columns (6) and (7) examine green innovation quantity. In Columns (2) and (3), the estimated coefficients for green finance are significantly positive, indicating that green finance development enhances both corporate green innovation and GTFP. The significantly positive coefficient for Total in Column (3) further confirms the mediating role of overall green innovation and supports Hypothesis 3a.

In Columns (4) and (5), the coefficients for green finance remain significantly positive, suggesting that it improves both green innovation quality and GTFP. Moreover, the significantly positive coefficient for Inva reported in Column (5) indicates that green finance enhances GTFP by improving the quality of corporate green innovation. Conversely, the coefficients for green finance in Columns (6) and (7) are not significant, and the coefficient for Uma is also insignificant. This finding suggests that green finance does not substantially increase the quantity of green innovation, indicating that green innovation quantity does not play a mediating role in this relationship.

Overall, the mediation test results suggest that corporate green innovation partially mediates the positive impact of green finance on GTFP. Importantly, this effect is primarily driven by improvements in the quality of green innovation rather than its quantity.

In Table 11, the dependent variable in Columns (1) and (3) is corporate GTFP, whereas the dependent variable in Column (2) is corporate carbon emissions. The results indicate that the coefficient of green finance in Column (2) is significantly negative, suggesting that green finance development reduces corporate carbon emissions. Conversely, the coefficients for green finance in Columns (1) and (3) are significantly positive, and the coefficient for Carbon in Column (3) is significantly negative, confirming the mediating role of carbon emissions. These empirical results demonstrate that green finance enhances corporate GTFP by reducing carbon emissions, thereby validating Hypothesis 3b.