Enabling Green Innovation Quality through Green Finance Credit Allocation: Evidence from Chinese Firms

Abstract

As one of the world’s largest economies and the biggest emitter of greenhouse gases, China plays a critical role in global environmental management. As China emphasizes new quality productive forces, understanding how green finance can enable green innovation quality (GIQ) is essential for projecting China’s influence in the sustainable development of the global ecological environment. This paper sets up a quasi-natural experiment using the Green Credit Policy (GCP) to examine the impact of green financial credit allocation on the enterprises’ GIQ. The findings demonstrate that the GCP has the potential to improve the GIQ of the green credit-restricted industries, compared to non-green credit-restricted ones. It is worth noting that as China speeds up its industrial digital transformation and productivity improvement, green financial credit allocation can elevate the digitization level and total factor productivity of green credit-restricted industries, leading to a higher GIQ by curbing corporate shadow banking. Further research shows that fintech and financial regulation can strengthen the positive influence of the GCP on GIQ. Moreover, regional intellectual property protection has a beneficial synergistic effect in combination with the policy.

1. Introduction

In view of the increasingly prominent impact of global climate change, the world is paying more and more attention to environmental protection. Improving the ecological environment to achieve sustainable development requires not only strong end-of-pipe measures but also the strategic mobilization of financial resources [1]. Green finance plays a pivotal role in facilitating the redirection of financial resources from polluting activities to eco-friendly initiatives and promoting significant investment in green projects. Studies by Li et al. (2018) and Haas and Popov (2019) have shown that green finance significantly enhances green innovation within enterprises globally [2,3]. As the largest developing country, China faces an urgent need to transition from traditional industrial practices to establishing a market-oriented green technological innovation system. However, there is a notable gap in China’s environmental regulatory policies concerning the integration of green finance, which is crucial for promoting high-quality green innovation and supporting environmentally sustainable development [4,5].

Green finance refers to financial activities that support and promote environmental protection and sustainable development through various financial instruments and products [6]. This type of financial activity is characterized by the integration of environmental and social responsibilities into financial decision-making, with the main objective of taking environmental, social and governance (ESG) factors into account in financial activities in order to promote sustainable economic growth. Through capital guidance, risk management, incubation and innovation, information disclosure and policy support, green finance attempts to promote low-carbon, environmentally and socially responsible investment and financing for sustainable development. However, green finance involves a wide range of fields, mainly including green credit, green bonds, green insurance, carbon finance, etc., and it is difficult to explore the role of green finance in a comprehensive manner in all fields. The former China Banking Regulatory Commission (CBRC) formulated the Guidelines on Green Credit in 2012, which is the first normative document of China dedicated to green credit. The Guidelines require banking financial institutions to promote green credit at a strategic level and effectively control environmental and social risks in credit business activities [5]. Distinguishing from environmental regulatory policies characterized by administrative penalties, the Guidelines aim to guide green credit through the allocation of credit resources to improve the green innovation of enterprises and promote green transformation of the economy, which also provides a good perspective for us to analyze the role of green finance from the perspective of green credit. According to the estimation of China’s Industrial Bank Research, the balance of green credit accounts for more than 90% of the balance of all green financing. And China’s State Administration of Financial Supervision and Administration states that, by the end of 2023, the green credit balance of China’s 21 major banks reached 27.2 trillion yuan, up 31.7% year-on-year, ranking first in the world in terms of scale. In the same period, the balance of green bonds in China was only RMB 3.62 trillion, less than 14% of the scale of green credit, and the scale of other green financing methods was even smaller. Therefore, it is plausible that the findings based on green credit can represent the role of green finance to a large extent. Thus, this paper takes the Guidelines issued by the former CBRC as the entry point and use the Differences-in-Differences (DID) method to assess the implementation effect of GCP. The objectives are as follows: firstly, to try to explore the logic of the influence between green finance and green innovation; secondly, to explore the intrinsic mechanism by which GCP influences corporate green innovation; and thirdly, to provide useful policy recommendations for promoting high-quality green innovation and supporting the environmentally sustainable development of enterprises. Specifically, considering that the Guidelines are committed to guiding the flow of funds to environmentally friendly and sustainable projects, promoting the green transformation of the economy and preventing environmental and social risks, this paper tries to analyze the micro impact of the GCP from the perspective of the GIQ.

Currently, two contrasting perspectives have surfaced from research on the influence of green finance on green innovation. One viewpoint suggests that green finance helps to increase corporate R&D spending [7], enhances the effectiveness of green innovation within enterprises [8] and leads to high-quality enterprise development. Simultaneously, it can reduce corporate agency costs, improve investment efficiency and act as credit restraints to drive green innovation among heavily polluting industries, thereby enhancing corporate environmental performance and firms’ green innovation performance [5,9]. On the other hand, an alternative perspective implies that green finance may aggravate the financing constraint faced by heavily polluting industries [10], raising the cost of credit financing for enterprises [11] and impeding green innovation. Specifically, the impact of green finance on the green innovation behavior of listed companies is detrimental, particularly in hindering corporate credit financing [12]. Additionally, green finance may intensify financial regulation challenges and diminish the effectiveness of financial services, thereby undermining the promotion of eco-friendly practices within the corporate sector [13]. However, from the perspective of the GIQ, this paper finds that GCP can enhance the digitization level and total factor productivity of green credit-constrained industries, and enhance the GIQ by suppressing corporate shadow banking. In this process, fintech and financial regulation can strengthen the contribution of GCP to the GIQ, and there is a positive synergy between regional intellectual property protection and GCP.

In order to investigate the influence of GCP on green innovation, this paper employs the “Green Credit Guidelines” introduced in China in 2012 as a quasi-natural experiment. To identify whether a listed company falls within a green credit-restricted industry, the study ascertains the company’s industry based on the presence of environmental and social risks classified as category A in the key evaluation indicators for green credit implementation. If a listed company is associated with an industry falling under category A, it is recognized as a green credit-restricted industry; otherwise, it is classified as a non-green credit-restricted industry. This paper creates an interaction term involving the policy time dummy variable and the grouping dummy variable as the principal explanatory variable. This variable primarily assesses the impact of GCP on the green innovation of firms subject to green credit restrictions and those that are not, both before and after the implementation of GCP. It is important to note that current methods for measuring enterprise green innovation are somewhat limited, making it challenging to fully capture the diverse nature of corporate environmental improvement. The existing studies have mainly measured corporate green innovation through metrics, such as the quantity of green patents [5,14]. However, as China’s economy transitions from quantity-driven growth to a focus on qualitative enhancements, the emphasis is no longer solely on the sheer volume of patents, but rather on the significance of enterprises’ commitment to green innovation. To address this shift, this paper utilizes the WIPO Green Patent List’s patent classification system and applies the Knowledge Breadth Method to assess the enterprises’ GIQ. By doing so, this paper aims to gauge the extent of green innovation within these enterprises, thereby addressing the limitations of relying solely on patent numbers to measure innovation activities. This approach ultimately yields a more accurate reflection of enterprises’ achievement in the realm of green innovation.

After establishing the indicators mentioned above, this paper discovers that GCP has a substantial impact on driving green innovation within green credit-constrained enterprises. This finding remains robust even after addressing several potential endogeneity issues and conducting thorough tests to validate the results. The findings of the impact mechanism test indicate that green credit effectively curbs corporate shadow banking and leads to spillover effects in digital technology, thereby facilitating the transformation of corporations towards green practices. Simultaneously, green credit also enhances the total factor productivity of enterprises, which in turn supports and encourages the transition towards greener operations.

Meanwhile, in the interconnected landscape of green finance and green technology innovation, fintech has the potential to enhance the supportive role of green finance in empowering entities [15]. Findings indicate that fintech can notably bolster green credit’s efficacy in promoting the greening of enterprises. However, the implementation of green credit may also add complexity to financial institutions’ portfolios and pose challenges to regulatory oversight, leading to increased difficulties in financial regulation. This paper also explores the moderating impact of financial regulation on the dynamics [13]. In order to enhance the precision of the regression results, this paper adjusts the financial regulation index to the city level based on the concentration of urban commercial banks, given that the financial regulation data are currently only available at the provincial level. The findings indicate that financial regulation reinforces the positive impact of fintech. Concurrently, the enhancement of the intellectual property protection system has been shown to bolster enterprises’ innovation capabilities, thereby facilitating technological advancements and industrial upgrades and effectively mitigating the constraints related to R&D investment and innovation risks [16]. Furthermore, this paper explores the influence of the intellectual property protection system and establishes its role in bolstering green innovation within firms that are both enabled and constrained by green credit.

In conclusion, this paper makes a meaningful contribution in three primary areas. Firstly, this paper provides a fresh perspective on green finance for promoting the high-quality development of green innovation. Instead of solely relying on the quantity of green patent applications or authorizations, as commonly performed in existing studies, this paper introduces the knowledge breadth approach to assess the enterprises’ GIQ. This method ensures a more objective and accurate measurement, important for China’s progression towards high-quality development. The second contribution lies in the comprehensive examination of the impact mechanism of green credit on enterprises’ green innovation. In light of China’s policy aim of digitization and innovation, this paper scrutinizes the influence of green credit through the lenses of commercial credit, digital technological innovation and total factor productivity. Lastly, this paper addresses the impact of fintech and financial regulation. By introducing city-level fintech and financial regulation indicators, the study explores how these factors can strengthen the role of green finance in promoting entity green innovation. This innovative approach digs deeper into the factors that influence the GIQ in companies.

The paper is structured as follows: the second part comprises the theoretical analysis and research hypotheses. The third and fourth parts detail the research design and the analysis of empirical results, respectively. The fifth part explores the impact mechanism, followed by the sixth part, which presents a further discussion. The final section contains the conclusions, policy recommendations and limitations.

2. Theoretical Analysis and Research Hypotheses

2.1. Green Finance Credit Allocation and Enterprises’ GIQ

Green financial credit allocation primarily occurs through bank credit channels to facilitate the efficient distribution of funds. This makes it more feasible for enterprises with a focus on energy conservation and environmental protection, as well as those engaged in green production activities, to secure loans and access more convenient and open financing channels [12]. Additionally, the reduced financing costs allow for the allocation of funds to green production initiatives [5]. The “Green Credit Guidelines” established by the former CBRC in 2012, provides valuable insight for studying green financial credit allocation. Within this system, commercial banks consider the environmental and social risks of their clients an essential factor in their ratings, credit approvals, management and disbursement. Commercial banks also identify industries subject to green credit restrictions. Poor environmental performance, high energy consumption and excessive pollution emissions make it less likely for a company to receive credit, resulting in higher financing expenses [17]. From a perspective of long-term and sustainable development, green credit restrictions gradually prompt industries to address short-term negative impacts. The constant reduction of environmental pollution costs, the mitigation of environmental risks and the acceleration of green transformation processes are vital [18]. Consequently, GCP may compel enterprises in green-credit-restricted industries to engage in green innovation initiatives. Simultaneously, as green finance fortifies banks’ monitoring functions concerning fund usage and possesses enduring credit monitoring, more enterprises shift from quantitative to substantive innovation [19], to bolster their standing in credit allocation. Accordingly, research hypothesis H1 can be posited:

Hypothesis 1.

Green financial credit allocation exerts a positive motivational effect on the GIQ in green credit-constrained industries.

2.2. Path Analysis of Green Financial Credit Allocation Affecting Enterprises’ GIQ

2.2.1. Based on Enterprise Digitization Level

The primary goal of green finance credit allocation is to boost financial support for environmentally responsible enterprises meeting green finance recognition standards, while concurrently reducing financing for heavy polluters. This ensures a reallocation of financial resources towards sustainable initiatives [20]. To secure additional financial backing, businesses in green-credit-restricted industries are motivated to invest in the research and development of green and clean technologies to diminish their ecological footprint [21]. Digital technology plays a pivotal role in steering companies away from traditional high-input, high-output, high-energy consumption and high-pollution production methods, toward low-carbon, energy-efficient production modes. This shift allows enterprises to enhance production efficiency by leveraging new technologies while addressing the negative environmental impact of their operations, ultimately achieving their own green transformation [22]. On the one hand, digital technology can effectively improve the production process and improve the efficiency of equipment operation, making the production process more green and low-carbon. In this way, enterprises can ensure the quality of production at the same time, but also reduce the pollution of the environment, to achieve the effect of energy saving and emission reduction [23]. On the other hand, the digital economy can optimize the resource allocation model. Through the digital infrastructure in the fields of industrial internet, big data, artificial intelligence, etc., various resource elements can be integrated and shared among different industries and enterprises. This not only enhances the efficiency of resource allocation, but also prompts enterprises to utilize resources more rationally, reduce waste and further promote green transformation [24]. Striving to navigate the pressures stemming from GCP, companies in green credit-restricted sectors may further enhance their digital technology capabilities and prioritize the synergy between production and environmental considerations, ultimately contributing to ecological improvement and heightened resource allocation efficiency [25]. Moreover, the strategic use of digital technology can assist heavy polluters in mitigating external environmental pressures. Based on this reasoning, the research hypothesis H2a is formulated:

Hypothesis 2a.

Green finance credit allocation has the potential to enhance the enterprises’ GIQ through the advancement of digitization in green-credit-restricted industries.

2.2.2. Based on Total Factor Productivity of Enterprises

GCP plays a crucial role in guiding the optimal allocation of resources between polluting and green industries and in enhancing the efficiency of resource allocation within green-credit-restricted industries [26]. In this context, less productive firms have an opportunity to access production resources at lower costs, allowing them to compete more effectively in the market. However, this could inhibit the flow of resources to high-productivity firms, thus limiting the market space for inefficient firms and enabling dynamic adjustments in firm size, leading to improved resource allocation efficiency [27]. Consequently, GCP can motivate enterprises to boost their total factor productivity through optimized resource allocation. On the one hand, an increase in total factor productivity means that firms are able to use resources more efficiently, and through optimization of the production process and technological innovation, firms can achieve the more efficient use of energy and reduce the loss of resources [28]. On the other hand, an increase in total factor productivity is usually accompanied by an increase in technological innovation and investment in research and development. Such innovation not only helps enterprises to improve product quality and value added, but also promotes the transition to green production and green products [29]. Overall, an increase in total factor productivity not only reduces the consumption of raw materials and the generation of waste, which in turn reduces the negative impact on the environment, but also provides enterprises with more resources and surplus funds so that they can increase their investment in green innovation activities and continuously improve the GIQ driven by efficiency [30]. Accordingly, research hypothesis H2b can be formulated as follows:

Hypothesis 2b.

Green finance credit allocation has the potential to enhance the enterprises’ GIQ by increasing total factor productivity in green-credit-constrained industries.

2.2.3. Based on the Shadow Banking of Enterprises

GCP typically compels financial institutions to thoroughly assess the environmental track records of borrowers. If borrowers are found to have engaged in serious environmental polluting activities, they may face higher loan interest rates and more stringent loan terms [17]. Consequently, companies operating in sectors with limited access to green credit may turn to shadow banking to shore up their liquidity [31]. However, the shadow banking system is characterized by high leverage, significant information asymmetry and ambiguous legal entities, rendering it more risky and potentially exposing firms to liquidity dilemmas and bankruptcy risks [32]. This hinders their ability to focus on core business development and dampens the prospects for green innovation. Nevertheless, in China’s strategic environment aims at preventing real enterprises from engaging in “de-realization” and promoting the green transformation of the manufacturing industry, green finance credit policies may compel enterprises in industries with restricted green credit access to refocus on their core business and pursue green credit support through green innovation [5]. Therefore, the influence of green finance credit policies on whether shadow banking is practiced in green-credit-constrained industries remains uncertain. As a result, the research hypothesis H2c can be formulated as follows:

Hypothesis 2c.

Green finance credit policies may impact the GIQ in firms operating in sectors with limited green credit through the practice of corporate shadow banking, but this effect is subject to uncertainty.

2.3. Analysis of Factors Influencing the Allocation of Green Financial Credit to Affect Enterprises’ GIQ

2.3.1. Based on Financial Regulation

The efficiency of green finance credit allocation may be influenced by various financial regulatory factors [33]. Specifically, financial regulators have the power to standardize the management regulations, auditing and assessment procedures for green credit business within financial institutions. They can also supervise and evaluate the implementation of GCP within these institutions, ultimately playing a key role in regulating green innovation [13]. Financial regulators can encourage financial institutions to actively and effectively engage in green financial business and investment, ensuring the quality of green credit and the feasibility of the regulatory mechanism, while also safeguarding the legitimacy of the regulation and promoting the development of enterprise green innovation [26]. Through financial supervision, regulations can effectively govern the behaviors of financial institutions and enterprises, ultimately supporting the stability and sustainable growth of the market. The regulation of green financial credit allocation not only helps prevent the accumulation of non-performing assets and risks but also guarantees the transparency of green credit allocation and fair competition and facilitates the establishment of a risk-prevention mechanism [34]. Therefore, the research hypothesis H3a can be stated as follows:

Hypothesis 3a.

Financial regulation can enhance the impact of green finance credit allocation on promoting the GIQ in enterprises.

2.3.2. Based on Local Fintech Development

Additionally, the use of fintech can significantly increase the efficiency of the pre-credit review and post-credit risk management for green credits [35], ultimately facilitating access to and the effective utilization of funds for environmental projects by enterprises. Fintech can collect business information from multiple channels and screen the credit needs of enterprises engaged in green innovation [36,37], enhancing the allocation efficiency of green credit. Furthermore, fintech, through means, such as blockchain and big data, can effectively reduce post-loan moral hazards and improve the ability of financial institutions to prevent risks [38]. Fintech can also better manage the destination of corporate credit through technological means, effectively controlling the operational and financial risks of enterprises and providing a stable environment for innovative activities [38]. Therefore, research hypothesis H3b can be stated as follows:

Hypothesis 3b.

Fintech can strengthen the positive impact of green financial credit allocation on the enterprises’ GIQ.

2.3.3. Based on Regional Intellectual Property Protection

Protecting intellectual property is a crucial tool for driving innovation, nurturing economic growth and ensuring fair competition, thereby offering financial returns and competitive advantages in the market. Additionally, intellectual property protection plays a vital role in advancing technological progress and fostering industrial development. A robust judicial framework for safeguarding intellectual property rights constitutes a critical institutional foundation for strategies promoting innovation growth [39]. Intellectual property infringement can be a serious impediment to R&D, and the strengthened enforcement of intellectual property protection by the government can enhance firms’ ability to innovate, mainly by reducing R&D spillover losses and easing external financing constraints [40]. Accordingly, the research hypothesis H3c can be formulated:

Hypothesis 3c.

Regional intellectual property protection can magnify the positive impact of green finance credit allocation on the enterprises’ GIQ.

3. Research Design

3.1. Sample Selection and Data Sources

Since China’s policy focus before 2007 tended to be in the traditional financial sector, rather than the green financial sector, this paper selects China’s A-share listed companies from 2007–2021 as the research sample, which can more accurately reflect the impact of the current GCP on corporate green innovation. Due to data quality issues, reference is made to Jing et al. (2021) [9], which excludes the data of 2017 and treats 2016 and 2018 as consecutive years. The data of listed companies originate from CSMAR and WIND databases, and the data of corporate patents originate from the WIPO Green Patent List. Provincial-level financial regulation data are obtained from the National Bureau of Statistics of China, and its provincial statistical yearbooks and bulletin data. In order to comprehensively measure the level of financial regulation at the city level, this paper manually organizes the data on the number of bank financial institutions’ outlets in Chinese cities and uses the ratio of the number of bank outlets in cities to the number of bank outlets in the province in the current year to construct the weights at the city level and then multiplies the weights by the financial regulation index at the provincial level to measure the level of financial regulation at the city level. To comprehensively measure the level of fintech, this paper manually collates the number and distribution of commercial banks in each city in China, which is obtained from the financial license information of the China Banking Regulatory Commission (CBRC). At the same time, the city AI agglomeration index is measured through a specific search of AI companies by Tianyancha. This paper collects and organizes 10,572 “company-year” observations. In order to ensure the feasibility of the study, on the basis of the original samples, this paper makes the following treatments to the raw data: first, 1 sample of financial enterprises is excluded; second, 266 samples of enterprises with a delisting risk are excluded; third, the number of enterprises with a delisting risk is excluded, and 939 samples with missing main variables are removed; fourth, remove 1 sample with insolvency and a book value of shareholders’ equity less than zero; fifth, perform a two-sided 1% shrinkage of all continuous variables to avoid the impact of outliers on the empirical results and ultimately obtain 10,572 “company-year” observations.

3.2. Model Establishment and Variable Setting

3.2.1. Model Establishment

In reference to the work by Xin and Ying (2021) [5], this paper utilizes the 2012 Green Credit Guidelines formulation as a quasi-natural experiment to examine green finance credit allocation and constructs the model in the following manner:

$$Ln\left({Patent}_{i,t}+1\right)={\alpha }_t+{\alpha }_i+{\beta }_1{Policy}_t+{\beta }_2{Gcres}_i\ast {Policy}_t+{\beta }_3{Gcres}_i+{\gamma }^{\prime }{X}_{i,t-1}+{\epsilon }_{i,t}$$

(1)

The explanatory variable “Ln(Patent + 1)” represents the firms’ GIQ. The main explanatory variables include the GCP (Policy), industry attributes (Gcres) and the interaction term between the two (Policy × Gcres). Control variables are denoted as X. Considering that innovative behavior may have cyclical effects, the control variables are lagged by one period. ε_i,t represents the random error term, where subscript i refers to the firm and t refers to the year.

3.2.2. Variable Setting

(1)

Explained Variable

The GIQ (Patent). Building on the work of Jie and Wenping (2018) [41], based on the patent classification numbers provided by the WIPO Green Patent List, the Knowledge Breadth Method is utilized to calculate the quality of green patents of enterprises and to measure their green innovations, which, to a certain extent, can overcome the shortcomings of reflecting the innovation activities of enterprises only by the number of patent applications:

$${Patent}_{n,t}=1-\mathsf{\Sigma }{\alpha }^2$$

(2)

In model (2), this paper uses Patent_n,t to represent the breadth of knowledge associated with various types of patents, serving as a proxy for the quality of environmental innovation within enterprises. Here, “n” and “t”, respectively, refer to the patent and the year, and α denotes the proportion of major group classifications within the patent classification number. The IPC classification number format in the patent documents of Chinese enterprises at the State Intellectual Property Office typically follows the structure “Department-Major Class-Minor Class-Major Group-Group”, such as “F24F11/00”. The first letter of the classification number spans A-H, representing eight major departments; the second and third digits indicate the major categories; the fourth letter denotes the minor categories, with major and minor groups separated by ‘/’. For example, one patent may have classification numbers F24F11/00, F24F11/10, and F24F11/20, while another patent similarly has F24F11/00, F24F12/00, and F24F13/20. Although the two patents share the same number of classification codes, they differ in that the first patent utilizes only F24F11 as major group information, whereas the second patent encompasses F24F11 as minor group information and information for three different major groups. According to the calculation rule of model (2), this indicates that the breadth of knowledge applied in the second patent exceeds that of the former. Hence, the greater the diversity between the patent classification numbers at the major group level, the wider the knowledge scope, reflecting a higher patent quality. To address the right-skewed distribution issue of green patent data, this paper employs the natural logarithm of green patent quality after adding 1 to obtain Ln(Patent + 1).

(2)

Core Explanatory Variables

The main explanatory variables under consideration are the GCP, industry characteristics and their cross-multiplier. Specifically, the variable “Policy” functions as a dummy variable, indicating the period before and after the implementation of the Guidelines. Following the implementation (2012 and beyond), the value of “Policy” becomes 1; otherwise, it remains 0. Gcres denotes the industry classification for the implementation of the GCP specified by the Guidelines. In this paper, the industry category to which the companies with environmental and social risks fall under in the Key Evaluation Indicators for Green Credit Implementation determines whether the listed company belongs to the green-credit-restricted industry. If the company falls under category A, it is labeled as a green-credit-restricted industry (Gcres = 1); if not, it is classified as a non-green-credit-restricted industry (Gcres = 0). The interaction term “Policy × Gcres” primarily assesses the impact of the GCP on green innovation within both the green-credit-restricted and non-green-credit-restricted industries before and after the policy’s implementation. A significantly positive coefficient of the cross-multiplier term, β₂, indicates the substantial advancement of green innovation in green-credit-constrained industries due to the GCP. Conversely, a non-significant coefficient suggests the lack of a significant promotional effect.

(3)

Moderator Variables

Financial regulation: Existing research often uses the ratio of regional financial regulatory expenditure to the value added by the financial sector as a measure of local financial regulation. Additionally, this paper manually organizes the data of urban bank financial institution outlets across the country, constructs city-level weights with the ratio of the number of urban bank outlets to the number of bank outlets in the province in the current year and then multiplies the financial regulatory index at the provincial level to obtain the regulation of financial regulatory indicators in each city.

Fintech: This paper leverages the approach taken by Fei et al. (2021) [42]. Initially, this paper utilizes Python tools to scrape 48 keywords associated with “fintech” from relevant news articles and conferences. Subsequently, this paper employs a Baidu News Advanced Search to look for news pages containing these keywords in conjunction with cities and municipalities. By crawling the source code of the Baidu News Advanced Search and extracting the number of search results, this paper aggregates a total of 254,456 searches containing “region + keywords” at the prefecture-level city or municipality directly under the central government. Next, this paper manually sorts the number of branches of each bank in each year and city using financial license information from the China Banking Regulatory Commission (CBRC) to construct the Herfindahl Index (HHI) for measuring bank competition. Additionally, this paper aggregates the number of AI enterprises at the city and provincial levels and applies the location entropy index to gauge the AI enterprise concentration at the regional level. Multiplying these three indexes yielded the comprehensive fintech evaluation index. To address right skewness and prevent missing values, this paper processes the index by adding one and taking the logarithm. Consequently, this paper obtains the fintech-level measure at the city level (Ln(fintech + 1)).

Intellectual property protection: This paper draws on Shengchao’s (2023) approach [43], which first divides the number of intellectual property protection infringement cases filed in each region by the total population, then divides the number of lawyers in the region by the total population and calculates the average of the two, while taking the logarithm to characterize the level of regional intellectual property protection (Lnprotect).

(4)

Control Variables

Based on previous literature [5,44], this paper incorporates the following control variables X_{i, t−1}: firm size (LnSize), expressed as the logarithm of the total assets of the firm; gearing ratio (LEV); net profitability of total assets (ROA); return on equity (ROE); accounts-receivable-to-revenue ratio (REC); dual chairmanship and CEO position (Dual); the percentage of ownership of the first largest shareholder (LnTop1); book market capitalization ratio (BM); firm value (LnTQ), expressed as the logarithm of the enterprise value multiple. Descriptive statistics for all variables are shown in

Table 1. Descriptive statistics of variables (2007–2021).

Variables	Obs	Mean	Median	P10	Std. Dev.	Min	Max
Ln(Patent + 1)	10,572	0.422	0.486	0.000	0.185	0.000	0.649
LnSize	10,572	3.206	3.232	3.083	0.085	3.030	3.350
LEV	10,572	0.539	0.567	0.352	0.137	0.130	0.786
ROA	10,572	0.043	0.040	0.006	0.034	−0.027	0.136
ROE	10,572	0.099	0.085	0.015	0.080	−0.060	0.337
REC	10,572	0.184	0.134	0.024	0.180	0.000	0.883
Dual	10,572	0.379	0.000	0.000	0.485	0.000	1.000
LnTop1	10,572	3.267	3.229	2.448	0.638	1.973	4.458
BM	10,572	0.771	0.795	0.419	0.242	0.225	1.220
LnTQ	10,572	2.776	2.696	2.084	0.547	1.749	4.466

Notes: Obs means observations, P10 denotes decile and Std. Dev. denotes standard deviation. The data set covers 15 years from 2007 to 2021.

.

4. Empirical Analysis

4.1. Baseline Regression

Table 2. Benchmark regression results.

Model	(1)	(2)
	Ln(Patent + 1)	Ln(Patent + 1)
DID	0.078 ***	0.058 ***
	(5.58)	(4.03)
Policy	−0.060 ***	0.003
	(−9.22)	(0.06)
Gcres	−0.065 ***	0.238 **
	(−4.67)	(2.08)
LnSize	0.276 ***	0.276 ***
	(7.22)	(5.08)
Ln(LEV + 1)	0.077 ***	0.025
	(3.42)	(0.96)
Ln(ROA + 1)	0.744 ***	−0.144
	(3.36)	(−0.57)
Ln(ROE + 1)	−0.217 **	0.086
	(−2.48)	(0.87)
Ln(REC + 1)	−0.059 ***	−0.074 ***
	(−4.73)	(−4.89)
Dual	0.010 **	0.028 ***
	(2.52)	(5.55)
LnTop1	−0.043 ***	−0.035 ***
	(−11.33)	(−6.44)
LnBM	−0.008	0.019
	(−0.61)	(1.08)
LnTQ	0.012 **	0.004
	(1.98)	(0.48)
Year	No	Yes
Sector	No	Yes
Area	No	Yes
N	10,572	10,572

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

displays the findings from the benchmarking study on the impact of green credit on firms’ green innovation. One of them, the significance level, is a criterion used in hypothesis testing to determine whether a statistic is significant or not. If the t-statistic of the regression coefficient is greater than the critical value, this paper rejects the original hypothesis (the independent variable is considered to have a significant effect on the dependent variable); otherwise, the original hypothesis is accepted. And the regression coefficient indicates the magnitude and direction of the effect of the independent variable on the dependent variable. In columns (1) and (2), the coefficient of the cross-multiplier term DID shows a highly significant positive correlation at the 1% level. After incorporating fixed effects, the resulting coefficient still indicates a notable increase in the GIQ within the green-credit-restricted sectors following the implementation of the GCP, underscoring the significant boost provided by the Guidelines in enhancing green innovation output in this sector. The reason may be that the main objective of the GCP is to curb the blind expansion of high-energy-consuming and high-polluting industries, and therefore, its policy design and coverage mainly focuses on the restriction of high-polluting industries, which supports the validity of Hypothesis 1.

4.2. Robustness Tests

4.2.1. Parallel Trend Test

Figure 1 presents the analysis of the impact of the GCP on corporate green innovation over time. Prior to the policy implementation, the estimated coefficient β within the 95% confidence interval shows no significant deviation from 0. However, one period after the policy shock, the coefficient becomes significantly different from 0, suggesting a delayed response to the policy implementation. Subsequent to this, there is a noticeable divergence in trends between the treatment group and the control group, particularly after the t + 1 period, indicating a positive promotion of enterprise green innovation by the GCP. Additionally, the parallel trend test confirms these findings.

4.2.2. Placebo Test

To test the reliability of the empirical findings, this paper introduces the placebo method to assess the robustness of the impact of the GCP. Following the guidelines, this paper randomly selects nine industries as the “pseudo-treatment group” (Gcresfalse), from which this paper constructs the dummy variable Patentfalse = Gcresfalse × Policy for the placebo test. This experiment was then repeated 1000 times, resulting in the p-value test plot shown in Figure 2. For the p-value test plot, the results show that the coefficient estimates are clustered around 0 and approximately follow a normal distribution, indicating that the regression results are not affected by unobservable factors and the results are more robust.

4.2.3. Replacement of Explanatory Variables

This paper draws on the approach of Jie and Wenping (2018) [41], extracting green invention patent data based on the WIPO Green Patent List, and calculating the green invention patent quality of the enterprises according to the provided patent classification number by using the model (2), and it takes the natural logarithm after adding 1 to obtain the green invention patent quality Ln(lnva + 1). The reason is because the application and authorization of invention patents require a longer investment cycle and higher maintenance costs, and the examination process is more rigorous. Therefore, a green invention patent can better reflect the quality of a company’s innovation. The results are shown in

Table 3. Robustness test based on proxy variables for corporate green innovation.

Model	(1)	(2)
	Ln(lnva + 1)	Ln(lnva + 1)
DID	0.056 ***	0.043 **
	(3.08)	(2.41)
Policy	−0.040 ***	0.036
	(−4.50)	(0.24)
Gcres	−0.100 ***	−0.329 **
	(−5.46)	(−2.22)
Control variable	Yes	Yes
Year	No	Yes
Sector	No	Yes
Area	No	Yes
N	8717	8684

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

, where the regression coefficients of DID are significantly positive at least at the 5% level, which indicates that the policy could also promote a higher degree of green innovation, which emphasizes environmentally sustainable technological solutions and are considerably more technical and innovative.

4.2.4. Change in Industry Definition Criteria

The Key Evaluation Indicators for Green Credit Implementation indicate that, aside from category A industries, category B industries also have adverse effects on the environment and society. Therefore, this paper broadens the identification of green credit restricted industries, including category B industries. Furthermore, by referencing the “Listed Company Environmental Verification Industry Classification Management Directory” and “Listed Company Environmental Information Disclosure Guidelines” and in conjunction with the “Guidelines for Industry Classification of Listed Companies”, this paper identifies the mining industry (industry code: B06, B07, B08, B09), manufacturing industry (industry code: C17, C19, C22, C25, C26, C28, C29, C30, C31) and polluting firms in electricity, heat, gas and water production and supply (industry code: D44) as the experimental group. Non-polluting firms, after removing green firms, are used as the control group. The regression results in

Table 4. Robustness test based on two criteria for the changing industry definition.

Model	(1)	(2)
	Ln(Patent + 1)	Ln(Patent + 1)
DID	0.048 ***	0.048 ***
	(3.67)	(3.59)
Policy	−0.019	−0.018
	(−0.45)	(−0.43)
Gcres	0.486 ***	0.317 *
	(3.53)	(1.93)
Control variable	Yes	Yes
Year	Yes	Yes
Sector	Yes	Yes
Area	Yes	Yes
N	10,572	10,572

Notes: *** and * indicate significance at the 1% and10% levels, respectively.

show that the coefficients of the DID are all significantly positive at the 1% level, consistent with the regression results in Table 2, indicating robust results for the first and second changes in industry definition criteria.

4.2.5. Tests Based on the PSM-DID Methodology

First, green innovation by heavily polluting firms is often influenced by policy interventions, such as environmental regulations, making it difficult to directly estimate the policy effects of green credit. Second, if heavily polluting firms themselves have high green innovation capabilities, this may also lead to biased estimates of policy effects. Therefore, to mitigate systematic differences between heavily polluting industries and other sectors and minimize estimation bias in the double-difference method, this paper employs the PSM-DID method for robustness testing. This involves conducting a logit regression of control variables using a dummy variable to determine if an industry is heavily polluting, in order to obtain the propensity score value. The matching industry for the heavily polluting industry is identified as the industry with the closest propensity score value, effectively minimizing systematic differences and reducing the DID estimation bias. This paper specifically utilizes the kernel matching method for estimation to assess the robustness of the GCP’s role in promoting firms’ green innovation.

Table 5. Robustness test based on PSM-DID.

Model	(1)	(2)	(3)
	Ln(Patent + 1)	Ln(Patent + 1)	Ln(Patent + 1)
	Before	After	DID
Increment	−0.060	0.003	0.063
Standard error	0.011	0.007	0.012
T-value	−5.250	0.450	5.320
p-value	0.000 ***	0.655	0.000 ***

Notes: *** indicates significance at the 1% level.

presents the regression results, which demonstrate that even after applying the PSM-DID methodology, the GCP consistently and significantly stimulates firms’ green innovation activities. Additionally, the firms’ GIQ increases by 6.3%, aligning with the previous results and indicating robustness.

4.2.6. The 2008 International Financial Crisis and the Impact of Omitted Variables

Due to potential impacts from the 2008 international financial crisis and omitted variables, this paper omits the 2008 data sample and introduces a new variable (Punish) in model (1) to represent environmental penalties, where Punish = 1 signifies that firms are subject to such penalties and Punish = 0 signifies that firms are not.

Table 6. Robustness tests based on excluding the 2008 sample and environmental penalties.

Model	(1)	(2)
	Ln(Patent + 1)	Ln(Patent + 1)
DID	0.044 ***	0.048 ***
	(3.29)	(3.11)
Policy	0.031	0.103 **
	(0.70)	(2.08)
Gcres	0.490 ***	0.213
	(3.56)	(1.49)
Punish	—	−0.002
	—	(−0.11)
Control variable	Yes	Yes
Year	Yes	Yes
Sector	Yes	Yes
Area	Yes	Yes
N	10,507	10,572

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

presents the results, with column (1) reporting the regression results excluding the 2008 data sample and column (2) reporting the results introducing environmental penalties. The regression coefficients of DID are significantly positive at the 1% level, aligning with the robust results displayed in Table 2.

4.2.7. Environmental Pollution Factors and Replacement Sample Year Intervals

In order to account for the impact of concurrent environmental policies, this paper incorporates three variables based on Xin and Ying (2021) [5]. The first variable is the annual PITI index, which measures the disclosure of pollution source regulatory information in the location where the firm is registered. The second variable is the distance, which represents the shortest distance between the registered location of the listed company and neighboring state-controlled air monitoring stations. To address the issue of skewed distribution, the natural logarithm of the nearest distance plus 1 (LnDistance) is used. The third variable assesses the level of haze pollution (PM2.5) in the location of the listed companies. The paper focuses on Chinese listed companies from 2007 to 2021, treating the 2012 introduction of the Green Credit Guidelines as a quasi-natural experiment to examine its impact on corporate green innovation. To minimize the influence of subsequent policies, the sample is restricted to 2010–2013, based on the methodology of Jie et al. (2022) and Juncheng et al. (2023) [45,46].

Table 7. Robustness test based on environmental pollution factors and replacement sample year intervals.

Model	(1)	(2)	(3)	(4)
	Ln(Patent + 1)	Ln(Patent + 1)	Ln(Patent + 1)	Ln(Patent + 1)
DID	0.050 ***	0.058 ***	0.068 ***	0.075 ***
	(3.06)	(4.07)	(4.66)	(3.04)
Policy	−0.036	0.006	−0.035	0.047
	(−1.44)	(0.14)	(−0.84)	(0.87)
Gcres	−0.048	0.454 ***	0.222 *	0.308 ***
	(−0.27)	(8.01)	(1.93)	(2.87)
PITI	−0.001 *	—	—	—
	(−1.89)	—	—	—
Lndistance	—	−0.002	—	—
	—	(−0.65)	—	—
PM2.5	—	—	0.003 ***	—
	—	—	(5.44)	—
Control variable	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Sector	Yes	Yes	Yes	Yes
Area	Yes	Yes	Yes	Yes
N	6642	10,677	10,085	9487

Notes: *** and * indicate significance at the 1% and 10% levels, respectively.

presents the regression results, showing that the regression coefficients of DID are significantly positive at the 1% level, consistent with previous robust results and supporting hypothesis H1.

5. Analysis of Impact Mechanisms

5.1. Mechanism Testing Based on Digital Transformation

Based on the primary classification number of patents matched with the Statistical Classification of Digital Economy and Its Core Industries, the number of digital economy patent applications filed by listed companies in that year is obtained. These data are then logarithmized to derive Ln(Innovate + 1), which measures the level of digitalization of enterprises. To assess the reasonableness of the mediating effect, the Bootstrap method is employed to sample 500 times, yielding an estimated value of ab of Ln(Patent + 1) at 0.008, with a 95% confidence interval of (0.003, 0.014). As the confidence interval does not encompass 0, and the two-tailed test is significant (p = 0.002), it is indicative of a mediating effect with Ln(Innovate + 1) as the conduction variable. In

Table 8. Mechanism test based on digital transformation.

Model	(1)	(2)	(3)
	Ln(Patent + 1)	Ln(Innovate + 1)	Ln(Patent + 1)
DID	0.058 ***	1.410 ***	0.045 ***
	(4.03)	(19.37)	(3.10)
Policy	0.003	0.148 ***	−0.038
	(0.06)	(5.45)	(−0.91)
Gcres	0.238 **	−0.611 ***	0.432 ***
	(2.08)	(−2.77)	(6.96)
Ln(Innovate + 1)	—	—	0.005 **
	—	—	(2.40)
Control variable	Yes	Yes	Yes
Year	Yes	Yes	Yes
Sector	Yes	Yes	Yes
Area	Yes	Yes	Yes
N	10,572	10,372	10,372

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

, column (1) represents the total effect of green finance credit allocation on firms’ green innovation. Column (2) demonstrates the effect of green financial credit allocation on firms’ digital transformation, with the regression coefficient significantly positive at the 1% level, suggesting that green credit can promote the digitalization level of green-credit-constrained industries. Column (3) provides the regression result of green credit and the firms’ digitization level on Ln(Patent + 1). The regression coefficients of DID and Ln(Innovate + 1) are significantly positive at least at the 5% level, signaling that green financial credit allocation can enhance the digitization level of firms, consequently promoting the enhancement of the GIQ in green-credit-restricted industries, thus proving hypothesis H2a.

5.2. Mechanism Test Based on Total Factor Productivity of Enterprises

The OP and LP methods are frequently utilized in existing studies to gauge a firm’s total factor productivity [47]. However, the LP method is more adaptable to handling sample loss and endogeneity issues than the OP method, and it provides more accurate estimates of a firm’s total factor productivity. Thus, following the approach of Min et al. (2021) [48], this paper employs the LP method to measure the firm’s total factor productivity TFP_LP. This paper utilizes the Bootstrap method with 500 samples, resulting in an estimated indirect effect ab of Ln(Patent + 1) at 0.016, with a 95% confidence interval of (0.011, 0.022). As the confidence interval does not include 0, the two-tailed test is significant (p = 0.001), suggesting the existence of a mediating effect with TFP_LP as the transmission variable. In

Table 9. Mechanism test based on firms’ total factor productivity.

Model	(1)	(2)	(3)
	Ln(Patent + 1)	TFP_LP	Ln(Patent + 1)
DID	0.058 ***	0.458 ***	0.055 ***
	(4.03)	(21.54)	(3.73)
Policy	0.003	−0.298 ***	0.028
	(0.06)	(−3.98)	(0.55)
Gcres	0.238 **	0.313 ***	0.408 ***
	(2.08)	(3.48)	(6.65)
TFP_LP	—		0.025 ***
	—		(3.67)
Control variable	Yes	Yes	Yes
Year	Yes	Yes	Yes
Sector	Yes	Yes	Yes
Area	Yes	Yes	Yes
N	10,572	10,186	10,186

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

, column (2) demonstrates the impact of green finance credit allocation on firms’ total factor productivity, with the regression coefficient significantly positive at the 1% level. Meanwhile, column (3) represents the combined effect of green financial credit allocation and firms’ total factor productivity on the GIQ, showing both DID and TFP_LP coefficients to be significantly positive at the 1% level. These results indicate that the GCP can enhance the total factor productivity of the green-credit-restricted industry, subsequently improving the GIQ of enterprises, thereby supporting hypothesis H2b.

5.3. Mechanism Test Based on Shadow Banking

Referring to Zhen et al. (2023) [31], the paper delves into the link between “green credit and corporate green innovation” by using the sum of corporate accounts receivable, notes receivable and prepayment divided by business revenue to denote commercial credit (TC). A higher TC value signifies a greater supply of corporate shadow banking. Employing the Bootstrap method for 500 samplings, the estimated Ln(Patent + 1) succinct effect ab is found to be 0.005, with a 95% confidence interval of (0.003 0.007). As the confidence interval does not encompass 0 and the two-tailed test is significant (p = 0.002), this indicates a mediation effect with TC as the transmission variable. The regression results in

Table 10. Mechanism test based on business credit of enterprises.

Model	(1)	(2)	(3)
	Ln(Patent + 1)	TC	Ln(Patent + 1)
DID	0.058 ***	−0.083 ***	0.056 ***
	(4.03)	(−6.33)	(3.83)
Policy	0.003	−0.027	−0.0004
	(0.06)	(−0.70)	(−0.01)
Gcres	0.238 **	0.498 ***	0.470 ***
	(2.08)	(8.40)	(7.03)
TC	—		−0.047 ***
	—		(−4.03)
Control variable	Yes	Yes	Yes
Year	Yes	Yes	Yes
Sector	Yes	Yes	Yes
Area	Yes	Yes	Yes
N	10,572	9394	9394

Notes: *** and ** indicate significance at the 1% and 5% levels, respectively.

reveal that the regression coefficient of DID in column (2) is significantly negative at the 1% level, suggesting that the implementation of GCP can curtail shadow banking in listed companies. Meanwhile, the regression coefficient of DID in column (3) is significantly positive at the 1% level, and the regression coefficient of TC is significantly negative at the 1% level. These findings imply that the GCP can improve GIQ in restricted industries by restricting shadow banking activities in listed companies, thus supporting hypothesis H2c.

6. Further Discussion

6.1. The Impact of Financial Regulation

Drawing from the theoretical analysis of how the allocation of green financial credit may complicate financial regulation, this paper delves deeper into the influence of financial regulation on the green innovation of enterprises enabled by green credit. Previous research has relied on the ratio of regional financial regulatory expenditure to financial sector value added as a proxy for local financial regulation. However, this indicator is limited to the provincial level, leading to internal bias in evaluating the level of financial regulation at the city level due to data constraints. To address this, this paper manually compiles data on the number of urban bank and financial institution outlets nationwide to estimate the level of financial regulation at the city level. It constructs financial weights for each city by using the ratio of the number of urban bank outlets to the number of bank outlets in the province and multiplies them by the financial regulation index at the provincial level to obtain the financial regulation indicator. According to the regression results in

Table 11. Impact based on financial regulation, fintech and intellectual property protection.

Model	(1)	(2)	(3)
	Ln(Patent + 1)	Ln(Patent + 1)	Ln(Patent + 1)
DID × Regulation	0.047 ***	—	—
	(3.90)	—	—
DID × Ln(fintech + 1)	—	0.091 ***	—
	—	(3.28)	—
DID × Lnprotect	—	—	0.043 ***
	—	—	(4.72)
Policy	−0.070 ***	0.096 *	−0.126 ***
	(−3.44)	(1.78)	(−5.30)
Gcres	0.458 ***	0.240 **	0.463 ***
	(7.46)	(2.12)	(7.55)
Regulation	−0.011 *	—	—
	(−1.70)	—	—
Ln(fintech + 1)	—	−0.134 ***	—
	—	(−6.05)	—
Lnprotect	—	—	0.085 ***
	—	—	(3.32)
Control variable	Yes	Yes	Yes
Year	Yes	Yes	Yes
Sector	Yes	Yes	Yes
Area	Yes	Yes	Yes
N	10,572	10,572	10,572

Notes: ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

, the coefficients of “DID × Regulation” are both significantly positive at the 1% level, with regulation further promoting green innovation activities in the green-credit-restricted industry and enhancing the GIQ. This provides evidence to support hypothesis H3a.

6.2. The Impact of Financial Technology

Given that fintech has the potential to enhance the efficient allocation of green credit to financial resources, this paper leverages the approach taken by Fei et al. (2021) [42]. Initially, this paper utilizes Python tools to scrape 48 keywords associated with “fintech” from relevant news articles and conferences. Subsequently, this paper employs a Baidu News Advanced Search to look for news pages containing these keywords in conjunction with cities and municipalities. By crawling the source code of the Baidu News Advanced Search and extracting the number of search results, this paper aggregates a total of 254,456 searches containing “region + keywords” at the prefecture-level city or municipality directly under the central government. Next, this paper manually sorts the number of branches of each bank in each year and city using financial license information from the China Banking Regulatory Commission (CBRC) to construct the Herfindahl Index (HHI) for measuring bank competition. Additionally, this paper aggregates the number of AI enterprises at the city and provincial levels and applies the location entropy index to gauge the AI enterprise concentration at the regional level. Multiplying these three indexes yielded the comprehensive fintech evaluation index. To address right skewness and prevent missing values, this paper processes the index by adding one and taking the logarithm. Consequently, this paper obtains the fintech level measure at the city level (Ln(fintech + 1)). According to the regression results in Table 11, the coefficient of “DID × Ln(fintech + 1)” significantly indicates at the 1% level, supporting the notion that fintech further advances the allocation of green financial credit, empowering businesses to enhance the GIQ and rendering hypothesis H3b as valid.

6.3. The Impact of Intellectual Property Protection

Research has indicated that intellectual property protection systems can lead to an increase in the quantity and quality of patent applications filed by firms, consequently enhancing the level of innovation within these firms [49]. Thus, leveraging Shengchao’s work (2023) [43], this paper examines the regional level of intellectual property protection (Lnprotect), characterizing it as the average of the number of intellectual property infringement cases divided by the total population and the number of lawyers divided by the total population in each region and then taking the logarithm. According to the regression results in Table 11, the coefficient of “DID × Lnprotect” is significantly positive at the 1% level, suggesting that after implementation of the GCP, reinforcing regional intellectual property protection can effectively elevate the GIQ in green-credit-restricted industries. This supports the validation of hypothesis H3c.

7. Conclusions, Policy Recommendations and Limitations

7.1. Conclusions

As an important force of market-guided resource allocation and key link between the financial sector and the ecological environment, how to effectively serve corporate green innovation is the key to promoting the financial empowerment of entities. Under the policy background that China vigorously develops new quality productivity and promotes the transformation of green innovation, this paper takes the implementation of the Green Credit Guidelines as a quasi-natural experiment and utilizes the Knowledge Breadth Method to measure the GIQ of enterprises, exploring the impact of the GCP on enterprises’ GIQ during the period of 2007–2021. The following conclusions are drawn.

Firstly, GCP significantly improves the GIQ in green-credit-constrained industries, and this conclusion still holds after multidimensional robustness tests, such as parallel trend tests, placebo tests and changing explanatory variable measures. Secondly, the mechanism test suggests that GCP can enhance firms’ GIQ by increasing firms’ digitization level and total factor productivity and suppressing firms’ shadow banking. Thirdly, the moderating mechanism test surfaces that fintech, financial regulation and local intellectual property protection can strengthen the promotion effect of GCP on firms’ GIQ.

7.2. Policy Recommendations

Based on the findings, this paper presents the following policy implications. Firstly, the green financial system, of which green credit is an important component, should be continuously improved to accelerate the formation of green productivity. Research has shown that the GCP can force enterprises to improve the GIQ and promote the development of new productivity. Green credit standards, assessment and monitoring systems should be further improved to create stronger positive incentives for green innovation and improve the risk-prevention ability of banks. At the same time, it is necessary to continuously improve the green financial service system, covering diversified green financial support tools, including green credit, green bonds, green insurance and green funds, and to promote the formation of a market-oriented green technological innovation system.

Secondly, the synergistic empowerment of the GCP and related financial instruments should be enhanced to jointly promote the improvement of the GIQ of enterprises. Promoting green innovation as the core driving force of the green transformation is a systematic project, though the research in this paper also shows that the credit policy aimed at promoting the green development of enterprises is also forcing enterprises to continue to promote the digital transformation and total factor productivity. Therefore, improving the GIQ should be incorporated into a unified framework of intelligence, greening and integration, and the synergy and consistency between green finance and other financial policy tools should be enhanced to accelerate the formation of new and green productivity.

Thirdly, it should strengthen the positive incentive mechanism of financial technology and financial regulation and improve the local intellectual property protection system. Driven by the new round of scientific and technological revolution, the technical advantages of financial science and technology should be fully utilized to enhance the accuracy and effectiveness of green credit. At the same time, financial institutions should be encouraged to actively carry out green credit business through the formulation of more scientific and reasonable regulatory policies, so as to promote the accurate flow of financial resources to real enterprises and promote the high-quality development of green innovation. At the same time, it should also increase the intensity of local intellectual property protection to stimulate enterprises to take the initiative to carry out green innovation and provide policy support for the green and high-quality development of enterprises.

7.3. Limitations

It is essential to acknowledge the limitations of the study, although this paper has considered as many factors as possible when studying the impact of the GCP on the GIQ of corporates, and the limitations of this paper may lie in the following: on the one hand, this paper is based on the patent classification number and utilizes the Knowledge Breadth Method to measure the GIQ of corporates, which, although more precise than using only corporate green patents to measure the GIQ of firms, future research may be able to examine the sustainable dynamic perspective of enterprise green innovation. On the other hand, this paper only explores three paths of GCP affecting corporate green innovation, and future research may be able to examine this from more possible perspectives.