Does Green Credit Policy Promote or Inhibit Firms’ Green Innovation in China? Moderating Effect of Environmental Information Disclosure

Abstract

Green credit policy (GCP) serves as an important tool for environmental protection and economy development. However, conflicting evidence exists regarding its role in affecting firms’ green innovation. China’s GCP practice provides an opportunity to explore this issue in the context of developing economies. Taking the implementation of the “Green Credit Guidelines” in China in 2012 as an exogenous shock, this paper adopts the difference-in-differences (DIDs) method to separately explore GCP’s effect on green innovation of non-heavily polluting firms (non-HPFs) and heavily polluting firms (HPFs). Based on the microdata of Chinese firms from 2008 to 2020, this study finds that: (1) GCP promotes green innovation of non-HPFs, but inhibits green innovation of HPFs. (2) GCP’s promoting effect on green innovation of non-HPFs is more prominent in large-sized firms, regions with a higher financial development level, and regions with a higher pollution level. (3) GCP’s inhibiting effect on green innovation of HPFs is less prominent in regions with higher financial development level. (4) Environmental information disclosure (EID) strengthens GCP’s promoting effect on green innovation of non-HPFs. Overall, these findings help practitioners to better understand the impact of GCP on firms’ green innovation in developing countries.

1. Introduction

With rapid economic development giving rise to frequent environmental emergencies and severe pollution problems, the Chinese government has issued a series of policies to regulate firms’ behaviors [1]. The green credit policy (GCP) of China, which was initially designed in 2007 and formally enacted in 2012 by the Chinese central government, is an important policy instrument to alleviate firms’ polluting behaviors [2,3,4]. According to the Equator Principles proposed by the International Finance Corporation (IFC) in 2002, GCP refers to specific guidelines and rules designed for financial institutions to assess the environmental risks of projects or firms when lending [5]. Specifically, through providing more loans or preferential interest rates, GCP aims to put more funds from credit facilities (e.g., commercial banks) to environmentally friendly projects and their operating firms, eventually reducing industrial pollution [6].

Previous studies have found that, at the macro level, GCP encourages clean production, transforms the economic development paradigm, and creates a virtuous loop between finance and the environment [5,7]. At the micro level, scholars explore GCP’s influences on organizational activities, such as banks’ lending activities, firms’ investment, firms’ financing activities, firms’ polluting behaviors, and firms’ technological innovation [4]. For instance, Sun et al. explored the effect of GCP on firms’ pollution prevention behaviors [8]. Yao et al. studied the effect of GCP on firms’ financial behaviors and performance [9]. Xie et al. [10] analyzed both the direct and indirect effects of GCP on firms’ innovation. This study builds on the latter approach exploring the impact of GCP on firms’ green innovation activities.

Green innovation, belonging to both green activities and innovation activities, refers to “hardware or software innovation … in technologies … involved in energy-saving, pollution-prevention, waste recycling, green product designs, or corporate environmental management” [11] (p. 332). Green innovation enables firms to match technologies with green market opportunities and environmental protection requirements to advance their competitive advantages [12]. Given the importance of environmental protection and technological innovation for a firm’s survival, increasing studies explore GCP’s effects on green activities, innovation activities, or both at the firm level [13]. In general, there are two streams of literature conflicting with each other. One stream of literature, based on the neoclassical economic perspective, mainly focuses on challenges brought by GCP including a higher threshold of banks’ financing access and higher environmental compliance costs of firms [14]. Thus, this stream of literature emphasizes GCP’s inhibition effect on firms’ business activities such as green innovation. Alternatively, another stream of literature, based on Porter’s hypothesis, highlights that GCP and other environmental regulations may show an “innovation compensation effect”. That is, if environmental laws and regulations are well designed and implemented, they can significantly promote corporate innovation, which further offsets the rising costs of environmental protection [14,15]. Although both streams of research are insightful individually, they create a fundamental tension when taken together exemplified by the research question:

RQ1. Does GCP promote or inhibit firms’ green innovation?

Meanwhile, there has been increasing interest in the intersection between economic environmental policy instruments (e.g., GCP) and information-based environmental policy instruments (e.g., environmental information disclosure) [16]. The approach that focuses on the GCP-innovation linkage at the firm level can be extended by exploring the effect of environmental information disclosure (EID) [17]. In particular, EID may impact the potential benefits or costs of GCP, should it interact with GCP to affect firms’ green innovation activities. EID refers to a series of firm behaviors that disclose content such as environmental performance, environmental protection expenditure, and environmental protection income to the public through corporate annual reports, social responsibility reports, corporate websites, etc. [18,19,20]. As Luo et al. argue, firms with a high level of EID can help mitigate manager–owner conflicts and alleviate information asymmetry with stakeholders, hence turning better credit advantages into green innovation [20]. Thus, it is important to study how EID moderates the relationship between GCP and firms’ green innovation. Surprisingly, the existing literature provides little knowledge about this important issue.

Therefore, the second research question of this paper is:

RQ2. Can environmental information disclosure (EID) help enhance the positive or reduce the negative effect of GCP on firms’ green innovation?

By answering the proposed research questions, this study brings three main contributions to the relevant field of literature. First, we analyze GCP’s opposite effects on green innovation of non-heavily polluting firms (non-HPFs) and heavily polluting firms (HPFs). Thus, this study partially addresses the critical question of whether GCP promotes or inhibits firms’ green innovation [7,8,9,10]. Second, few studies examine EID’s moderating effect on the GCP–green innovation linkage, so little is known about how variations in EID affect GCP’s influence on firms’ green innovation [17]. We examine the moderating role of EID and the findings provide a deeper understanding of GCP. Third, we investigate the differences in GCP’s impact on green innovation of firms from the perspectives of firm heterogeneity and regional heterogeneity separately, which makes the research more comprehensive and is of great significance in helping governments to formulate detailed environmental regulation policies according to specific contexts.

Overall, the purpose of this study is to establish a framework to study whether GCP facilitates firms’ green innovation in developing economies such as China. The remaining part of this paper is organized as follows. Section 2 provides a theoretical background and literature review. Section 3 develops several hypotheses. Section 4 describes the research design. Section 5 presents the empirical results. Section 6 sets forth the conclusion, implications, and limitations.

2. Theoretical Background

2.1. Environmental Regulation and Firms’ Green Innovation

Firms’ green innovation refers to new or modified techniques, processes, and products which help firms better reduce pollution, control emissions, save energy, and ultimately achieve sustainable development [21]. Unlike highly symbolic and visible activities such as green ads, green innovation is characterized by high investments, high risks, and long cycles [22]. Meanwhile, compared with general innovation activities, green innovation involves both innovation issues and social issues, which lead to increased managerial complexity. In addition, green innovation has the characteristic of the “spillover effect”, which means green innovation achievements may be easily imitated by other entities and spread continuously [23]. Therefore, many firms lack the internal motivation to implement green innovation strategies and their green innovation activities are strongly driven by external institutional forces [21,22].

Institutional theory posits that firms’ behaviors are affected by political, economic, and social forces exerted by external institutes (e.g., governments, financial banks, suppliers, customers, etc.) [24]. Prior green innovation studies have identified a number of institutional forces, especially environmental regulation [21]. As an important policy instrument for governments’ social governance, environmental regulation restricts firms’ production and operation behaviors by imposing administrative penalties, emission standards, and emission taxes to reduce or even eliminate industrial pollution and facilitate sustainable economic development [25,26].

There are two main views (i.e., neoclassical economic perspective and Porter’s hypothesis) on whether environmental regulation can promote firms’ green innovation. According to the neoclassical economic perspective, environmental regulation significantly increases related firms’ environmental compliance costs, thereby preventing them from carrying out green innovation [6]. Chen et al. found that China’s carbon emission trading schemes have an evident lagging effect on restraining firms’ green innovation [12]. Tao et al. took China’s Target Responsibility System of Environmental Protection as a quasi-natural experiment and discovered that the policy leads to a decline in the quality of green innovation activities [27]. On the contrary, based on Porter’s hypothesis, other studies argue that more stringent but properly designed environmental regulation could trigger technological innovation and enhance firm competitiveness [15]. As Peng et al. showed, environmental regulation promotes green innovation behaviors through increasing green innovation intention [21]. Cui et al. found that the Cleaner Production Audit has a positive effect on both radical and incremental green innovation [1]. Lu et al. further argued that green finance reform and innovation can significantly enhance firms’ green technology innovation [28]. Therefore, it remains unclear whether environmental regulation facilitates or inhibits firms’ green innovation.

2.2. Green Credit Policy and Firms’ Activities

According to the Equator Principles proposed by the International Finance Corporation (IFC) in 2002, green credit policy (GCP) refers to specific guidelines and rules designed for financial institutions to assess the environmental risks of projects or firms when lending [5]. Customers of these financial institutions who do not comply with environment protection requirements will not be granted credit. GCP is thus a market-driven environmental policy instrument that promotes firms’ energy conservation, emission reduction, and other green activities through differential and dynamic credit policies of banking financial institutions [5,6,9,14,29]. Compared with many developed economies, China issued the “Green Credit Guidelines” recently in 2012.

Given the importance of GCP in China, increasing studies have explored GCP’s impact on firm activities, such as links to polluting behaviors, investment, financing activities, and innovation [4,8,9,10]. A review of the literature shows that the research themes into GCP’s impact at the firm level can be grouped into three types: (1) GCP’s impact on the economic activities of firms; (2) GCP’s impact on the environmental activities of firms; and (3) GCP’s impact on the innovation activities of firms. By reviewing related studies, we can find the following three gaps. First, the literature is unequivocal regarding the role of GCP in affecting firms’ activities in the context of developing countries. As Liu et al. pointed out, based on Porter’s hypothesis, prior studies have verified the positive effect of environmental regulation (e.g., GCP) on innovation, but whether this positive effect is valid in developing economies remains to be examined [7]. While some studies, in the context of developing economies, find that GCP has a positive effect on firm activities, others find insignificant or negative impacts [6,7,10,13,14,29,30,31,32,33,34,35]. Second, the previous literature has often explored GCP’s impact on green activities or innovation activities separately, however, rarely discusses GCP’s influence on green innovation. Third, the existing research on the effect of GCP on firm activities neglect environmental information disclosure [17].

2.3. The Moderating Role of Environmental Information Disclosure

Scholars increasingly recognize that, despite facing broadly similar environmental regulation forces (e.g., GCP), the variance in firms’ green innovation outputs persist. In other words, the relationship between environmental regulation and green innovation may be contingent on some contextual factors. While most related studies focus on external contextual factors [1,22,29,36], they neglect internal ones such as environmental information disclosure (EID). In accord with mounting interest in the interface between GCP and EID [17], we suggest that the analysis of the GCP–green innovation linkage can be extended by exploring EID’s moderating effect.

EID refers to a series of behaviors where market subjects (e.g., firms) disclose content such as environmental protection expenditure, environmental protection income, and environmental performance to the public through corporate annual reports, social responsibility reports, corporate websites, etc. [20]. According to agency theory, EID can effectively deal with the problem of information asymmetry to reduce the agency cost between various stakeholders and the focal firm [20]. Scholars who treat EID as a key firm characteristic have found that businesses with a high level of EID always perform better in green innovation [37]. As Luo et al. (2022) showed, EID significantly promotes firm innovation, especially exploratory innovation [20]. Zhang et al. found that the improvement of environmental information transparency has significantly increased the green innovation of firms, especially the number of green invention patent applications [38]. Ding et al. (2022) point out that EID increases the green patents of high-polluting firms [16].

In summary, previous studies have explored EID’s consequences at the firm level. However, there are few investigations on the moderating mechanism of EDI on the GCP–green innovation linkage. This research gap provides a good entry point for our study. The conceptual model is shown in Figure 1.

3. Hypotheses Development

3.1. Green Credit Policy and Green Innovation

We posit that green credit policy (GCP) has a positive effect on firms’ green innovation in non-heavily polluting industries. First, with the promulgation of GCP, non-HPFs have more funds to support green innovation. GCP compels credit-granting financial institutions to restrain HPFs’ loan applications while supply non-HPFs with a larger scale of debt financing [6,7,32]. With more funds, non-HPFs are more willing to innovate, take risks, and be more proactive than competitors toward emerging green opportunities. Second, GCP urges non-HPFs to invest more in green innovation projects through bringing a strong sense of crisis. Although GCP creates a relatively loose credit environment for non-HPFs in a short term, it poses more challenges for non-HPFs in the long term. Managers of non-HPFs easily realize that green credit standards are getting higher in the future and they will face a harsher credit environment [21,35,39]. The expectation of more intense credit competition makes non-HPFs’ managers attach greater importance to green value-added initiatives such as green technological innovation. Third, the implementation of GCP sends signals to the whole capital market that non-HPFs will be granted more credits [6]. As signaling theory posit, information is an important factor affecting the capital market. In this way, private investors may follow banks to invest in non-HPFs. Based on these arguments, this paper proposes the first hypothesis.

Hypothesis 1.

Green credit policy promotes green innovation in non-heavily polluting firms.

We predict that GCP has a negative effect on firms’ green innovation in heavily polluting industries. First, GCP may limit HPFs’ loan financing capacities and subsequently discourage their efforts in sustainable business activities such as green innovation [9]. Specifically, GCP stipulates that banks should increase loans lent to green projects but restrict loans lent to high pollution, high emission, and high energy consumption projects, thereby exerting high financing constraints on HPFs [9]. In this way, if HPFs want to secure financial support from banks under the supervision of GCP, they have to divert their limited green resources from those long-term oriented behaviors (e.g., green technological innovation) into short-term oriented ones (e.g., emission reduction, green ads, etc.) [5,35]. Second, the existence of GCP requires firms, especially HPFs, to invest more on daily environmental protection, thereby limiting their innovation capacities [21]. In other words, strict GCP significantly increases HPFs’ environmental protection costs which may further crowd out green innovation budget [21]. Third, after financial institutions comprehensively assess HPFs’ environmental risks according to GCP, they may send warning signals to external market subjects, especially related investment institutions. Under these circumstances, those institutions may reduce or even withdraw their investments. Without sufficient capital inflows, HPFs tend to behave cautiously and avoid taking high-risk actions such as green innovation [6,31]. Fourth, according to Porter’s hypothesis, HPFs have the motivation to invest more in green technological innovation when facing stricter environmental regulations [15]. Although GCP has both positive and negative impacts on HPFs’ green innovation, we believe the net effect will be negative. Therefore, this paper proposes the second hypothesis.

Hypothesis 2.

Green credit policy inhibits green innovation in heavily polluting firms.

3.2. The Moderating Effect of Environmental Information Disclosure

We predict that EID can enhance the positive effect of GCP on green innovation in non-HPFs. First, EID can reduce agency conflicts between firm owners and managers, and then help non-HPFs better exploit green credit opportunities in green innovation projects. As Luo et al. argued, EID reduces owner–manager agency conflicts and increases managers’ intentions in investing high-risk activities, especially green innovation projects [20]. By contrast, in firms with serious owner–manager conflicts, managers may seek short-term interests and object to many uncertain projects such as green innovation [40]. Second, by alleviating information asymmetry between firms and their stakeholders (e.g., customers, governments, investors, etc.), EID could help non-HPFs augment GCP’s information value for green innovation. Although GCP can partly convey non-HPFs’ environmental information to their external stakeholders, it is insufficient for investors to make sound investment decisions. As EID enables private investors to learn more detailed information about the focal firm, it markedly improves non-HPFs’ information transparency as well as legitimate status [41]. In this situation, non-HPFs are more likely to get funds from external investors (e.g., banks, private investors, etc.), contributing to higher levels of green innovation. Third, as a specific type of voluntary environmental regulation, EID can reflect one firm’s tendency to green initiatives. In this way, firms with a high level of EID are more motivated to capture the first-mover advantage in the future market [35]. Based on these arguments, we hypothesize:

Hypothesis 3.

Environmental information disclosure strengthens the promoting effect of GCP on green innovation in non-heavily polluting firms.

We suggest that EID weakens the inhibiting effect of GCP on green innovation in HPFs. First, EID could help HPFs decrease the difficulty of obtaining debt capital. According to institutional theory, legitimacy is critical to a firm’s survival because it ensures support and also continuous capital inflow from external stakeholders such as creditors [42]. Firms obtain legitimacy by signaling their conformity to various stakeholders’ expectations [42]. After the implementation of GCP, environmentally inferior firms, especially HPFs, are exposed to greater institutional pressures, including compliance pressure, public opinion pressure, and moral condemnation [43,44]. In this situation, by disclosing key green information to investors and meeting their environmental protection demands, EID helps to prove HPFs’ legitimacy, which decreases the level of credit financing difficulty and as such enhances green innovation [43]. Second, by providing incentives to shift toward green processes, EID augments the motivation value of GCP for HPFs’ green innovation. According to Porter’s hypothesis, HPFs are actually motivated to conduct green innovation after the promulgation of GCP [15]. However, the motivation level is low considering strict financing regulations that HPFs face. Under these circumstances, HPFs’ EID behaviors stimulate the public’s concerns about their green behaviors [38]. To better cater to these concerns, HPFs are inclined to adopt more green initiatives such as green innovation. Therefore, we hypothesize:

Hypothesis 4.

Environmental information disclosure weakens the inhibiting effect of GCP on green innovation in heavily polluting firms.

4. Data and Methodology

4.1. Data

This paper uses the implementation of the “Green Credit Guidelines” in China as a quasi-natural experiment. Since the guidelines were promulgated in February 2012, this paper selects all Chinese A-share listed firms from 2008 to 2020 as the original sample, excluding ST firms, *ST firms, PT firms, financial firms, firms with missing data, and firms whose asset–liability ratio is greater than 1, and finally obtains 3698 listed firms and 26,766 observations. In particular, we use the interpolation method to fill in missing values for data, and winsorize continuous variables at 1% and 99% to exclude the outlier effect. This paper focuses on the empirical analyses of both non-HPFs and HPFs. According to the industry category of listed firms and their pollution intensity, firms in the sample are classified into non-heavily polluting firms (non-HPFs) and heavily polluting firms (HPFs). Non-HPFs affected by the green credit policy (i.e., GCP) are regarded as treatment group 1, all HPFs are regarded as treatment group 2, and non-HPFs not affected by GCP are regarded as the control group.

According to the industry classification guidelines in China, the industry codes referring to firms in heavily polluting industries are B01, B03, B05, B07, C01, C03, C05, C11, C14, C31, C35, C41, C43, C61, C65, C67, C81, D01, H01, and H03 [14]. In this way, HPFs in our study are from industries including steel, electrolytic aluminum, thermal power, cement, coal, metallurgy, petrochemicals, chemicals, paper making, building materials, pharmaceuticals, fermentation, brewing, leather, textiles, and mining [45]. We have 3698 firms in the research sample. Finally, treatment group 1 includes 14,605 non-HPF observations which are not affected by GCP; treatment group 2 includes 7814 HPF observations, and the control group includes 4347 non-HPF observations which are affected by GCP. The data of listed firms come from the China Stock Market and Accounting Research Database (CSMAR) and Chinese Research Data Services (CNRDS).

4.2. Variables

4.2.1. Dependent Variable

Green innovation (GI) is the dependent variable in this study. Existing literature uses either R&D investment or the number of patents to measure firms’ innovation capacity. While green R&D investment emphasizes the pre-innovation stage and cannot accurately measure firms’ innovation output [7], green patents are introduced to reflect firms’ green innovation capacity and also green innovation performance. Furthermore, green patents can be subdivided into green invention patents and green utility patents [32]. Compared to green utility patents, green invention patents are a more convincing measure of innovation capacity because they represent substantial green technological innovation [32]. Therefore, we choose green invention patent grants as the indicator of firms’ green innovation. After obtaining the total number of green invention patents applied by firms in each year, we take the natural logarithm of the number of green invention patents plus one [22]. The higher the GI value is, the higher the green innovation level the firm has. Specifically, we use NHPFGI to denote non-HPFs’ green innovation, and HPFGI to represent HPFs’ green innovation.

4.2.2. Independent Variable

Model 1 is a multi-period DID model because the impact of GCP on non-HPFs is not at a single point in time, but begins when non-HPFs establish lending relationships with banks which implement GCP. Time_i is the main test independent variable in Model 1. In this paper, each bank’s first disclosure of green credit balances in its annual report and related reports marks the basic completion of its green credit system structure. It is then considered as the time point when the bank formally begins to implement GCP. By comparing the year in which the firm establishes a lending relationship with one bank with the time when that bank implements its GCP, if the former is later than the latter, the year in which the lending relationship is established is taken as the policy impact point, and vice versa, the time when the bank implements its green credit policy is taken as the policy impact point. After determining the policy impact point for each firm, Post_t is a year dummy variable that equals 1 if the observation year is after that point and 0 if the observation year is before that point.

Treat_i*Post_t is the main test independent variable in Model 2, which is a general DID model. Since the Green Credit Guidelines were issued on 24 February 2012, Post_t is a year dummy variable that equals 1 if the observation year is from 2012 to 2020 and 0 if the observation year is from 2008 to 2011. Moreover, if a listed firm belongs to the above 16 heavily polluting industries, we allocate it to the treatment group and Treat_i equals 1, and if a listed firm does not belong to the above 16 heavily polluting industries, we class it into the control group and Treat_i equals 0. The coefficient of Treat_i*Post_t indicates that, after the implementation of GCP, compared to the control group, the green innovation in the treatment group has changed, which directly reflects the effect of GCP.

4.2.3. Control Variables

In terms of the selection of control variables, this paper selects firm growth (Growth), returns on asset (ROA), firm age (Age), asset–liability ratio (Lev), fixed assets (FixAss), cashflow (CaF), and institutional investors (Inst) as control variables. Δ_i denotes the individual fixed effect and μ_t denotes the time fixed effect (See

Table 1. Definitions and measures of major variables.

Variable Name	Variable Symbol	Definition
Green innovation of non-HPFs	NHPFGI	Green invention patent grants of non-heavily polluting firms
Green innovation of HPFs	HPFGI	Green invention patent grants of heavily polluting firms
Treatment group 1	Treat	The value of non-HPFs affected by GCP is 1, and the value of non-HPFs not affected by GCP is 0
Treatment group 2	Treat	The value of HPFs is 1, and the value of non-HPFs not affected by GCP is 0
Treatment period	Post	For HPFs, the value of the sample is 1 in and after 2012, otherwise it is 0; for non-HPFs, the value of the sample is 1 in and after the policy impact point, otherwise it is 0.
Environmental information disclosure	EID	With content analysis method, environmental information is first classified, then each category is assigned a value according to its content, and finally, the score for each firm is calculated
Firm growth	Growth	The growth rate of business revenue
Returns on asset	ROA	Return on assets equal to earnings before interest, tax and abnormal items at fiscal year-end divided by average total assets
Firm age	Age	The number of years from the issue period to the current period, taking the logarithm, ln(current year − year of incorporation + 1)
Asset–liability ratio	Lev	Total debt divided by total assets at fiscal year-end
Fixed assets	FixAss	Fixed assets divided by total assets at fiscal year-end
Cashflow	CaF	Net cash flow from operating activities divided by total assets
Institutional investors	Inst	The percentage of shares owned by institutional investors in the enterprise
Individual fixed effect	δi	Control of individual firm
Time fixed effect	μt	Control of year

for definitions of variables).

4.2.4. Moderation Variable

Environmental information disclosure (EID) is the moderation variable. In line with many existing studies, we use content analysis to measure EID. With this method, we classify environmental information into seven sections, namely environmental management, environmental liabilities, environmental performance, environmental governance, information disclosure vehicles, and certification by independent agencies. Under each section, a number of sub-items are set up and scored according to the content of EID; the rules are 0 for no description, 1 for general qualitative description and 2 for quantitative description; a few sub-items are 0 for no disclosure and 1 for disclosure; and finally, the scores of each indicator are summed to obtain each sample firm’s EID score. To avoid the problem of right-skewed distribution, the score is taken as the logarithm of EID score.

4.3. Methodology

The difference-in-difference (DID) model is widely used in policy evaluation because it can suggest causality [45]. In this study, GCP meets the preconditions of the DID model as a purely exogenous event, because it does not occur for experimental purposes and will not be affected by a single firm. It can be regarded as a quasi-natural experiment. This paper constructs a quasi-natural experiment using the implementation of the “Green Credit Guidelines” in China in 2012 and uses the DID model to examine GCP’s effect on firms’ green innovation. In order to distinguish GCP’s different effects on different pollution types of firms, we divide firms into non-HPFs and HPFs according to their pollution levels.

To test H1, a multi-period DID model is set up in this paper, as GCP’s impacts on non-HPFs are not at the same point in time, but begin when establishing credit relationships with banks which implement GCP.

$$NHPFG{I}_{i,t}={\alpha }_0+{\beta }_{0}Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(1)

$NHPFG{I}_{i,t}$ represents green innovation of the non-heavily polluting firm i in the year t. $Pos{t}_t$ is an event dummy variable, while the value is 1 in and after the policy impact point, otherwise it is 0. $X_{i,t}$ is a series of firm-level control variables. ${\delta }_i$ is an individual fixed effect. ${\mu }_t$ is a time-fixed effect. ${\epsilon }_{i,t}$ is a random perturbation term. We are interested in β₀, and if β₀ is significantly positive, it indicates that the introduction of GCP has a positive effect on green innovation of non-HPFs, and vice versa.

To test H2, the paper sets the DID model as

$$HPFG{I}_{i,t}={\alpha }_0+{\beta }_{0}Trea{t}_i\times Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(2)

${\mathrm{HPFGI}}_{\mathrm{i},\mathrm{t}}$ represents green innovation of the heavily polluting firm i in the year t. In Model 2, $Trea{t}_i$ is a group dummy variable, while the value of HPFs is 1 and the value of non-HPFs not affected by GCP is 0. $Trea{t}_i\times Pos{t}_t$ is the interaction item in the DID model. The other variables are defined as in Model 1. We are interested in β₀, and if β₀ is significantly negative, it indicates that the introduction of GCP has a negative effect on green innovation of HPFs, and vice versa.

To test H3 and H4, this paper sets the DID model as

$$\mathrm{NHPF}G{I}_{i,t}={\alpha }_0+{\beta }_{0}Pos{t}_t\times EI{D}_i+{\beta }_{1}EI{D}_i+{\beta }_{2}Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(3)

$$\begin{gathered} HPFG{I}_{i,t}={\alpha }_0+{\beta }_{0}Trea{t}_i\times Pos{t}_t\times EI{D}_i+{\beta }_1\times Trea{t}_i+{\beta }_{2}EI{D}_i\times Trea{t}_i+ \\ {\beta }_{3}EI{D}_i\times Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t} \end{gathered}$$

(4)

EID_i represents environmental information disclosure of the firm i. $Trea{t}_i\times Pos{t}_t\times EI{D}_i$ is the three-way interaction item. The other variables are defined as in Model 1. We are interested in β₀, and if β₀ is significantly positive, it indicates that environmental information disclosure is positively moderating GCP’s effect on green innovation, and vice versa.

5. Empirical Results

5.1. Descriptive Statistics

The descriptive statistics are divided into three categories, as shown in

Table 2. Descriptive Statistical Results.

	Control Group		Treatment Group 1		Treatment Group 2
Variable	Mean	SD	Mean	SD	Mean	SD
GI	0.600	1.020	0.570	0.950	0.420	0.780
EID	1.680	0.750	1.550	0.680	2.090	0.810
ROA	0.0600	0.0600	0.0400	0.0600	0.0500	0.0600
CaF	0.0500	0.0700	0.0400	0.0700	0.0600	0.0700
FixAss	0.180	0.150	0.170	0.140	0.300	0.170
Lev	0.380	0.210	0.440	0.210	0.420	0.210
INST	0.370	0.260	0.370	0.230	0.390	0.240
Age	2.820	0.350	2.820	0.370	2.830	0.340
Growth	0.470	1.360	0.550	1.400	0.210	0.830

. The control group denotes non-HPFs not affected by GCP. Treatment group 1 denotes non-HPFs firms affected by GCP. Treatment group 2 denotes HPFs.

5.2. Baseline Regression Results

Table 3. Baseline Regression Results.

	(1)	(2)
	NHPFGI	HPFGI
Treati*Postt		−0.119 ** (−2.33)
Postt	0.0331 ** (2.17)
Age	0.405 *** (3.00)	0.289 (1.70)
ROA	0.261 ** (2.10)	0.405 * (2.39)
Cashflow	−0.207 ** (−2.55)	−0.0195 (−0.20)
Lev	0.253 *** (3.48)	0.289 *** (3.43)
Inst	−0.0219 (−0.53)	0.0341 (0.65)
Growth	−0.00272 (−0.69)	−0.00454 (−0.56)
FixAss	0.0157 (0.17)	0.0633 (0.69)
Constant	−0.685 * (−1.80)	−0.433 (−0.91)
Year-fixed	YES	YES
Firm-fixed	YES	YES
N	18,168	11,576
Adjusted R2	0.668	0.645

Note: *, ** and *** indicate that the estimated coefficients are significant at the levels of 10%, 5%, and 1%, respectively. T-values in parenthesis are provided.

shows GCP’s effect on firms’ green innovation, where column (1) is for non-HPFs and column (2) is for HPFs. To address endogeneity, we use clustered standard errors at the firm level to ensure that the regression above is reliable. For non-HPFs, the coefficient in column (1) is 0.0331, which is significant at the 5% level. This is consistent with H1, indicating that GCP stimulates non-HPFs’ green innovation. For HPFs, the coefficient in column (2) is −0.119, significant at the 5% level. This result means that GCP inhibits green innovation of HPFs, thereby validating H2.

5.3. Moderating Effects of Environmental Information Disclosure

To test H3 and H4, we further analyze the moderation effects, and their results are listed in

Table 4. Moderating effects of EID.

	(1)	(2)
	NHPFGI	HPFGI
Treati*Postt*EID		0.019 (0.47)
Treati* EID		−0.108 *** (−4.07)
Postt*EID	0.055 *** (2.80)	0.100 *** (3.38)
Treati*Postt		−0.122 * (−1.66)
Age	0.397 *** (2.95)	0.318 * (1.89)
ROA	0.252 ** (2.04)	0.421 ** (2.50)
Cashflow	−0.214 *** (−2.65)	−0.012 (−0.12)
Lev	0.254 *** (3.51)	0.286 *** (3.41)
Inst	−0.022 (−0.54)	0.034 (0.65)
Growth	−0.003 (−0.69)	−0.005 (−0.65)
FixAss	0.009 (0.09)	0.050 (0.54)
Constant	−0.687 * (−1.82)	−0.558 (−1.20)
Year-fixed	YES	YES
Firm-fixed	YES	YES
N	18,168	11,576
Adjusted R2	0.670	0.650

Note: *, ** and *** indicate that the estimated coefficients are significant at the levels of 10%, 5%, and 1%, respectively. T-values in parenthesis are provided.

. Column (1) is for non-HPFs and column (2) is for HPFs. The main focuses are the triple intersections of Treat_i, Post_t, and the moderator, which imply the differences of Treat_i*Post_t when the moderator changes. As the table shows, in column (1), the coefficient of Treat_i*Post_t*EID is 0.066, which is significant at the 1% level. In column (2), the coefficient of Treat_i*Post_t*EID is 0.019, which is not significant. These results indicate that any increase in EID will facilitate the impact of Treat*Post on GI in non-HPFs, but not affect the impact of Treat*Post on GI in HPFs. Therefore, H3 is proved but H4 is not proved.

5.4. Heterogeneity Analysis

In order to examine the heterogeneity of GCP’s effects on green innovation, this paper further examines the heterogeneity in terms of firm size, the level of regional financial development, and the level of regional pollution. The specific estimation results are shown in

Table 5. Heterogeneity test.

	(1)	(2)	(3)	(4)	(5)	(6)
	NHPFGI	NHPFGI	NHPFGI	HPFGI	HPFGI	HPFGI
Postt*Size	0.137 *** (5.05)
Postt*Findev		0.040 ** (1.15)
Postt*Pollute			0.076 *** (2.96)
Treati*Postt*Size				0.106 (1.57)
Treati*Postt*Findev					0.134 * (1.75)
Treati*Postt*Pollute						0.072 (1.14)
Constant	−0.651 * (−1.72)	−0.678 * (−1.79)	−0.679 * (−1.79)	−0.525 (−1.12)	−0.450 (−0.95)	−0.411 (−0.87)
Control Variables	YES	YES	YES	YES	YES	YES
Year-fixed	YES	YES	YES	YES	YES	YES
Firm-fixed	YES	YES	YES	YES	YES	YES
N	18,168	18,168	18,168	11,576	11,576	11,576
Adjusted R2	0.670	0.668	0.668	0.647	0.645	0.645

Note: *, ** and *** indicate that the estimated coefficients are significant at the levels of 10%, 5%, and 1%, respectively. T-values in parenthesis are provided.

, where columns (1) to (3) are for non-HPFs and columns (4) to (6) are for HPFs.

(1) Heterogeneity of firm size

Schumpeter’s innovation theory suggests that firm size is closely related to innovation [12]. Firms of different sizes face varying levels of capabilities, costs, and risks, thereby creating distinct micro-environments for firm-level innovation activities. This paper takes the logarithm of the total assets of the firm to measure firm size. The Treat_i*Post_t*Size is obtained by multiplying firm size with the policy dummy variable.

The coefficient of Treat_i*Post_t*Size in column (1) is significantly positive at the 1% level, which indicates that the size can influence GCP’s promotion effect on green innovation of non-HPFs. That is to say, GCP’s promotion effect on green innovation is more obvious for large-sized non-HPFs, compared to small-sized non-HPFs. The coefficient of Treat_i*Post_t*Size in column (4) is insignificantly positive, indicating that firm size plays a limited role in the GCP-HPFs’ green innovation linkage.

(2) Heterogeneity of regional financial development

The regional financing environment faced by firms may also have a significant impact on the GCP–green innovation linkage. In regions with well-developed financial markets, firms have abundant access to external capital. In addition to credit financing, firms have other capital options such as bond financing and equity financing. In contrast, in regions with less developed financial markets, firms usually face limited capital options and rely heavily on bank credit as a major source of financing [32]. As GCP influences firms’ green innovation by affecting their credit financing, regional financial development can provide or restrict local firms’ financing channels. Therefore, GCP’s impacts on green innovation may vary in regions with different levels of financial development.

We measure regional financial development by the ratio of the sum of banking loans to GDP [46]. The regression results in columns (2) and (5) show that the coefficients of Treat_i*Post_t*Findev are both positive at the 10% level. This indicates that the level of regional financial development can positively moderate GCP’s effects on both non-HPFs’ and HPFs’ green innovation. To be specific, the higher the level of regional financial development, the more obvious GCP’s incentivizing effects on non-HPFs’ green innovation and the less significant GCP’s inhibitory effect on HPFs’ green innovation.

(3) Heterogeneity of regional pollution level

Pollution intensity varies greatly across regions in China. The more polluted a region (e.g., a city) is, the more its local government may emphasize high-pollution industries. This paper measures regional pollution by the haze pollution of the city where a firm’s headquarter locates. By referring to the study of Zhang et al. [47], the annual average concentration of fine particles with a diameter of 2.5 µm or less (PM_2.5) is employed as the proxy indicator of haze pollution. The coefficient of Treat_i*Post_t*Pollute in column (3) is significantly positive, which indicates that the regional pollution level can influence GCP’s promotion effect on green innovation of non-HPFs. That is, GCP’s promotion effect on green innovation is more obvious for non-HPFs located in high pollution regions. The coefficient of Treat_i*Post_t*Pollute in column (6) is insignificantly positive, indicating that regional pollution intensity plays a limited role in the GCP-HPFs’ green innovation linkage.

5.5. Robustness Testing

To guarantee the robustness of the results, both the parallel trend test and placebo test are carried out.

5.5.1. Parallel Trend Test

When using the DID method, the treatment and control groups need to pass the parallel trend test. That is, without the implementation of GCP, the development trend of treatment and control groups’ variables remain consistent. Hence, drawing on the study of Jcobsons et al. and adopting the event study method, this paper analyzes GCP’s dynamic impact on firms in the treatment group, thereby testing the parallel trend premise [13].

$$HPFG{I}_{i,t}={\alpha }_0+{{\displaystyle \sum }}_{j=-M}^N{\beta }_{t}Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(5)

$$HPFG{I}_{i,t}={\alpha }_0+{{\displaystyle \sum }}_{j=-M}^N{\beta }_{t}Trea{t}_{i}Pos{t}_t+{\gamma }_0{X}_{i,t}+{\delta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(6)

In the above equations, M and N denote the number of periods before and after GCP’s implementation year, respectively. In Equation (5), the coefficient of Post_t (${\beta }_t$) measures the difference between the treatment and control groups in period t. In Equation (6), the coefficient of the interaction term between Treat_i and Post_t (${\beta }_t$) measures the difference between the treatment and control groups in period t. Other variables are defined to remain consistent with Equation (1). Figure 2 represents the estimation results of the regression coefficients ${\beta }_t$ at 95% confidence intervals. It can be seen that ${\beta }_t$ is insignificant for a number of years prior to GCP’s promulgation, indicating that the treatment and control groups are not significantly different before GCP’s promulgation point and basically satisfy the parallel trend.

5.5.2. Placebo Test

As noted in the previous analysis, the DID presupposes that there is no large variation in firms’ green innovation prior to the policy event (i.e., GCP). The estimated coefficients on the core variables will be insignificant if the policy event is not set in 2012; otherwise, it is likely that there are other unobservable factors that may influence firms’ green innovation [48]. By referring to existing research [5], the placebo test in this study sets the hypothetical time of policy shock as random between 2008 and 2020 to examine whether a policy effect still exists. Figure 3a and Figure 3b report the probability density distribution of the estimated coefficients. The distribution centers around zero and our true baseline estimates from column (1) and (2) of Table 3 (0.0331 and −0.119) are outliers in the placebo test. The placebo test provides evidence that our estimation results are not seriously biased because of any omitted variables [16].

6. Conclusions, Implications, and Limitations

Based on the data of listed firms in China from 2008 to 2020, this study empirically investigates the effect of the Green Credit Guidelines (i.e., GCP) on firms’ green innovation by using the DID model and two classic robustness tests. The influence of heterogeneity and the moderating effect of environmental information disclosure (EID) on the relationship between GCP and firms’ green innovation are also analyzed. The study finds that: (1) GCP promotes green innovation of non-heavily polluting firms (non-HPFs). (2) GCP inhibits green innovation of heavily polluting firms (HPFs). (3) EID strengthens GCP’s promoting effect on green innovation of non-HPFs. (4) GCP’s promoting effect on green innovation of non-HPFs is more prominent in large-sized firms, regions with higher financial development levels, and regions with higher pollution levels. (5) GCP’s inhibiting effect on green innovation of HPFs is less prominent in regions with higher financial development levels.

These above conclusions have several practical implications. First, governments, banks, and firms should cooperate to improve GCP’s design and implementation. We find that GCP improves non-HPFs’ green innovation but imperils HPFs’ green innovation. Consequently, governments should not only continue to enforce GCP as an environmentally friendly regulation but also improve the design of GCP. For example, it could update some clauses of GCP and strengthen the supervision of banks’ lending to firms, especially HPFs. Banks should continue to play the guiding role of GCP for firms and enhance environmental and social risk management in the credit financing process [9,49]. Facing the introduction of GCP, non-HPFs should actively carry out green innovation to better leverage credit resources. HPFs should augment their awareness of environmental protection, accelerate transformation, and realize their own sustainable development through promoting green innovation.

Second, governments should eliminate the “one-size-fits-all” policy as soon as possible. The results of additional analysis indicate that GCP’s promoting effect on green innovation of non-HPFs is more prominent in large-sized firms, regions with higher financial development levels, and regions with higher pollution levels. Additionally, GCP’s inhibiting effect on green innovation of HPFs is less prominent in regions with higher financial development levels. Therefore, governments should consider the differences of all firms’ governance pressures and formulate specific control measures.

Finally, our research finds that EID plays a positive moderating role in the GCP–non-HPFs’ green innovation linkage. Therefore, when formulating and implementing policies, governments, financial institutions, and firms all need to give full consideration to the role of corporate environmental information disclosure, and guide firms, especially non-HPFs, to eliminate environmental information asymmetry through EID.

The research has some limitations which need to be improved in the future. First, we take the implementation of the “Green Credit Guidelines” in China in 2012 as an exogenous shock, and then only select Chinese A-share listed firms as the research objects. We believe the results are country-specific. For example, we find that GCP inhibits HPFs’ green innovation in China. Scholars could use samples in other developing countries to explore this issue. Second, this study only analyzes the heterogeneity of firm size, regional financial development, and regional pollution level. However, the effect of GCP on firms’ green innovation may also vary depending on the level of marketization, the industry, and firms’ internal control systems. Hence, subsequent studies can do more in-depth research.