How Does Climate Policy Uncertainty Affect Green Innovation Among Chinese Companies?

Abstract

Climate change is a major challenge for humanity, with important implications for corporate development. Based on the data of China’s A-share listed companies from 2000 to 2021, this study evaluates the impact of climate policy uncertainty (CPU) on firms’ green innovation using a two-way fixed-effects model. It is found that CPU inhibits firms’ green innovation by reducing government support and external investment, with corporate green innovation declining by 3.3% for each unit increase in CPU. The mechanism studies reveal that CPU has impeded corporate green innovation by reducing corporate social responsibility and increasing corporate financing constraints. The heterogeneity analyses indicate that the negative effect of CPU was more noticeable in state-owned enterprises (SOEs), heavily polluting firms, and firms with a high equity concentration. Further exploration of the economic consequences suggests that CPU has weakened enterprise market competitiveness, including market share, market value, and management efficiency. Climate policy uncertainty is an important external factor for sustainable development, and this paper discusses the impact of climate policy uncertainty on corporate green innovation, which can help to realize the coordinated development of climate policy and corporate green innovation.

1. Background and Introduction

1.1. Introduction

Corporate green innovation is an enterprise economic activity in which enterprises use green energy and green technology to innovate and transform their products and services, in order to achieve energy conservation and consumption reduction. Government policies have a significant impact on green innovation due to green innovation’s high risk, long payback cycle, and uncertain returns [1,2].

However, maintaining the long-term stability of government policies is challenging, as the government bears multiple responsibilities such as economic incentives, environmental protection, and social governance [3,4]. This means that corporate green innovation will lack a stable policy environment. Additionally, global climate change has exacerbated policy instability. Due to the high dependence of corporate green innovation on government policies, there is great value in exploring the impact of CPU. Currently, based on firm behaviors, some scholars have found that climate policy affects corporate finance and governance [5,6,7]. It has also been argued that stable climate policies can promote the development of green technology and protect exchange rate stability [8,9]. Although the existing literature empirically tests the impact of climate policies on firms’ behavior, these studies mainly base their discussion on stable climate policies, ignoring the direct impact of CPU on firms’ green innovation. In addition, most of the existing literature focuses on the positive effects of climate policies on firms’ behavior, and there is little literature that discusses the negative effects of CPU on firms’ green innovation. Thus, this study centers on China and investigates the effects of CPU on corporate green innovation.

In recent years, corporate social responsibility (CSR) and financing constraints have become important factors that affect corporate green innovation [10,11,12,13,14]. This provides ideas for the mechanism analysis in this paper. Some scholars found that CSR can encourage corporate green innovation, while financing constraints are mainly seen as disincentives [15]. Nonetheless, the existing research has neglected how external environmental changes affect CSR and financing constraints. This impact may interfere with corporate green innovation. Therefore, this paper will introduce CSR and financing constraints in the mechanism analysis.

This paper explores the effects of CPU on Chinese firms’ green innovation. Firstly, the benchmark regression shows that CPU has inhibited firms’ green innovation. After considering robustness concerns including endogeneity, alternative dependent variable, alternative independent variables, lagged dependent variable, lagged independent variables, and high-dimensional fixed effect, the baseline conclusion remains valid. Secondly, the willingness of SOEs, highly polluting corporations, and highly concentrated ownership corporations to reduce green innovation in the face of CPU was more pronounced. Moreover, CPU has inhibited firm green innovation by reducing CSR and increasing financing constraints. Finally, the dilemma caused by CPU will further inhibit the competitiveness of enterprises. This means that CPU has reduced the market share, market value, and management efficiency of corporations.

In light of the study above, this paper has three main contributions: First, existing studies remain at the level of the first-order moment characteristics of policies (i.e., a range of climate policies developed and implemented by governments or relevant institutions to address climate change, achieve greenhouse gas emissions’ reduction, and adapt to the impacts of climate change) and do not fully examine their second-order moment characteristics (i.e., CPU, which indicates the degree of change in climate policy). Based on theoretical and empirical analysis, this paper explores the correlation between CPU and corporate green innovation, to deepen the research on the influencing factors of corporate green innovation. Additionally, this paper comprehensively explores the heterogeneous impacts of CPU on corporate green innovation from different levels. These analyses offer various options for firms of different types, with different ownership concentration levels and in different industries, to adapt to CPU.

Second, the existing literature ignores the impact of internal corporate factors such as corporate social responsibility and financing constraints on corporate green innovation. This paper introduces CSR and financing constraints into the analytical framework on the relationship between CPU and corporate green innovation.

Finally, we further discuss how CPU affects firms’ competitiveness, so as to offer empirical findings that can help firms deal with CPU effectively.

The remainder of this paper is organized as follows. Section 2 provides the theoretical analysis and formulates the research hypotheses. Section 3 gives the research design of the paper. Section 4 presents the empirical analysis. Conclusions and policy implications are presented in Section 5.

1.2. Literature Review

1.2.1. Studies on CPU

The global climate environment is currently facing increasingly severe challenges. Current studies on CPU cover three main areas. The first focuses on measuring CPU. Lee measured the Chinese CPU and found that it has rapidly risen recently [16]. This measurement provides a basis for a further exploration of the impacts of CPU. The second is to evaluate how CPU affects macroeconomic development. CPU can raise external investment thresholds and discourage regional economic growth [17,18]. In the face of environmental volatility, the financing cost for high-polluting and high-carbon firms has increased dramatically [19,20]. The third area focuses on assessing how CPU affects corporate risk. Such studies are mainly focused on corporate debt structures. CPU can change corporate debt structures and lead to resource misallocation [21]. Moreover, the prolonged use of short-term debt undermines the competitive advantage of SMEs.

1.2.2. Studies on Policy Uncertainty-Related Corporate Innovation

Scholars today are increasingly concerned with how corporate innovation is influenced by economic policy uncertainty (EPU).

Initially, a number of academics posited that EPU can stimulate corporate innovation [22,23,24,25]. These studies center on the ways in which EPU impacts corporate innovation capacity. External investment and R&D investment are crucial determinants for enhancing corporate innovation capabilities. In the face of EPU, corporations often seek to improve their innovation and R&D efficiency by streamlining decision-making processes, allocating resources more effectively, and fostering a culture of innovation within the organization [26,27,28]. Certain researchers have pointed out the differences between family firms and general firms. They argue that family firms tend to prioritize long-term interests. As a result, when confronted with EPU, family firms are more likely to increase their investment in innovation [29]. Additionally, EPU has the potential to enhance corporate organizational resilience, which contributes to an improvement in innovation capacity [30,31].

Conversely, some experts have contended that EPU acts as a disincentive for firms to innovate [32,33,34,35,36,37,38]. This perspective is underpinned by two primary rationales. The first rationale is related to risk aversion. Economic policy fluctuations may lead some enterprises to avoid investment risks and cut down on R&D investment [39,40,41]. The second rationale is associated with information asymmetry. EPU is closely linked to information asymmetry. External investors prefer a stable market environment due to asymmetric information [42]. In addition, a few scholars have discovered that EPU has a non-linear impact on firm innovation.

While the impacts of EPU on corporate innovation are diverse and complex, other aspects of policy uncertainty also merit attention. Meanwhile, certain scholars have identified the presence of current uncertainty in both monetary and trade policies. This uncertainty has been shown to exert significant impacts on firm innovation. According to Ross et al. [43], both internal and external pressures can encourage company innovation when trade policy is unclear. Yu et al. [44] have argued that monetary policy uncertainty inhibits corporate green innovation.

1.2.3. A Summary

The existing research has extensively discussed the logical relationship between EPU and firm innovation. Some scholars have also found that CPU inhibits macroeconomic development. The existing literature provides new ideas for the study in this paper. Given the aggravation of environmental problems, it is urgent to integrate green innovation into economic development. In particular, whether there is a correlation between CPU and corporate green innovation deserves in-depth research, which has been less addressed in the existing literature. As the basic units of economic development, corporations are worth examining to see whether they will be negatively affected by CPU.

2. Research Hypotheses

How does corporate green innovation become affected by climate policy uncertainty (CPU)? Generally, firms tend to adapt their decision-making and investment choices in the face of environmental volatility [45,46]. Nonetheless, this behavior reduces firms’ productivity, which in turn influences incentives for corporate green innovation.

Unlike general decision-making, corporate green innovation is highly dependent on financial support and is characterized by a high risk and long cycles. Therefore, stable internal support and external investment are key to enterprises achieving green innovation. However, if CPU increases, it may be detrimental to firms’ green innovation behavior.

On the one hand, continued fluctuations in climate policy will diminish external investment. Currently, there are a large number of irrational investors in the market, and frequent adjustments in climate policy can amplify investors’ risk expectations of disasters and uncertainty. In the face of CPU, investors tend to avoid risks and curtail investment in order to maximize profits [47,48]. When external investment decreases, firms are less motivated to engage in green innovation.

Moreover, CPU changes corporate decision-making and reduces green innovation investment. According to the real options theory, in the face of environmental fluctuations, firms adjust their internal corporate decisions to maximize the option value [45]. In the face of CPU, firms are more inclined to hold financial assets with higher option values and reduce green innovation inputs. However, excessive financial investment can distort normal corporate production behaviors, which is not conducive to corporate green innovation.

In addition, policy support also directly impacts firms’ green innovation motivation [49]. When government support is reduced, enterprises are more willing to produce low-cost and high-energy-consumption products, which reduces enterprise green innovation. On this basis, this paper puts forward the following hypotheses.

Hypothesis 1. 

CPU could inhibit corporate green innovation by reducing government support and external investment.

Industry differences expose enterprises to diverse production conditions and varying production levels. High-polluting enterprises generally possess weaker production technologies and management capabilities. Such enterprises are more focused on production costs. As a result, they face greater cost pressure when it comes to carrying out green innovation. Therefore, when confronted with CPU, high-polluting enterprises are less willing to engage in green innovation [50,51,52].

On the other hand, low-polluting firms are more technologically mature and are not over-dependent on natural resources. Moreover, they have a higher probability of improving productivity by leveraging their technological advantages. Additionally, low-polluting firms typically have a stronger risk tolerance and a more pronounced willingness to innovate. Consequently, in the face of CPU, high-polluting enterprises lack incentives for green innovation. On this basis, this paper puts forward the following hypothesis.

Hypothesis 1a. 

The inhibitory effect of CPU on green innovation in high-polluting corporations is greater due to a lower willingness to innovate.

SOEs play a crucial role as the “stabilizer” and “ballast” in economic and social development, effectively promoting economic growth. Compared with non-state-owned corporations, SOEs generally possess greater social influence. Additionally, they often operate on a larger production scale. However, when facing climate policy uncertainty (CPU), SOEs are more likely to reduce their efforts in green innovation. On the one hand, SOEs are burdened with multiple responsibilities. Besides fulfilling their production duties, they must also meet various social and environmental obligations. As a result, the proportion of their R&D expenditure allocated to green innovation is relatively low. In addition, compared with private enterprises, SOEs have more serious principal–agent problems and a higher risk of innovation R&D failure [53]. Due to multiple considerations concerning their own promotion and enterprise risk management, the management of SOEs has a lower willingness to make innovation decisions. On the other hand, some SOEs enjoy monopoly advantages and hold a high market share in their respective industries. These enterprises are able to amass excessive profits by capitalizing on their monopoly positions and market dominance. In contrast, non-state-owned enterprises (NOEs) rely more on their own competitiveness to survive and thrive in the market. Accordingly, in the face of policy changes, SOEs tend to have less incentive for green innovation. Based on this, this paper proposes the following hypothesis.

Hypothesis 1b. 

The inhibitory effect of CPU on green innovation is more pronounced in SOEs than in non-SOEs.

Equity concentration has a significant impact on green innovation in companies. Generally, the equity ownership of major shareholders reflects the firm’s support for innovation and R&D [54]. A relatively balanced equity ownership is conducive to enterprises’ adopting rational decision-making and focusing on long-term corporate development to promote corporate green innovation. However, if the equity ratio of large shareholders is too high, the personal preferences of large shareholders can have a significant impact on corporate innovation. In fact, when confronted with policy uncertainty, large shareholders often tend to avoid high-risk projects and reduce green innovation investment. Therefore, the personal wills of the largest shareholders may act as a deterrent to firms’ green innovation.

In addition, concentrated ownership diminishes the willingness of small and medium-scale shareholders to invest and gives rise to the phenomenon of “free riding”. When small and medium-scale shareholders perceive that their influence on corporate decisions is limited due to concentrated ownership, they may be less motivated to invest in innovation, expecting to benefit from the efforts of others. This phenomenon, in turn, reduces the overall incentive for corporate innovation. On this basis, this paper puts forward the following hypothesis.

Hypothesis 1c. 

When corporations have a concentrated ownership, a higher decision risk leads to a greater inhibitory effect of the CPU on corporate green innovation.

Compared with traditional technological innovation, green technological innovation is more dependent on the resource support of enterprise stakeholders. The existing research shows that corporate social responsibility (CSR) is inextricably linked to corporate environmental governance [55]. A good CSR performance can strengthen the connection between enterprises and stakeholders [56] and prompt enterprises to obtain external support for green innovation, such as technology, capital, and talent. However, CPU weakens CSR and hinders firms’ green innovation. On the one hand, CPU increases the cost of corporate social responsibility. The traditional view is that when the investment is irreversible, the uncertainty of the external environment increases the cost of corporate decision-making and inhibits long-term corporate investment [57]. CSR, as a special long-term holding asset, carries the risk of irreversible investment and an uncertain return. Consequently, in the face of CPU, firms tend to reduce the fulfillment of social responsibility and reduce corporate green innovation. On the other hand, CPU changes the liquidity preference of enterprises. According to risk preference theory, facing policy uncertainty, firms are more cautious in their investment decisions [58]. Therefore, CPU leads firms to retain more cash and reduce CSR. On this basis, this paper puts forward the following hypothesis.

Hypothesis 2. 

CPU inhibits corporate green innovation by reducing CSR.

Financing constraints are a key factor affecting corporate green innovation [44,59,60,61]. The conventional view is that green finance has positive incentives for corporate green innovation. However, this paper argues that CPU leads to financing constraints that are detrimental to corporate green innovation. On the one hand, the information asymmetry between banks and corporations is aggravated by CPU. Due to limited access to information, banks tend to raise lending standards and increase auditing efforts in the face of CPU. This behavior increases corporate financing costs. Currently, most listed companies in China are in transformation. Policy uncertainty leads to increased corporate financing constraints [62]. Reduced bank investment increases the difficulty for enterprises in introducing technology and raises the threshold for technological innovation, which reduces enterprise innovation performance. On the other hand, corporate financing pressure is highly correlated with the financing environment. CPU reduces the risk-taking capacity of enterprises and causes low investor confidence, which in turn exacerbates the difficulty of corporate financing. On this basis, this paper puts forward the following hypothesis.

Hypothesis 3. 

CPU can exacerbate financing constraints, thereby discouraging corporate green innovation.

3. Research Design

3.1. Data Sources

This paper conducts an empirical study using the data of Chinese A-share listed companies spanning from 2000 to 2021, with a total of 2404 companies in the initial sample. The CPU index is from https://figshare.com/articles/dataset/China_s_CPU_index/24071193/1, accessed on 1 July 2024. The data on corporate green innovation are sourced from the China Stock Market & Accounting Research (CSMAR) Database (http://data.csmar.com/, accessed on 1 July 2024). Except for CPU and corporate green innovation, the data in this paper mainly come from two sources: the Wind database (https://www.wind.com.cn/, accessed on 1 July 2024) and CSMAR database (http://data.csmar.com/, accessed on 1 July 2024). Moreover, this paper excludes enterprise data with abnormal values. The final sample obtained consists of 32,314 observations.

Table 1. Descriptive statistics.

Variables	Obs	Mean	SD	Min	Max
CPU	32,314	1.4938	0.6879	0	4.0567
Green	32,314	0.2183	0.5919	0	5.7807
Size	32,314	21.8326	1.2362	17.1218	28.2565
Lev	32,314	0.4478	0.2295	0.0070	9.6988
Roa	32,314	0.0325	0.0882	−1.6806	0.8795
Roe	32,314	0.0210	0.8553	−75.8922	2.3789
Ato	32,314	0.6724	0.5464	0.0015	11.4529
Cashflow	32,314	0.0451	0.0781	−0.7418	0.8759
Rec	32,314	0.1218	0.1067	0	0.8133
Inv	32,314	0.1485	0.1342	0	0.9426
Fixed	32,314	0.2450	0.1713	−0.2062	0.9599
Growth	32,314	12.7253	1169.73	−0.9860	134,607.1
Dturn	32,314	−0.0996	0.5204	−6.8859	4.3290
KZ	31,530	2.0034	2.7812	−15.2971	28.8377
CSR	24,313	3.9756	1.0315	1	8
EnterMarketShare	32,288	0.0257	0.0737	0	1
Bm	30,704	0.9979	1.1524	0.0098	27.8578
Htd	26,607	4.0062	0.2837	1.0986	4.8202

lists the descriptive statistics of the variables in this paper.

3.2. Methodology

On the one hand, considering that factors such as firms’ own characteristics and entrepreneurs’ decision-making preferences are unpredictable and may affect green innovation, the introduction of individual fixed effects can eliminate the interference of such effects. On the other hand, unobservable time-varying factors such as macro-level economic fluctuations, institutional reforms, and environmental changes may have exogenous impacts on green innovation, and the introduction of time-fixed effects can control the influence of these potential factors. Based on the above two reasons, this paper adopts a two-way fixed-effects model to study the impact of CPU on corporate green innovation:

$${Green}_{it}={\beta }_0+{\beta }_1{CPU}_{it}+\gamma Contro{l}_{it}+{\lambda }_i+{\mu }_t+{\epsilon }_{it}$$

(1)

where the firm and year are indicated by $i$ and $t$. CPU is represented by the independent variable ${CPU}_{it}$. The dependent variable ${Green}_{it}$ represents corporate green innovation. Theoretically, there should be a negative correlation between ${Green}_{it}$ and ${CPU}_{it}$, as measured by the coefficient ${\beta }_1$. $Contro{l}_{it}$ is the control variable that may affect corporate green innovation. Firm and year fixed effects are denoted by ${\lambda }_i$ and ${\mu }_t$, respectively.

The specific theoretical logic is shown in Figure 1:

3.3. Variable Selection

3.3.1. Measuring Corporate Green Innovation

Based on the existing research literature [63], we consider the number of green utility models and green invention patents filed to measure the green innovation of enterprises, and the specific method is shown in

Table 2. Variable definitions.

Variables	Definition
CPU	CPU (MacBERT method)
Green	Green (ln(number of green invention patent applications + number of green utility model applications + 1))
Size	Total assets
Lev	Asset/liability ratio
Roa	Net profit/average balance of total assets
Roe	Net profit/average balance of shareholders’ equity
Ato	Operating income/average total assets
Cashflow	Net cash flows from operating activities divided by total assets
Rec	Net accounts receivable to total assets
Inv	The ratio of net inventories to total assets
Fixed	The ratio of net fixed assets to total assets
Growth	Operating income of the current year/operating income of the previous year-1
Dturn	Average monthly turnover of stocks in the current year/average monthly turnover of stocks in the previous year
KZ	KZ (KZ index method)
CSR	CSR (China Securities ESG rating)
EnterMarket-Share	EnterMarketShare (Lerner index) corporate market share
Bm	Bm (book value/total market capitalization) corporate market value
Mfee	Corporate Management Efficiency (overhead divided by revenue)
Htd	Extreme heat

. Compared with design patents, the quantity of applications for green utility models and green invention patents more accurately represents the caliber of corporate innovation. Additionally, the review thresholds for design patents are lower, making it easier for enterprises to meet the review thresholds. Therefore, this paper does not consider applying design patents to measure enterprise green innovation. The limitation of this paper in constructing green innovation indicators for firms is that green patents may not accurately represent actual innovation output, and replacement indicators are needed for robustness testing.

3.3.2. Measuring CPU

This paper’s primary explanatory variable is ${CPU}_{it}$. Facing climate change, there has been an increase in the uncertainty surrounding the development of government policies to address climate change. Based on the existing literature, this paper applies the MacBERT model to measure CPU between 2000 and 2021 [64]. The specific steps are as follows: first, six authoritative Chinese newspapers, including the People’s Daily, Guangming Daily, Economic Daily, Global Times, Science and Technology Daily, and China News Service, are selected as texts based on their credibility, influence, and internationalization level; second, the MacBERT model is constructed to initialize the model parameters, conduct training and assessment, and classify the text news; third, keywords describing climate policy uncertainty are selected from the text news; fourth, the number of news items containing keywords in the total number of articles in the month is recorded in terms of the month and used to calculate the monthly CPU; and fifth, an arithmetic average is applied to the monthly CPU and then multiplied by 100 to obtain the annual CPU data of this paper. According to the regression results for ${\beta }_1$, corporate green innovation is hindered by CPU if the ${\beta }_1$ value is significantly smaller than 0. The limitation of constructing the CPU in this paper is that the CPU is urban data, which may ignore differences in the extent that heterogeneous firms are disturbed by policy uncertainty, and this limitation requires the substitution of variables for robustness testing.

3.3.3. Mechanism Variables

This paper mainly selects corporate financing constraints (KZ) and corporate social responsibility (CSR) as mechanism variables. Among them, this paper selects ESG ratings to indicate corporate social responsibility. Based on the existing literature, this paper comprehensively considers the period of application and coverage of each ESG rating and uses Huazheng ESG ratings as the proxy variable of corporate social responsibility. Additionally, in order to measure corporate financial constraints, this research uses the KZ index approach. Referring to the existing literature, the following steps are taken to construct the KZ index [65]: Firstly, the sample companies are classified according to the ratio of operating cash flow to total assets, the ratio of cash dividends to total assets, the cash holding ratio, the gearing ratio, and the Tobin’s Q. If the ratio of operating cash flow to total assets is lower than the median, KZ1 = 1; if the ratio of cash dividends to total assets is lower than the median, KZ2 = 1; if the cash holding ratio is lower than the median, KZ3 = 1; if the gearing ratio is higher than the median, KZ4 = 1; if the Tobin’s Q is higher than the median, KZ5 = 1. Secondly, the KZ index is calculated as KZ = KZ1 + KZ2 + KZ3 + KZ4 + KZ5. The larger the KZ index, the higher the degree of corporate financing constraints.

3.3.4. Control Variables

This study makes the case that ${Green}_{it}$ may be impacted by a few control variables. The control variables include firm microdata such as firm size (Size), gearing ratio (Lev), and net profit margin on total assets (Roa). Additionally, year fixed effects (Year), industry fixed effects (Ind), and firm fixed effects (Firm) are all controlled for in this article. Table 2 lists the definitions of the variables.

4. Empirical Results and Analysis

4.1. Benchmark Regression

Table 3. Benchmark regression.

	(1)	(2)	(3)	(4)	(5)	(6)
	Green	Green	Green	Green	Green	Green
CPU	−0.032 ***	−0.035 ***	−0.035 ***	−0.032 ***	−0.033 ***	−0.033 ***
	(−2.781)	(−3.149)	(−2.646)	(−4.274)	(−4.456)	(−2.974)
Size		0.100 ***	0.100 ***		0.054 ***	0.054 ***
		(9.317)	(9.128)		(6.291)	(5.385)
Lev		0.093 **	0.093 **		0.053 **	0.053 **
		(2.574)	(2.328)		(2.214)	(2.092)
Roa		0.171 ***	0.171 **		0.053	0.053
		(2.666)	(2.361)		(1.183)	(1.221)
Roe		0.004 *	0.004 *		0.003 *	0.003
		(1.681)	(1.669)		(1.689)	(1.645)
Ato		−0.005	−0.005		−0.043 ***	−0.043 ***
		(−0.497)	(−0.467)		(−3.549)	(−3.714)
Cashflow		0.108 **	0.108 *		−0.025	−0.025
		(2.034)	(1.969)		(−0.641)	(−0.512)
Rec		0.391 ***	0.391 ***		0.322 ***	0.322 ***
		(4.352)	(5.051)		(4.498)	(4.761)
Inv		−0.174 ***	−0.174 **		−0.097 **	−0.097 *
		(−2.880)	(−2.580)		(−2.128)	(−1.918)
Fixed		−0.136 **	−0.136 **		0.108 ***	0.108 **
		(−2.574)	(−2.447)		(3.008)	(2.659)
Growth		−0.000 ***	−0.000 ***		0.000 *	0.000
		(−4.503)	(−4.483)		(1.754)	(1.609)
Dturn		−0.006	−0.006		0.002	0.002
		(−0.846)	(−0.921)		(0.334)	(0.333)
_cons	0.275 ***	−1.955 ***	−1.955 ***	0.266 ***	−0.952 ***	−0.952 ***
	(13.023)	(−8.760)	(−8.410)	(24.001)	(−5.168)	(−4.376)
Year	√	√	√	√	√	√
Ind	√	√	√
Firm				√	√	√
N	30,703	30,703	30,703	32,314	32,314	32,314
R2	0.090	0.129	0.129	0.548	0.552	0.552
F	7.735	11.291	12.281	18.269	7.253	7.273

Notes: The standard errors are the t-value; *, **, and *** represent significance levels of 10%, 5%, and 1%, respectively.

reports the impact of CPU on green innovation in Chinese firms. Column (1), column (2), column (4) and column (5) are clustered at the firm level. Column (1) shows that the estimates are significantly negative, with only year fixed effects and industry fixed effects considered. Based on column (1), control variables are added in column (2), and the estimation results are significantly negative. Column (4) shows that the estimates are significantly negative when considering year fixed effects and firm fixed effects. Based on column (4), control variables are added in column (5), and the estimates are significantly negative. Due to CPU in city-level data, column (3) and column (6) are clustered regressions at the city level. Column (3) and column (6) show that the estimates are significantly negative. The results of the benchmark regression show that CPU in China inhibits corporate green innovation. On the one hand, CPU reduces government support. Policy support is of great significance for firms to carry out innovation activities. Environmental uncertainty may lead local governments to reduce financial expenditure on science and technology, which thus increases the risk of green innovation for firms; on the other hand, CPU reduces external investment. Enterprise green innovation relies on external investment, and environmental fluctuations are prone to distort external investment behavior, resulting in a high degree of unpredictability in enterprise green innovation.

Moreover, there is an increase in the value of R² in columns (2), (3), (5), and (6) compared to columns (1) and (4), indicating that the selected control variables are reasonably effective.

4.2. Robustness Tests

4.2.1. Endogeneity Test

In this research, we examine the impact of CPU on corporate green innovation, and it is possible that endogeneity issues are present in the benchmark regression results. To enhance the reliability of the regression results, this work conducts endogeneity testing using the instrumental variable approach. Referring to the existing literature [66], this paper constructs extremely high temperature (Htd) as an instrumental variable to measure CPU. On the one hand, Htd is associated with climate policy changes, since CPU mainly comes from the frequency of climate policy changes. On the other hand, Htd is an objective climate characteristic, which it is difficult to influence by corporate green innovation. Therefore, Htd satisfies the criteria for instrumental variables in terms of relevance and exogeneity. The process of constructing the indicator of extremely high temperatures includes three steps: first, 1971–2000 is selected as the climatic reference period; second, the crucial value of extremely high temperatures in various cities is ascertained by gathering daily observation data from weather stations; and third, extremely high temperatures are measured by counting the number of days when extremely high temperatures occur in various cities annually.

This paper uses extremely high temperatures as an instrumental variable and applies the two-stage least-squared method (2SLS) for estimation. The 2SLS method is a common solution for analyzing the endogeneity problem. There are two steps in this procedure. In the first step, the dependent variable is regressed with instrumental variables. In the second step of regression, the independent variable is then used to regress the dependent variable.

Table 4. Robustness tests.

	(1)		(2)	(3)	(4)	(5)	(6)
	IV		Lagged Independent Variable	Lagged Dependent Variable	Control for High-Dimensional Fixed Effects	Replacement of Independent Variables	Replacement of the Dependent Variable
	CPU	Green	Green	L.Green	Green	Green	GTFP
CPU		−0.274 *		−0.023 ***	−0.038 ***		−0.063 **
		(−1.687)		(−3.115)	(−3.572)		(−2.550)
L.CPU			−0.018 ***
			(−3.040)
CPU * CIV						−0.060 ***
						(−5.450)
Htd	0.091 ***
	(6.690)
_cons			−0.753 ***	−0.888 ***	−0.939 ***
			(−5.300)	(−5.090)	(−4.126)
Control	√	√	√	√	√	√	√
Year	√	√	√	√		√	√
Firm	√	√	√	√	√	√	√
Year * Province					√
N	26,607	26,607	29,630	29,459	32,314	24,635	22,676
R2		−0.060	0.514	0.536	0.562	0.607	0.987
F		7.313	5.610	6.927	8.810	5.000	92.940

*, **, and *** represent significance levels of 10%, 5%, and 1%, respectively.

’s column (1) reports the regression results. The results show that, first, the instrumental variable extremely high temperature (Htd) significantly increases CPU, and second, corporate green innovation (Green) has a substantial negative correlation with CPU. The benchmark regression’s robustness is confirmed by the regression findings.

4.2.2. Other Robustness Tests

First, the lagged independent variable. CPU is the independent variable in this study. In this paper, the lagged independent variable is used in the regression test. Lagged independent variable regression is beneficial for testing the robustness of the regression results.

Second, the lagged dependent variable. Considering a lag relationship between CPU and corporate green innovation, this paper adopts the lagged dependent variable for the regression test. Since the economic effects generated by CPU may not occur in the current period, lagged dependent variable regression is conducive to testing the robustness of the regression results.

Third, control for high-dimensional fixed effects. Due to the vastness of China, factors such as the economic level and industrial structure of different provinces are difficult to predict and may affect green innovation. The introduction of high-dimensional fixed effects can control for the influence of these potential factors. Consequently, this paper introduces high-dimensional fixed effects such as year*province for regression.

Fourth, replacing the independent variables. Considering that CPU consists of city-level data, and it ignores the differences in the degree of disturbance of heterogeneous firms by policy uncertainty, this paper takes a further step. We simultaneously introduce the interaction term between firm revenue volatility and CPU to construct a firm-level CPU indicator. To this end, this paper replaces the independent variable for robustness testing [67]. Climate policy uncertainty (CPU*CIV) is represented by the interaction term between CPU and corporate income volatility (CIV). This paper refers to existing studies and uses the standard deviation of corporate sales revenue as the basis to measure the abnormal sales revenue of corporations and obtain the CIV. The regression results after replacing the independent variables are shown in column (5) of Table 4.

Fifth, replacing the dependent variable. Since green patents can only reflect the outcomes of enterprises’ green innovation efforts and fail to accurately represent the actual innovation output, this paper substitutes the dependent variable with green total factor productivity (GTFP). Specifically, this paper refers to the super-efficient SBM model and GML index method and chooses the super-efficient SBM-GML model to measure GTFP [68]. The regression results after replacing the dependent variable are shown in column (6) of Table 4.

After lagging the dependent variable, the estimates are significantly negative, indicating a long-running effect of CPU on corporate green innovation. When controlling for high-dimensional fixed effects, the estimates remain significantly negative. Moreover, after replacing both the independent and dependent variables, the significantly negative estimates further confirm that CPU has an inhibitory effect on corporate green innovation. Overall, the robustness tests strongly support the conclusion that CPU inhibits corporate green innovation.

4.3. Mechanism Tests

The above study reveals that CPU has caused enterprise green innovation dilemmas and has significantly inhibited corporate green innovation levels. This study primarily examines how corporate green innovation is impacted by CPU through a combination of corporate financing constraints (KZ) and corporate social responsibility (CSR).

First, this study explores how corporate financing constraints are affected by CPU. Previous studies have shown that the KZ index is designed as a proxy variable for financing constraints, and an increase in the KZ index indicates an increase in financing constraints; second, this study investigates how corporate social responsibility is affected by CPU. Corporate social responsibility mainly manifests itself in enterprises’ contributions to society and the fulfillment of social obligations, which in turn help shape their brand image. This paper selects the Huazheng ESG rating as a proxy variable to measure corporate social responsibility.

The mechanism test results are shown in

Table 5. Mechanism tests.

	(1)	(2)
	KZ	CSR
CPU	0.079 ***	−0.045 ***
	(3.230)	(−3.600)
_cons	4.676 ***	−0.543
	(3.240)	(−1.282)
Control Variables	√	√
Yea	√	√
Firm	√	√
N	31,530	24,313
R2	0.793	0.541
F	2650.890	475.680

*** represent significance levels of 1%.

. Among these, the effect of CPU on corporate financing constraints (KZ) is illustrated in column (1). The estimations are notably negative at the 1% level, as indicated in column (1). The financial constraints increase by 7.9% for every unit increase in CPU. The results show that CPU has affected corporate financing constraints. The effect of CPU on corporate social responsibility (CSR) is illustrated in column (2). The estimations are notably negative at the 1% level, as indicated in column (2). Corporate social responsibility declines by 4.5% for every unit increase in CPU. The findings demonstrate that corporate social responsibility has been impacted by CPU. This paper finds that corporate financing constraints and corporate social responsibility are important factors through which CPU inhibits corporate green innovation. Therefore, hypotheses 2 and 3 of this paper are verified.

Overall, on the one hand, financing constraints imply an increase in corporate financing risk. Due to higher financing constraints, firms can hardly obtain funds from outside sources, and this leads to an increase in corporate financing costs. Faced with CPU, companies with low corporate risks are preferred by banks and external investors. This tendency reduces investors’ investment in green innovative enterprises. Increased pressure on financing constraints can continue to squeeze the space for green innovation, inhibiting corporate green innovation. Therefore, increasing enterprise financing constraints has inhibited enterprise green innovation; on the other hand, due to CPU, corporations are more inclined to exercise risk aversion and reduce information disclosure. Changes in the external environment have exacerbated the difficulty of corporate development and reduced the willingness of companies to participate in ESG ratings. Moreover, external investors’ investment confidence declines due to reduced corporate social responsibility and continuously reduced corporate investment in green innovation. Therefore, corporate social responsibility will decline due to greater CPU, which will limit corporate green innovation. In contrast to the existing literature, this paper reveals how CPU affects corporate green innovation through financing constraints and corporate social responsibility. These findings can help government departments focus on achieving coherence between climate policies, corporate financing, and corporate social responsibility and strengthening the cooperation between governments and financial institutions.

4.4. Heterogeneity Tests

The preceding research concentrates on how CPU affects corporate green innovation on average. The heterogeneous impact of CPU is further examined in this section.

4.4.1. Industry Heterogeneity

Among the 16 industries listed as highly polluting in the China Securities Regulatory Commission’s 2012 revision of the Guidelines for the Industry Classification of Listed Companies are the manufacturing of non-metallic mineral goods, textile manufacturing, metal product manufacturing, ferrous metal smelting and pressing, non-ferrous metal smelting and pressing, and pharmaceutical manufacturing. Except for these highly polluting industries, the industries are non-highly polluting.

Table 6. Heterogeneity tests.

	(1)	(2)	(3)	(4)	(5)	(6)
	Highly Polluting Enterprises	Non-Highly Polluting Enterprises	State-Owned Enterprises	Non-State-Owned Enterprises	Highly Concentrated Ownership Corporations	Lowly Concentrated Ownership Corporations
	Green	Green	Green	Green	Green	Green
CPU	−0.054 ***	−0.024 ***	−0.041 ***	−0.022 **	−0.032 ***	−0.019 *
	(−3.728)	(−2.749)	(−2.945)	(−2.587)	(−3.234)	(−1.826)
_cons	−1.323 ***	−1.086 ***	−0.433	−0.999 ***	−0.836 ***	−1.089 ***
	(−3.516)	(−4.768)	(−1.086)	(−3.844)	(−3.286)	(−4.082)
Control	√	√	√	√	√	√
Year	√	√	√	√	√	√
Firm	√	√	√	√	√	√
N	8639	23,651	10,741	17,073	17,951	14,208
R2	0.442	0.589	0.584	0.596	0.566	0.612
F	2.734	6.392	4.205	4.081	4.964	3.629

*, **, and *** represent significance levels of 10%, 5%, and 1%, respectively.

’s columns (1) and (2) display the estimation findings. The findings indicate that corporate green innovation in highly polluting industries is more severely inhibited by CPU than in non-highly polluting industries. For a one-unit increase in CPU, green innovation decreases by 5.4% for high-polluting firms and by 2.4% for non-highly-polluting firms. High-polluting industries have high costs and low profits compared to non-highly polluting industries. High-polluting firms focus more on short-term gains and aim to maximize short-term economic profits. Therefore, the risk-bearing ability of high-polluting enterprises is relatively poor, and they are more inclined to reduce green innovation. Hypothesis 1a of this paper is verified.

4.4.2. Property Rights Heterogeneity

Given the diversity of property rights held by firms, the impact of CPU on corporate green innovation may vary according to the type of property rights. According to the difference in enterprise ownership, this paper divides enterprises into SOEs and NOEs. Table 6’s columns (3) and (4) display the findings. The findings indicate that CPU has a stronger inhibitory impact on green innovation for SOEs than for non-SOEs. For a one-unit increase in CPU, green innovation decreases by 4.1% for SOEs and 2.2% for non-SOEs. SOEs have to consider both economic and social factors. Consequently, SOEs will formulate development plans according to government requirements and lack operational autonomy. Furthermore, SOEs need to assume social responsibility. SOEs are more inclined to decrease green innovation when confronted with CPU. Hypothesis 1b of this paper is verified.

4.4.3. Concentrated Ownership Heterogeneity

Corporations with varying levels of concentrated ownership may experience different impacts of CPU in terms of green innovation. This paper classifies firms into highly concentrated ownership corporations and lowly concentrated ownership corporations. Enterprises with the proportion of shares held by the first largest shareholder greater than the sample median are regarded as highly concentrated ownership corporations. The findings are shown in Table 6’s columns (5) and (6). The results show that the inhibiting impact of CPU on green innovation is stronger for highly concentrated ownership corporations than for lowly concentrated ownership corporations. A one-unit increase in CPU reduces green innovation by 3.2% for high-equity-concentration firms and 1.9% for low-equity-concentration firms. Firms’ resource allocation is at risk when the largest shareholder’s shareholding is too high. The largest shareholders are more willing to avoid risks to obtain short-term interests, which reduces corporate green innovation. Hypothesis 1c of this paper is verified.

4.5. Economic Consequences Tests

As previously mentioned, corporate green innovation may be negatively affected by CPU. What are the economic consequences of this negative impact? This paper measures enterprise competitiveness through three aspects: enterprise market share (EnterMarketShare), enterprise market value (Bm), and enterprise management efficiency (Mfee). The regression results are displayed in

Table 7. Economic consequences tests.

	(1)	(2)	(3)
	EnterMarketShare	Bm	Mfee
CPU	−0.002 *	−0.026 *	−0.009 *
	(−1.754)	(−1.866)	(−1.844)
_cons	−0.266 ***	−11.212 ***	1.421 ***
	(−6.132)	(−13.612)	(2.614)
Control Variables	√	√	√
Year	√	√	√
Firm	√	√	√
N	32,288	30,704	30,701
R2	0.576	0.641	0.114
F	74.020	172.130	22.092

*, and *** represent significance levels of 10%, and 1%, respectively.

’s columns (1), (2), and (3). The outcome demonstrates that enterprise market share (EnterMarketShare) has been limited by CPU. Enterprise market share will drop by 3.2% for every unit increase in CPU. The result indicates that CPU has a restraining effect on enterprise market value (Bm). Specifically, a one-unit increase in CPU would lead to a 2.6% reduction in the enterprise market value of firms. The result shows that CPU has inhibited enterprise management efficiency (Mfee). Enterprise management efficiency will drop by 0.9% for every unit increase in CPU. Overall, CPU has inhibited enterprise competitiveness, and the negative effect of CPU on corporate development is well reflected in the decline in market share, the reduction in market value, and the decrease in the efficiency of corporate management. On the one hand, facing CPU, enterprises may adopt conservative innovation strategies and policy choices, thus missing market opportunities and ultimately reducing their market share, On the other hand, CPU makes it difficult for enterprises to accurately predict the external environment, resulting in a decrease in the efficiency of enterprise resource allocation and a subsequent reduction in the enterprise’s value. In addition, CPU may reduce the enterprise’s risk appetite and limit the enterprise’s investment behavior, which leads to a decrease in the enterprise’s investment decision-making efficiency. In the long run, corporate governance and corporate innovation are essential to achieving sustainable economic development [69]. Therefore, these negative impacts need to be taken into account.

5. Conclusions and Policy Implications

Based on the data of China’s A-share listed companies from 2000 to 2021, this paper explores the impact of CPU on corporate green innovation, and the study shows the following: (1) CPU inhibits corporate green innovation, and corporate green innovation decreases by 3.3% for every one-unit increase in CPU. (2) The impact of CPU on corporate green innovation shows an obvious heterogeneity, mainly reflected in the more obvious inhibition of the green innovation of state-owned enterprises, high-pollution enterprises, and enterprises with highly concentrated shareholdings by CPU. (3) On the one hand, CPU causes enterprises to face greater financing constraints, inhibiting corporate green innovation, and on the other hand, CPU reduces the fulfilment of corporate social responsibility, leading to a decline in corporate green innovation. (4) The analysis of economic consequences finds that CPU leads to a significant decline in corporate competitiveness, manifested in a decline in corporate market share, market value, and management efficiency. Compared with the existing literature, this paper provides a new perspective (CPU) and comprehensively explores the impact of CPU on corporate green innovation from different levels. In addition, this paper introduces corporate social responsibility and financing constraints into the analytical framework of the relationship between CPU and corporate green innovation.

Based on the above research findings, the following policy recommendations are put forward: First, central policymaking bodies and local institutions need to maintain stability and regularity in climate policy. On the one hand, the government should clarify the continuity of climate policy, stabilize market expectations, and avoid a high frequency of policy adjustments. On the other hand, the government should increase the transparency of climate policy implementation, strengthen communication with corporation entities, and reduce the adverse impacts of climate policy changes.

Second, relevant departments should promote the coordinated development of green innovation in different types of enterprises. On the one hand, differentiated policies should be formulated for different types of enterprises to avoid one-size-fits-all system management and to achieve the coordinated development of green innovation among Chinese enterprises. On the other hand, it is necessary to optimize the market structure, expand the demand for products, and enhance the confidence of enterprises in green innovation.

Third, it is necessary to improve the financing environment and enhance corporate social responsibility. On the one hand, there is a need to improve financing support policies, build a service-oriented financial system, and mitigate the unfavorable impact of the CPU on enterprise financing. On the other hand, CSR should be included in the screening criteria for enterprise green innovation funding recipients to strengthen the sense of corporate social responsibility.

Fourth, it is necessary to enhance the competitiveness of enterprises in the market. Governments should guarantee orderly market competition and bolster support for green innovation among small and medium-sized enterprises. Furthermore, governments should strengthen market regulation and guarantee accurate information disclosure. It is also necessary to increase publicity to encourage enterprises to actively fulfill their social responsibilities and improve their level of green innovation.