Does CSR Improve the Quality of Economic Growth? Based on the Perspective of Green Innovation

Abstract

Taking into consideration existing research on corporate social responsibility (CSR) and green innovation, this study categorizes green innovation into substantive and strategic types. For this study, we constructed a general equilibrium model that integrates the effects of CSR on substantive and strategic green innovation, exploring changes in emission reduction technologies caused by firms’ CSR efforts and their impact on economic growth quality. We derived the economic growth trajectory for substantive green innovation and strategic green innovation as a function of CSR. The theoretical model that we developed in this study shows that CSR can improve economic growth quality. To empirically test our theoretical model, we used data at the provincial level in China from 2013 to 2022; these empirical results are consistent with the theoretical model. In addition, robustness tests and endogeneity issues were conducted; our findings from these tests show that substantive green innovation acts as a transmission mechanism through which CSR promotes economic growth quality. Additionally, the credit preferences of financial institutions positively moderate the relationship between CSR and economic growth quality. This study provides valuable insights for firms aiming to fulfill CSR obligations and enhance their capability in substantive green innovation.

1. Introduction

The long-term survival of humanity is challenged by environmental issues. Environmental issues such as climate change, air pollution, and biodiversity loss have been widely disseminated in society to raise public awareness. The rapid economic growth of China in the early years had a negative impact on the environment, impeding the holistic and sustainable development of society [1]. Sustainable economic growth is innovative, efficiency, energy-saving, and environmentally friendly, and its essence is green economic growth [2]. With the new development concept, sustainable economic growth focuses on technological innovation and ecological civilization as part of green development [3]. Successful implementation of sustainable economic growth is not only dependent on changes in industrial structure but also on the “firms” as key actors [4]. Firms play a key role in driving green economic development. Environmental legislation further forces firms to take appropriate and timely steps to mitigate their economic, environmental, and social impacts [5]. Thus, both academia and industry are interested in exploring ways to accelerate the implementation of green development principles by firms and to improve economic growth quality.

As high-quality economic development has become increasingly ingrained in society, Corporate Social Responsibility (CSR) has gained increasing importance [6]. Firms play an extremely important role as both creators of wealth and emitters of pollution in achieving high-quality economic development [5,7]. In the course of corporate development, the focus has shifted from simply maximizing profits to fulfilling CSR while pursuing profits [8]. To practice CSR, firms must consider the interests of stakeholders such as the environment, society, and consumers. Firms should implement sustainable growth targets by actively responding to social development needs and investing in green innovation and product transformation [9]. In terms of national policy, Xi Jinping has delivered a number of important speeches on the social responsibility work of state-owned central enterprises since the 18th CPC (the Communist Party of China) National Congress. With the establishment of the Social Responsibility Bureau in March 2022, SASAC (State-owned Assets Supervision and Administration Commission of the State Council) further emphasizes the importance that the government attaches to firms fulfilling their social responsibility. In December 2023, the Blue Book on Social Responsibility of Central Enterprises (2023) and other similar publications were centrally published, emphasizing the comprehensive integration of social responsibility into low-carbon strategies, major decision-making processes, operations, and management, as well as the cultural construction of enterprises. Therefore, the impact mechanism of CSR on economic growth quality and its effectiveness are of great concern to policymakers and academicians.

To accomplish CSR objectives and to drive economic growth, green innovation is an essential component. Addressing environmental concerns, green innovation has become a key tool for energy restructuring and industrial technological upgrading [4]. Furthermore, different motivations for innovation can lead to differential effects on environmental improvement when firms implement high-quality economic development objectives. Consequently, scholars have categorized green innovation into substantive and strategic types [10,11,12]. Substantive green innovation refers to a strategic goal that focuses on the symbiotic development of firms and the ecological environment, aiming to reduce emissions through changes in green technology [13,14]. Strategic green innovation refers to “innovative” activities aimed at obtaining government subsidies for emission reduction and avoiding environmental regulation, which is not an enhancement of real emission reduction technological innovation capacity and is difficult to achieve substantive reduction effects [14,15]. Compared with previous economic growth macro models, our research model incorporates the endogenization of green innovation and further examines the impact of two types of green innovation on high-quality economic growth.

In the context of high-quality economic development, our study addresses the following questions:

• Does CSR significantly influence the two types of green innovation?
• How do the two types of green innovations affect the quality of economic growth and what are the transmission mechanisms?
• How does the differential interest rate policy implemented by financial institutions based on a firm’s CSR performance contribute further to improving the quality of economic growth?

To answer the aforementioned questions, we develop a general equilibrium model that integrates the effects of CSR on substantive and strategic green innovation, exploring the changes in emission reduction technologies resulting from firms fulfilling CSR and their impact on the quality of economic growth. Subsequently, to empirically test our theoretical model, this paper uses data at the provincial level in China from 2013 to 2022. Our findings indicate that CSR can effectively reduce environmental pollution per unit of output by promoting substantive green innovation, thereby improving the quality of economic growth. In addition, the empirical test shows that the credit preferences of financial institutions positively moderate the relationship between CSR and the quality of economic growth. The empirical evidence strongly supports the mechanism proposed by our theoretical model, and robustness tests and endogeneity issues have been conducted. Our study has significant implications for enhancing and expanding existing theories of CSR and economic growth quality.

This study makes contributions to the existing research from three aspects. The first is the literature on the driving force of economic growth. Existing research has primarily focused on exploring the impact on economic quality in terms of industrial structure, human capital investment, institutional environment, technological progress, and trade openness [16,17,18]. These studies concentrate on the impact of macro-external factors on economic growth, whereas our study considers the impact of CSR on firms’ innovation decisions and economic quality. Drawing upon the inherent attributes of CSR, this study analyzes it as an internal driving force for high-quality development. The second is related to the literature on green innovation. Most of the existing literature categorizes green innovation into substantive and strategic types based on innovation motivations [13,15]. This body of research considers factors such as carbon emissions trading policies, government R&D subsidies, and environmental regulations to explore their impact on the two types of green innovation [10,11,19]. Unlike previous studies in this field, our study employs CSR as endogenous information regarding firms’ selection of two types of green innovations, examining their impact on the quality of economic growth. Finally, this study contributes to the literature on the extension of the general equilibrium model [20,21,22]. By endogenizing green innovation and technological change into a general equilibrium model, the aforementioned study primarily investigates its impact on environmental pollution. Based on the above research, we integrate CSR as an endogenous driver of substantive green innovation in our general equilibrium model. We derive the economic growth trajectory at market equilibrium, incorporating both substantial green innovation and strategic green innovation. Our study expands the theory of synergistic promotion of CSR and green innovation, as well as the integration of CSR and high-quality economic development. This research is crucial for promoting the development of a policy system that ensures corporations fulfill their social responsibilities and formulate appropriate environmental policies.

According to Figure 1, this paper is structured as follows: In Section 2, we review the related literature. In Section 3, we construct the general equilibrium model and derive the equilibrium outcome. In Section 4, we introduce the data and empirical design. In Section 5, we provide the empirical results. In Section 6, we summarize this study and offer policy implications.

2. Literature Review

As the global society’s demand for sustainable development continues to increase, the concept and practice of Corporate Social Responsibility (CSR) are constantly evolving and expanding. The concept of CSR involves companies creating economic, social, and environmental value for a wide range of stakeholders, including shareholders, suppliers, consumers, governments, communities, and the environment. This is achieved through the implementation of strategic practices such as corporate governance and operations management [23].The ethical foundation of CSR lies in actively balancing economic interests with social responsibilities to promote the enhancement of societal well-being and achieve sustainable development [24,25]. CSR practices can enhance a company’s social image and reputation, simultaneously contributing to the realization of low-carbon sustainable development. In organizational behavior studies, CSR is regarded as a factor influencing organizational behavior. It not only shapes the social image of the organization but also impacts employee behaviors, attitudes, and interactions with external stakeholders [26,27]. As micro-units of the economy and society, firms face critical questions regarding whether they should assume social responsibility, what types of social responsibility they should undertake, and how to fulfill these responsibilities. These issues are widely researched and discussed in both academia and business [28]. Previous scholars have extensively studied the impact of CSR on corporate value, risk, and governance using empirical methods [29,30,31]. Furthermore, China’s goal of high-quality development is closely tied to CSR concepts, which emphasize the importance of integrating social and environmental concerns into economic development.

Research has examined a wide range of factors affecting the quality of economic growth, including industry structure, human capital investment, institutional environment, technological progress, and trade openness [16,17,18]. CSR concepts are closely linked to China’s goal of high-quality development. There are two main categories of research on CSR and its impact on quality economic development. The first category focuses on the relationship between CSR and environmental sustainability from the perspectives of systemic risk, institutional theory, and multidimensional CSR. Duanmu et al. [32] describe socially responsible investment (SRI) as a differentiated emission reduction strategy. Yu et al. [33] found that firms actively fulfill their environmental responsibility by reducing carbon emissions, which not only reduces systemic risks but also stimulates consumers’ low-carbon preferences. He et al. [34] analyzed corporate responsibility for emission reduction from five perspectives: identification, implementation, compensation, disclosure, and assessment of its effect. In the other category, game theory is used to analyze the optimization strategy problem of CSR on environmental sustainability in supply chains. To examine the effect of manufacturers practicing CSR on the equilibrium results of the supply chain, Panda et al. [35] developed a two-stage supply chain model. From a green supply chain management perspective, Seman et al. [36] studied CSR’s impact on environmental performance. Additionally, there is a body of literature that explores the impact of CSR on economic growth from the perspectives of different countries and industries. Yousefian et al. [37] analyze the relationship between CSR performance and the economic growth of European mining corporations. Sharma and Sathish [38] examine whether CSR expenses shown by the banks promote the sustainability of an emerging economy like India. In this paper, we extend the literature on CSR and high-quality economic development to incorporate green innovations. By integrating CSR, green innovation, and economic growth quality, the economic impacts of CSR can be further enriched.

Firms have increasingly embraced green innovation as a strategic approach to fulfilling CSR. Institutional theory asserts that firm behavior is shaped by social, political, and cultural institutions, in addition to market forces. To gain societal and market acceptance, firms must bolster their legitimacy. Implementing CSR and green innovation helps firms align with societal expectations and regulatory standards, thereby enhancing their legitimacy and reputation [39]. Economic growth and sustainable development can be reconciled through green innovation, which can decrease the negative environmental impact of firms’ production processes [2]. The existing literature on the factors influencing firm green innovation can be categorized into two main types [40]. The first category involves macro-level research, which mainly focuses on investigating the effects of external regulations such as environmental regulations [2], environmental policies [41], and government environmental supervision [19] on firm green innovation, and has reached valuable conclusions. The second type is micro-level research, which primarily investigates the impact of green innovation in five aspects: corporate environmental management factors, corporate governance factors, corporate resources, capabilities, and strategies, as well as stakeholder factors [42,43,44]. Furthermore, different motives for innovation may lead to different effects when firms pursue sustainable growth goals. Therefore, several studies have classified innovation into substantive green innovation and strategic green innovation [13,15]. Based on risk aversion theory, firms facing stringent environmental regulations and growing public awareness may engage deeply in substantial green innovation to mitigate market, legal, and reputational risks [45]. The existing literature explores the impact of these two types of green innovations on firms’ environmental performance, social reputation, and market competitiveness [3,8,43]. However, few studies have integrated CSR, substantive green innovation, and strategic green innovation into a general equilibrium model framework to investigate the mechanisms influencing economic growth quality. In our study, we extend the “Porter hypothesis” by illustrating how to encourage companies to comply with their social responsibilities through substantive green innovation activities.

The above literature review indicates that CSR and environmental performance, as well as the drivers of green innovation, have been extensively studied. However, from the perspective of the green innovation category, existing literature has not thoroughly investigated the impact of CSR on the quality of economic growth. Furthermore, few studies have examined the mechanisms through which CSR contributes to improving economic growth quality by incorporating CSR and substantive/strategic green innovation into general equilibrium models of sustainable growth. Using a general equilibrium model of endogenous emission reduction technology (a flow chart of theoretical analysis is shown in Figure 2), we aim to analyze how CSR influences the two types of green innovation. Additionally, we examine the changes in emission reduction technologies caused by firms fulfilling CSR and their impact on economic growth quality.

3. Theoretical Model

Firstly, this paper categorizes green innovation into substantive green innovation and strategic green innovation based on motivation theory. Secondly, this study constructs a general equilibrium model that integrates the effects of CSR on substantive and strategic green innovation, exploring the changes in emission reduction technologies resulting from firms fulfilling CSR and their impact on economic growth quality. Finally, the paper analyzes the transmission mechanism through which CSR affects economic growth quality by comparing the economic growth paths in the theoretical model.

3.1. Model Structure

3.1.1. Households

In this economic framework, representative households can benefit from both consumption and a favorable environment. The utility function can be defined as follows in the literature [46,47]:

$$U={\displaystyle \frac{C^{1-\sigma }-1}{1-\sigma }}+{\displaystyle \frac{E^{1+\omega }}{1+\omega }},\sigma >0,\omega >0$$

(1)

where $C$ represents the consumption, $\sigma$ represents the relative risk aversion coefficient, E represents the environmental quality and $\omega$ denotes the preference of the people for the environment. As $\omega$ increases, people’s environmental awareness and concern also increase.

3.1.2. Production and Environment Pollution

According to the most relevant literature [2,48], the labor market need not be considered since it does not affect the dynamic relationship between the economy and the environment. As firms utilize capital and natural resources for production, while the environment also influences production processes, the production function is as follows:

$$Y=F(E,K_Y,N)=A(E)K_Y^{\alpha }{N}^{\beta }=A{E}^{\delta }{K}_Y^{\alpha }{N}^{\beta }$$

(2)

where $K_Y$ denotes capital used for production, $N$ denotes collected resources, and $A(E)$ denotes the efficiency parameter of production [49]. A favorable work environment can enhance both the physical and psychological efficiency of workers. $E$ denotes the environmental quality, which includes clean air, clean water sources, and forests, while $N$ refers to the natural resources utilized in production, such as collected water, harvested trees, and extracted minerals.

The environment and economy are interconnected. On one hand, the environmental quality $E$ affects residents’ utility and production. On the other hand, firms need resources $N$ for production. Moreover, resource extraction can harm the environment. Based on the classical assumptions in the environmental economics literature [20,50], the environmental movement equation is as follows:

$$\stackrel{•}{E}=\vartheta E(E_0-E)-P(N,H)=Z(E,N,H),(0<E<E_0)$$

(3)

where $E$ is the environmental quality, and $\stackrel{•}{E}$ is the rate of change of the environmental quality over time, which is determined by the environmental self-recovery capacity $\vartheta E(E_0-E)$ and pollution $P$. The self-recovery capacity of the environment is related to the deviation between the current level $E$ and the initial state $E_0$. Firms extract resources $N$ for production, and resource extraction can harm the environment and cause pollution. Simultaneously, emission reduction technology reduces the level of pollution. Therefore, environmental pollution $P$ is related to $N$ and emission reduction technology $H$. Setting $P={\displaystyle \frac{N}{H}}$, where $N$ represents the level of natural resource consumption, $H$ represents emission reduction technology and a larger $H$ indicates that pollution resulting from unit resource consumption is lower.

3.1.3. Green Innovation Sector

The main innovations are the theoretical model that incorporates CSR in the innovation sector and derives the economic growth trajectory for substantive green innovation and strategic green innovation as a function of CSR. In this paper, green innovation differs from the clean technology selection model, which is an endogenous technology developed through innovative research to mitigate pollution. Consequently, there is a green innovation sector that develops innovative emission reduction technology that can reduce pollution emissions directly. However, owing to the high costs, long-term nature, significant risks, and substantial financial investment associated with emission reduction technology research and development, all of which entail “cost–benefit” uncertainty, the innovation sector has to choose between substantive green innovation and strategic green innovation.

Using $H$ to measure the level of innovative technology that reduces pollution:

$$H=\{\begin{array}{cc}\hfill \lambda CSR{A}_0{K}_H^{\mu },& adoptingsubstantivegreeninnovation\hfill \\ \hfill A_0{H}_0,& adoptingstrategicgreeninnovation\hfill \end{array}$$

(4)

where $\lambda$ represents the degree of improvement in research and development (R&D) capabilities after adopting substantive green innovation, and CSR represents the degree of an enterprise’s fulfillment of corporate social responsibility. $A_0$ represents the existing R&D capabilities, $K_H$ represents the R&D investment in substantive green innovation. Knowledge spillover implies $\mu >1$. As R&D investment in green innovation has certain thresholds, minimal investments are insufficient to achieve substantive green innovation in emission reduction technology. Hence, the innovation sector opting for strategic green innovation will choose inherent emission reduction technology, and $H_0$ representing the existing level of emission reduction technology. In this case, R&D investment $K_H=0$.

Firms’ strategic decision-making regarding green innovation, coupled with their commitment to CSR, contributes to the implementation and advancement of innovative strategies for emission reduction technology research and development. Firms must transform from strategic green innovation to substantive green innovation to achieve multiple strategic objectives, including pollution reduction and energy consumption reduction. Firms can enhance their investment in innovation and foster a positive innovation climate by fulfilling their social responsibilities. As a result, innovation sectors integrating CSR principles into emission reduction technology research and development can drive substantive green innovation. Investments in emission reduction innovative technology are provided by financial institutions.

3.1.4. Financial Institutions

Another innovative aim of our paper is to establish a functional relationship between the preferential credit offerings by financial institutions and the level of CSR. According to our model, banks serve as the sole financial institutions in the economy, receiving deposits and disbursing loans. Financial institutions are assumed to offer preferential interest rates to the green innovation sector. However, the potential strategic green innovation behaviors of the green innovation sector pose challenges to the screening of financial institutions, leading to mismatches in financial resources. Therefore, our study uses CSR in the green innovation sector as a criterion for interest rate incentives. In the modern financial system, CSR plays a key role in balancing the relationship between the economy, society, and the environment. Considering the resource allocation based on green innovation investment, a higher level of CSR implies that firms allocate more resources to emission reduction technology research, thereby strengthening the investment of substantive green innovation. As a result, firms that demonstrate good CSR are more likely to engage in substantive green innovation.

The firm’s total capital $K$ is provided by banks, which is allocated to both the product manufacturing sector $K_Y$ and green innovation sector for research investments $K_H$, so we have $K=K_Y+K_H$.The profit function of banks is ${\pi }_B=(CSR)R_H{K}_H+R_Y{K}_Y-rK$, where $R_H$ represents the loan interest rates for the green innovation sector and $R_Y$ represents the loan interest rates for the product manufacturing sector, while $r$ representing the deposit interest rate. Taking into account the level of CSR fulfillment, green innovation sectors receive different interest rates, and in an equilibrium of competition, these different interest rates will be reflected in different credit amounts. The greater the CSR, the smaller K_H in equilibrium, indicating lower interest rates and stronger subsidies in the green innovation sector. In addition, banks are allocating a proportion of their capital, denoted by ${\eta }_{CSR}$ to be invested in the green innovation department, so we have: $K_H={\eta }_{CSR}K$, $K_Y=(1-{\eta }_{CSR})K$. Furthermore,

$${\pi }_B=(CSR){\eta }_{CSR}{R}_{H}K+R_Y(1-{\eta }_{CSR})K-rK$$

(5)

3.1.5. Environment Policy

The government imposes pollution taxes, with a tax rate of $p_p$ per unit of pollution. The generated tax revenue is transferred to households through transfer payments. Simultaneously, resources are publicly owned, and companies are obligated to purchase resources from households rather than exploiting them for free. The unit price of resources is determined by the equilibrium in the resource trading market. Additionally, it is assumed that both sellers and buyers in the resource trading market are price takers.

3.2. Social Planner’s Problem

The Planner’s Problem is

$$\max {\displaystyle {\int }_0^{\infty }[\frac{C_t^{1-\sigma }-1}{1-\sigma }+\frac{E_t^{1+\omega }}{1+\omega }]e^{-\rho t}}dt$$

$$s.t.\hspace{1em}\begin{array}{c}Y=A{E}^{\delta }{(1-\eta )}^{\alpha }{N}^{\beta }\end{array}$$

$$\stackrel{•}{K}=Y-C+\nu N-p_H\cdot H-p_p\cdot E$$

$$\stackrel{•}{E}=\vartheta E(E_0-E)-{\displaystyle \raisebox{1ex}{$N$}\!\left/ \!\raisebox{-1ex}{$H$}\right.}$$

At this point, it is necessary to solve this maximization problem. The current value Hamiltonian function is set as:

$$H={\displaystyle \frac{C^{1-\sigma }-1}{1-\sigma }}+{\displaystyle \frac{E_t^{1+\omega }}{1+\omega }}+{\lambda }_1(rK-C+\nu N-p_H\cdot H-p_p\cdot E)+{\lambda }_2[\vartheta E(E_0-E)-{\displaystyle \raisebox{1ex}{$N$}\!\left/ \!\raisebox{-1ex}{$H$}\right.}]$$

C, Y and $N$ are control variables. $K$ and $E$ are state variables in this optimal control problem. The first-order conditions are:

$$H_{\eta }=0\to \eta ={\displaystyle \frac{\beta \mu }{\beta \mu +\alpha }}$$

Due to the Cobb-Douglas production function and additive utility function, both trade-offs for consumption and the environment affect growth paths, but this effect is only reflected in the equilibrium evolution path of the environment and not directly in capital allocation calculations. As a result, in the theoretical optimum, the social planner makes the optimal decision in two steps, allowing the production sector to choose the optimal capital allocation and the consumer to choose the optimal consumption path. Economic growth can be described by the following system of equations:

$$\eta ={\displaystyle \frac{\beta \mu }{\beta \mu +\alpha }}$$

(6)

$${\displaystyle \frac{\stackrel{•}{C}}{C}}={\displaystyle \frac{1}{\sigma }}(Y_{K_Y}+Y_{N}H{Z}_H-\rho )$$

(7)

$${\displaystyle \frac{E^{\omega }}{U_C{Y}_{N}H}}+{\displaystyle \frac{\delta N}{\beta HE}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{Y_N}}{Y_N}}-{\displaystyle \frac{\stackrel{•}{H}}{H}}$$

(8)

Equations (6)–(8) and the transversality condition $\underset{t\to \infty }{\lim }{\lambda }_{1}K{e}^{-\rho t}=0;\underset{t\to \infty }{\lim }{\lambda }_{2}E{e}^{-\rho t}=0$ constitute the solution of this dynamical system. According to the Social Planner’s Problem, Equation (6) shows the optimal credit resource allocation between the green innovation sector and the production sector. Equation (7) is the Ramsey-Keynes rule. Equation (8) is the movement path of environmental pollution, and this path is related not only to consumption but also to the level of abatement technology in the green innovation sector.

3.3. Competitive Equilibrium

3.3.1. The Firm Decisions

The firm does not pay for environmental impact. The price of resources $\nu$ is determined by market equilibrium, while pollution tax $p_p$ is set by the government. Capital price $R_Y$ is determined by credit market equilibrium. The price of emission reduction technology $p_H$ is determined by equilibrium transactions between firms and the green innovation sector. To begin with, the price of the final product is normalized to 1. Then, using the profit maximization objective function, the firm’s first-order condition can be derived as $F_{K_Y}=R_Y$. $F_N$ and $p_H$ depend on the choice of the green innovation sector.

When the green innovation sector chooses substantive green innovations,

$$F_N={\displaystyle \frac{p_p}{H}}+\nu ={\displaystyle \frac{p_p}{\lambda CSR{A}_0{K}_H^{\mu }}}+\nu , p_H=p_{p}N{(\lambda CSR{A}_0{K}_H^{\mu })}^{-2}$$

(9)

When the green innovation sector chooses strategic green innovations,

$$F_N={\displaystyle \frac{p_p}{H}}+\nu ={\displaystyle \frac{p_p}{A_0{H}_0}}+\nu ,p_H=p_{p}N{(A_0{H}_0)}^{-2}$$

(10)

3.3.2. Green Innovation Sector Decisions

When the green innovation sector chooses substantive green innovation, it utilizes innovative R&D capital $K_H$ for emissions reduction. The first-order condition for profit maximization is $R_H=p_H{H}_{K_H}$, indicating that the marginal product of capital used for substantive green innovation equals its price. When the green innovation sector chooses strategic green innovation, $K_H=0$.

3.3.3. Financial Institutions Decisions

Assuming financial institutions operate in a perfectly competitive market, the zero-profit condition is ${\pi }_B=(CSR)R_H{K}_H+R_Y{K}_Y-rK=0$. Capital is allocated towards both firm production and the development of emission-reducing technology in the green innovation sector. Based on the level of CSR, financial institutions implement preferential loan interest rates, yielding the following result: $(CSR)R_H=R_Y$. The equation signifies that the actual interest rate $(CSR)R_H$ in the green innovation sector is equal to the interest rate $R_Y$ in the production sector. If the $(CSR)R_H$ and $R_Y$ are not equal, banks will allocate all credit funds to the sector with the higher interest rate, leaving the other sector without capital, resulting in the entire system losing economic significance.

3.3.4. Utilities of Consumers

The optimal consumer problem is:

$$\max U=\max {\displaystyle {\int }_0^{\infty }[\frac{C_t^{1-\sigma }-1}{1-\sigma }+\frac{E_t^{1+\omega }}{1+\omega }]e^{-\rho t}}dt$$

$$\begin{array}{ll}s.t.& \hspace{1em}\stackrel{•}{K}=rK-C+\nu N-p_H\cdot H-p_{p}E\\ & \hspace{1em}\stackrel{•}{E}=\vartheta E(E_0-E)-{\displaystyle \raisebox{1ex}{$N$}\!\left/ \!\raisebox{-1ex}{$H$}\right.}\end{array}$$

The solution to the consumer’s optimal optimization problem is:

$${\displaystyle \frac{\dot{C}}{C}}={\displaystyle \frac{1}{\sigma }}(\gamma -\rho )$$

(11)

When the green innovation sector chooses substantive green innovations,

$${\displaystyle \frac{E^{\omega }}{U_C(\nu \lambda CSR{A}_0{K}_H^{\mu }+{\rho }_p)}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{\nu }}{\nu +p_p/\lambda CSR{A}_0{K}_H^{\mu }}}-{\displaystyle \frac{\stackrel{•}{(\nu \lambda CSR{A}_0{K}_H^{\mu })}}{\lambda CSR{A}_0{K}_H^{\mu }+p_p/\nu }}$$

(12)

When the green innovation sector chooses strategic green innovations,

$${\displaystyle \frac{E^{\omega }}{U_C(\nu A_0{H}_0+{\rho }_p)}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{\nu }}{\nu +\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$A_0{H}_0$}\right.}}-{\displaystyle \frac{\stackrel{•}{(A_0{H}_0)}}{A_0{H}_0+\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$\nu $}\right.}}$$

(13)

3.3.5. Market Equilibrium with Two Categories of Green Innovation

Market equilibrium conditions are determined by the combined market clearance conditions of households, producers, green innovation sectors, and financial sectors.

When the green innovation sector chooses strategic green innovations, the steady state and optimal growth path is:

$${\displaystyle \frac{E^{\omega }}{U_c{F}_N{A}_0{H}_0}}+{\displaystyle \frac{(CSR)\delta N}{\beta \lambda A_0{H}_{0}E}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{F_N}}{N}} \mathrm{and} {\displaystyle \frac{{(1-{\eta }_{CSR})}^{\mu +1}}{{({\eta }_{CSR})}^{1-\alpha }}}={\displaystyle \frac{(CSR)p_p\mu A_0{H}_0}{\alpha A{E}^{\delta }{N}^{\beta -1}{K}^{\alpha +\beta }}}$$

(14)

When the green innovation sector chooses substantive green innovations, the steady state and optimal growth path is:

$${\displaystyle \frac{E^{\omega }}{U_c{F}_N\lambda CSR{A}_0{K}_H^{\mu }}}+{\displaystyle \frac{\delta N}{\beta \lambda CSR{A}_0{K}_H^{\mu }E}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{F_N}}{N}}-{\displaystyle \frac{\mu \lambda A_0{K}_H^{\mu -1}}{\lambda A_0{K}_H^{\mu }}}$$

because $K_H={\eta }_{CSR}K$, we can obtain:

$${\displaystyle \frac{E^{\omega }}{U_c{F}_N\lambda A_{0}CSR{({\eta }_{CSR}K)}^{\mu }}}+{\displaystyle \frac{\delta N}{\beta \lambda A_{0}CSR{({\eta }_{CSR}K)}^{\mu }E}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{F_N}}{N}}-{\displaystyle \frac{\mu }{{\eta }_{CSR}{K}_H}}$$

(15)

$${\displaystyle \frac{{(1-{\eta }_{CSR})}^{\mu +1}}{{({\eta }_{CSR})}^{1-\alpha }}}={\displaystyle \frac{(CSR)p_p\mu \lambda A_0{({\eta }_{CSR}K)}^{\mu }}{\alpha A{E}^{\delta }{N}^{\beta -1}{K}^{\alpha +\beta }}}$$

(16)

Based on the above numerical expression, we can conclude that in the market equilibrium, the allocation of capital is determined by CSR, pollution taxes $p_p$, productivity level $A$, and the capital $K$ and resource consumption $N$ and the allocation of capital ${\eta }_{CSR}$ further influences the choices of firms regarding two types of green innovation. When CSR is high or pollution tax policies are strengthened, ${\eta }_{CSR}$ increases, leading to a shift in the allocation of financial resources towards substantive green innovation.

When ${\eta }_{CSR}={\displaystyle \frac{\beta Y\lambda A_0(CSR)}{p_{p}N}}$, the allocation of capital for substantive green innovation in the green innovation sector can reach an optimal level. Furthermore, ${\eta }_{CSR}={\displaystyle \frac{\beta Y\lambda A_0(CSR)}{p_{p}N}}$ means that the optimal green finance policy for the green innovation sector is not only related to the output level $Y$, and pollution tax $p_p$, but also to the level of CSR. On one hand, the optimal resource allocation proportion in the green innovation sector also increases when CSR or output level increases under the premise of stable resource consumption. On the other hand, the optimal resource allocation proportion in the green innovation sector decreases when the pollution tax increases. In other words, the optimal resource allocation proportion in the green innovation sector exhibits two characteristics. Firstly, it increases with the expansion of CSR and economic scale. Secondly, it complements the pollution tax policy. It means that R&D capital allocation in the green innovation sector will become more critical at an advanced stage of economic development. In addition, green innovation policies must be coordinated with other environmental policies when formulated. Green innovation policies should be weakened when environmental policies are strengthened, and vice versa.

3.4. The Combination of Economic Growth Path, CSR, and Environmental Policy

The optimal level of capital allocation by financial institutions for substantive green innovation is dependent upon the amount of capital allocated to the innovation sector. In addition to the level of CSR, the optimal level is also influenced by the stage of economic development and other environmental policies. We can further compare the economic growth trajectory at market equilibrium with the social planner’s problem:

When the green innovation sector chooses strategic green innovations, we have the following economic growth trajectory:

$${\displaystyle \frac{{(1-{\eta }_{CSR})}^{\mu +1}}{{({\eta }_{CSR})}^{1-a}}}={\displaystyle \frac{(CSR)p_p\mu A_0{H}_0}{aA{E}^{\delta }{N}^{\beta -1}{K}^{a+\beta }}}$$

(17)

$${\displaystyle \frac{\stackrel{•}{C}}{C}}={\displaystyle \frac{1}{\sigma }}(F_{K_Y}-\rho )$$

(18)

$${\displaystyle \frac{E^{\omega }}{U_C(A_0{H}_0+p_p)}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{\nu }}{\nu +\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$A_0{H}_0$}\right.}}-{\displaystyle \frac{\stackrel{•}{(A_0{H}_0)}}{A_0{H}_0+\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$\nu $}\right.}}$$

(19)

When the green innovation sector chooses substantive green innovations, we have the following economic growth trajectory:

$${\displaystyle \frac{{(1-{\eta }_{CSR})}^{\mu +1}}{{({\eta }_{CSR})}^{1-\alpha }}}={\displaystyle \frac{(CSR)p_p\mu \lambda A_0{({\eta }_{CSR}K)}^{\mu }}{\alpha A{E}^{\delta }{N}^{\beta -1}{K}^{\alpha +\beta }}}$$

(20)

$${\displaystyle \frac{\stackrel{•}{C}}{C}}={\displaystyle \frac{1}{\sigma }}(F_{K_Y}-\rho )$$

(21)

$${\displaystyle \frac{E^{\omega }}{U_C[v\lambda A_{0}CSR{({\eta }_{CSR}K)}^{\mu }+p_p]}}+\vartheta (E_0-2E)=\rho +\sigma {\displaystyle \frac{\stackrel{•}{C}}{C}}-{\displaystyle \frac{\stackrel{•}{\nu }}{\nu +\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$v\lambda A_{0}CSR{({\eta }_{CSR}K)}^{\mu }$}\right.}}-{\displaystyle \frac{\stackrel{•}{[v\lambda A_0{({\eta }_{CSR}K)}^{\mu }}]}{\lambda A_{0}CSR{({\eta }_{CSR}K)}^{\mu }+\raisebox{1ex}{$p_p$}\!\left/ \!\raisebox{-1ex}{$\nu $}\right.}}$$

(22)

Equations (19) and (22) represent the environmental equation of motion on the path of economic growth. By substituting the production function (2) and environmental movement Equation (3) into (19) and (22), we can obtain:

When the green innovation sector chooses strategic green innovations,

$${\displaystyle \frac{\stackrel{•}{\Gamma }}{\Gamma }}={\displaystyle \frac{\stackrel{•}{P}}{P}}-{\displaystyle \frac{\stackrel{•}{Y}}{Y}}={\displaystyle \frac{E^{\omega }}{U_C{A}_0{H}_0{F}_N}}+\vartheta (E_0-2E)(A_0{H}_0{F}_N+p_p)-F_{K_Y}(A_0{H}_0{F}_N+p_p)$$

(23)

When the green innovation sector chooses substantive green innovations,

$${\displaystyle \frac{\stackrel{•}{\Gamma }}{\Gamma }}={\displaystyle \frac{\stackrel{•}{P}}{P}}-{\displaystyle \frac{\stackrel{•}{Y}}{Y}}={\displaystyle \frac{E^{\omega }}{U_C\lambda CSR{A}_0{({\eta }_{CSR}K)}^{\mu }{F}_N}}+\vartheta (E_0-2E)[\lambda CSR{A}_0{({\eta }_{CSR}K)}^{\mu }{F}_N+p_p]-F_{K_Y}[\lambda CSR{A}_0{({\eta }_{CSR}K)}^{\mu }{F}_N+p_p]$$

(24)

where $\Gamma ={\displaystyle \frac{P}{Y}}$ represents the environmental pollution per unit of output. In much literature on green sustainable development, it is called the environmental cost of economic growth and is an important indicator of the quality of economic growth. Equations (23) and (24) describe how substantive green innovation and strategic green innovation affect the quality of economic growth along the path of economic growth.

Taking the derivative of Equation (24) with respect to CSR, we obtain:

$${\displaystyle \frac{\partial \Gamma }{\partial CSR}}=[-{\displaystyle \frac{1}{{(CSR)}^2}}]{\displaystyle \frac{E^{\omega }}{U_C\lambda A_0{({\eta }_{CSR}K)}^{\mu }{F}_N}}-\vartheta (E_0-2E)\lambda A_0{({\eta }_{CSR}K)}^{\mu }{F}_N-F_{K_Y}\lambda A_0{({\eta }_{CSR}K)}^{\mu }{F}_N<0$$

(25)

Proposition 1:

Under competitive equilibrium, CSR can reduce environmental pollution per unit of output (improve economic growth quality).

Considering substantive green innovation $H=\lambda CSR{A}_0{K}_H^{\mu }$, we have

$$\Gamma ={\displaystyle \frac{E^{\omega }}{U_{C}H{\eta }_{CSR}^{\mu }{F}_N}}-\vartheta (E_0-2E)[H{\eta }_{CSR}^{\mu }{F}_N+p_p]-F_{K_Y}[H{\eta }_{CSR}^{\mu }{F}_N+p_p]$$

(26)

Taking the derivative of Equation (26) with respect to $H$, we obtain

$${\displaystyle \frac{\partial \Gamma }{\partial H}}=(-{\displaystyle \frac{1}{H^2}}){\displaystyle \frac{E^{\omega }}{U_C{\eta }_{CSR}^{\mu }{F}_N}}-\vartheta (E_0-2E){\eta }_{CSR}^{\mu }{F}_N-F_{K_Y}{\eta }_{CSR}^{\mu }{F}_N<0$$

(27)

Proposition 2:

Under competitive equilibrium, CSR prompts the green innovation sector to adopt substantive green innovation to improve economic growth quality.

Taking the derivative of Equation (24) and considering ${\eta }_{CSR}$, we obtain

$${\displaystyle \frac{\partial \Gamma }{\partial {\eta }_{CSR}}}=-\mu {{\eta }_{CSR}}^{-\mu -1}{\displaystyle \frac{E^{\omega }}{U_C\lambda CSR{A}_0{K}^{\mu }{F}_N}}-\vartheta \mu {{\eta }_{CSR}}^{\mu -1}(E_0-2E)\lambda CSR{A}_0{K}^{\mu }{F}_N-F_{K_Y}\mu {{\eta }_{CSR}}^{\mu -1}\lambda CSR{A}_0{K}^{\mu }{F}_N<0$$

(28)

Proposition 3:

Under competitive equilibrium, by implementing preferential credit policies through financial institutions, firms with higher CSR levels can further strengthen the effect of CSR on improving the quality of economic growth.

Does empirical evidence support the theoretical model of this paper? How do substantive and strategic green innovations affect economic growth? How does CSR affect the capital allocation of financial institutions to further leverage the effect of high-quality economic growth? This study uses data at the provincial level in China from 2013 to 2022 to test the theoretical model.

4. Empirical Design

4.1. Data and Sample Selection

This study uses panel data from 30 provinces (i.e., Taiwan, Hong Kong, Macau, and Tibet were not included due to a lack of data) for the period 2013–2022 to examine the relationship between CSR and economic growth quality in China. Our sample data is sourced from two main components: provincial-level data obtained from the China Provincial and Municipal Statistical Yearbook and China National Intellectual Property Administration (CNIPA), and firm-level data retrieved from the CSMAR database and the Wind database. Additionally, CSR data is collected from www.hexun.com (accessed on 1 July 2023). These databases meticulously detail indicator values across provinces, firms, and annual periods. Through further analysis and organization of the data, we can accurately compile the dataset for all variables in the study, facilitating empirical analysis. Therefore, our data sources are reliable.

The purpose of our study is to conduct a mediation analysis using green innovation and to test for the presence of moderating effects through green finance. Using stock concept labels, we categorize Chinese A-share listed companies and filter out companies in the energy conservation and environmental protection sectors. In particular, we refer to the concept labels provided by Wind and select companies associated with the following 12 labels: “Renewable Energy, Clean Technology, Carbon Capture and Storage, Green Building, Wind Power Generation, Electric Vehicles, Charging Stations, Sustainable Agriculture, Energy Conservation and Environmental Protection, Circular Economy, Exhaust Gas Treatment, Solar Photovoltaics”. Companies listed with the above 12 labels are considered to be the energy conservation and environmental protection industries at the provincial level.

4.2. Variables Definitions

4.2.1. Dependent Variable

The dependent variable is economic growth quality. The quality of economic growth encompasses multiple dimensions; however, this paper specifically focuses on the green development dimension. Existing research primarily utilizes three types of indicators to measure green development: environmental pollution [51], environmental pollution per unit of output, and natural resource consumption per unit of output [52]. Our study does not use environmental pollution indicators due to the multitude of environmental pollution indicators, the significant heterogeneity among different indicators, and the lack of a recognized and reliable way to aggregate all pollutants. The theoretical model indicates that environmental damage resulting from resource consumption is a core variable of economic development. Therefore, considering the data, the existing literature, and the theoretical model presented in this paper, energy consumption per unit of GDP has been chosen as the indicator to evaluate the quality of economic growth. Energy consumption indicators possess a more standardized measurement criterion than environmental pollution indicators. Energy consumption per unit of GDP is measured by the National Bureau of Statistics, which adds up the heat values of eight major fuels (including coal, natural gas, coke, gasoline, diesel, crude oil, kerosene, and fuel oil) and divides the total by the regional GDP, as dictated by ENERGYPD.

4.2.2. Key Explanatory Variables

The key explanatory variable is Corporate Social Responsibility (CSR). The measure of CSR fulfillment is based on the comprehensive CSR scores published by www.hexun.com (accessed on 1 July 2023). The data for this assessment system comes from social responsibility reports and annual financial reports published on the websites of Chinese-listed companies. The framework consists of 13 s-level indicators and 37 third-level indicators organized around five dimensions: shareholder responsibility, employee responsibility, supplier and customer rights and interests, environmental responsibility, and public responsibility. Based on this comprehensive framework, a company’s social responsibility performance can be objectively assessed [11].

By categorizing listed companies into energy conservation and environmental protection firms and separating them by province, the assessment of corporate social responsibility fulfillment for each province can be conducted as follows:

$$CS{R}_{it}={\displaystyle \frac{{\displaystyle \sum _{j=1}^{n}CS{R}_{ijt}}}{n}}$$

where $t$ indicates year, $i$ indicates the province, $j$ indicates the screened listed companies in the province, ${\displaystyle \sum _{j=1}^{n}CS{R}_{ijt}}$ indicates the total CSR score of the screened listed companies in the province $i$. We use the total CSR score in the province plus 1. This makes the logarithm the proxy variable of CSR, as suggested by INCSR.

In order to assess robustness, referring to Ding et al. [11], our study uses the disclosed comprehensive CSR evaluation scores from the RKS ESG serve as a proxy variable of CSR, denoted by INRCSR.

4.2.3. Mediating and Moderating Variables

In our theoretical model, CSR promotes substantive green innovation in the innovation sector to reduce environmental pollution per unit of output, thus green innovation serves as a mediating variable. According to the existing literature [14,53,54,55], green innovation is usually measured by the number of green patents granted by firms. We conducted patent searches by the province in the CNIPA’s “Patent Retrieval and Analysis” database using keywords related to “low-carbon”, “emission reduction”, and “energy-saving”. In our study, green innovation is categorized into substantive green innovation (GIP) and strategic green innovation (GUP) based on motivational factors. According to the literature [13,15], since invention patents have a high technological content are difficult to obtain, and are generally high-level technical innovations, this study uses the number of green invention patents plus 1 to take the logarithm as the proxy variable of GIP, denoted by INGIP. Green utility patents have low technological content and are less difficult to obtain. Therefore, this study uses the number of green utility patents plus 1. This makes the logarithm a proxy variable of GUP, as dictated by INGUP.

The moderating variable is green finance, denoted by GF. In accordance with the literature [53,56], the measure of green finance involves dividing the total borrowing of energy-saving and environmental protection firms by the total borrowing of all listed companies. Our paper selects the period from 2013 to 2022, as there is a delay in the availability of statistical data for environmental variables. All firms are classified according to whether they fall into the low-carbon and environmental protection sectors, and then they are summed up by province in the year $t$ to represent the green finance indicator for that province. In the theoretical model, the core variable ${\eta }_{CSR}$ means the percentage of capital invested in the green innovation sector with substantive green innovation R&D investment, thus this indicator is consistent with the theoretical model.

4.2.4. Control Variables

The control variables were selected with reference to the literature on environmental economics to control for other factors that might affect economic growth [34,54,56]. Our study proposed eight control variables: INTEREST, LNGDP, IND, LNINVEST, FIN, LNPGDP, EDU, and OPEN. Considering financing costs are vital for influencing corporate green innovation, it is important to control the average borrowing interest rate of eco-friendly firms. INTEREST is calculated by dividing the current year’s interest expenses by the previous year’s total borrowings. LNGDP (the natural logarithm of GDP), IND (the share of the secondary industry in GDP at the provincial level), LNINVEST (the natural logarithm of government investment in pollution control), FIN (the ratio of annual bank loans to provincial GDP as a proxy indicator for financial development), LNPGDP (the natural logarithm of GDP per capita). EDU is controlled using the ratio of provincial education expenditure to provincial GDP as a proxy indicator. OPEN (a province’s total exports and imports of goods divided by GDP).

4.3. Empirical Specification

According to our theoretical model, CSR reduces environmental pollution per unit of output (i.e., improve economic growth quality) by supporting the green innovation sector to conduct substantive green innovation. By implementing preferential credit interest rates, financial institutions can enhance the effect of CSR on reducing environmental pollution. Our study conducted the following empirical specification based on these findings.

To assess the impact of CSR on environmental pollution per unit of output (Proposition 1), the following baseline empirical model is used:

$$Energyp{d}_{it}={\beta }_0+{\beta }_{1}CS{R}_{it}+{\beta }_m{X}_{it}+Yea{r}_t+Provinc{e}_i+{\epsilon }_i$$

(29)

where $Energyp{d}_{it}$ represents the environmental pollution per unit of output of the province $i$ in year $t$, consistent with $\Gamma$ in the theoretical model. $CS{R}_{it}$ represents the level of corporate social responsibility fulfillment by province $i$ in year $t$, $X_{it}$ denotes the control variables. $Yea{r}_t$ and $Provinc{e}_i$ denote year and province fixed effect, ${\epsilon }_i$ is the residual term. If the coefficient ${\beta }_1$ is significantly negative, it indicates that CSR contributes to reducing the environmental pollution per unit of output, thereby enhancing the quality of economic growth.

To evaluate the mediation mechanism effect of green innovation (Proposition 2), the following mediation mechanism empirical model is developed:

$$Energyp{d}_{it}={\beta }_0^A+{\beta }_1^{A}CS{R}_{it}+{\beta }_m^A{X}_{it}+Yea{r}_t+Provinc{e}_i+{\epsilon }_i$$

(30)

$$M_{it}={\beta }_0^B+{\beta }_1^{B}CS{R}_{it}+{\beta }_m^B{X}_{it}+Yea{r}_t+Provinc{e}_i+{\epsilon }_i$$

(31)

$$Energyp{d}_{it}={\beta }_0^C+{\beta }_1^{C}CS{R}_{it}+{\beta }_2^C{M}_{it}+{\beta }_m^C{X}_{it}+Yea{r}_t+Provinc{e}_i+{\epsilon }_i$$

(32)

where $M_{it}$ represents substantive green innovation (GIP) and strategic green innovation (GUP). Specifically, Equation (30) is the baseline regression model. Equation (31) is conducted to analyze the effects of CSR on the mediating variables. A mediating effect is further tested by Equation (32) if the impact is significant.

To assess the moderation mechanism effect of green finance (Proposition 3), the following moderation mechanism empirical model is developed:

$$Energyp{d}_{it}={\beta }_0^D+{\beta }_1^{D}CS{R}_{it}+{\beta }_2^{D}G{F}_{it}+{\beta }_3^{D}G{F}_{it}\cdot CS{R}_{it}+{\beta }_m^C{X}_{it}+Yea{r}_t+Provinc{e}_i+{\epsilon }_i$$

(33)

where $G{F}_{it}$ represents the green finance indicator of the province $i$ in year $t$, consistent with the financial institution credit incentives ${\eta }_{CSR}$ in the theoretical model. If ${\beta }_1^D<0,{\beta }_3^D<0$ holds simultaneously, it indicates that credit preferences of financial institutions play a positive moderating role in reducing environmental pollution per unit of output through CSR.

5. Empirical Results and Analysis

5.1. Descriptive Statistics

Table 1. Descriptive statistics.

Variable	N	Mean	Min	Max	SD
ENERGYPD	300	0.871	0.208	2.342	0.499
LNCSR	300	9.696	7.042	11.590	0.893
LNGIP	300	5.435	1.946	8.006	1.324
LNGUP	300	3.084	0.000	7.029	1.490
GF	300	0.416	0.016	0.717	0.151
LNGDP	300	3.067	2.328	3.704	0.206
LNINVEST	300	4.727	2.701	6.617	0.633
INTEREST	300	0.017	0.000	0.100	0.009
IND	300	0.444	0.162	0.590	0.087
LNPGDP	300	9.426	7.214	10.631	0.673
EDU	300	0.041	0.021	0.109	0.015
OPEN	300	0.287	0.013	1.464	0.309
FIN	300	1.419	0.671	2.577	0.458

shows the descriptive statistics of the variables. The table shows that the mean value of ENERGYPD is 0.871, the maximum value is 2.342, and the minimum value is 0.208. These data show that there are large differences in the level of energy consumption per unit of GDP between different provinces. The mean value of the CSR (LNCSR) is 9.696, the minimum value is 7.042, and the maximum value is 11.590, indicating that the overall CSR performance of listed companies in different provinces is good, but some firms do not pay much attention to practicing CSR, and the CSR performance of different sample provinces varies greatly.

Correlation analysis is a statistical method employed to examine the degree and direction of relationships between two or more variables. Therefore, we further analyzed the Pearson correlation coefficients between the variables in

Table 2. Correlation analysis and multicollinearity test results.

Variable	ENERGYPD	LNGDP	LNINVEST	INTEREST	IND	LNPGDP	EDU	OPEN	FIN
ENERGYPD	1
LNGDP	−0.070	1
LNINVEST	−0.058	−0.069	1
INTEREST	0.063	0.049	−0.094	1
IND	−0.129	−0.042	0.162	0.053	1
LNPGDP	−0.299	−0.107	0.079	−0.091	0.023	1
EDU	−0.384	−0.206	0.021	−0.070	0.109	0.015	1
OPEN	0.119	0.287	0.013	0.024	0.057	0.009	0.013	1
FIN	0.335	0.082	−0.125	−0.080	0.110	0.058	0.016	−0.026	1
VIF	2.381	3.190	2.007	1.890	1.312	2.036	1.523	1.952	1.311

. The correlation coefficients between the variables are all below 0.50, and the average variance inflation factor (VIF) for the variables in models (29) to (33) is below the empirical value of 2, with the maximum VIF in all models being 3.190, far below the empirical value of 10. Therefore, there is no severe multicollinearity problem in the regressions of the models.

5.2. Baseline Regression Analysis

Table 3. Baseline regression analysis.

	(1)	(2)	(3)
	ENERGYPD	ENERGYPD	ENERGYPD
LNCSR	−0.391 *** (0.028)	−0.442 *** (0.130)	−0.794 *** (0.152)
INTEREST			−0.963 ** (0.482)
LNGDP			0.075 (0.053)
LNPGDP			0.807 *** (0.066)
IND			0.496 * (0.254)
LNINVEST			0.041 ** (0.021)
FIN			−0.155 *** (0.043)
EDU			7.327 *** (1.663)
OPEN			−0.243 *** (0.054)
Constant	4.661 *** (0.284)	5.159 *** (1.266)	0.315 (1.203)
Year FE	NO	YES	YES
Province FE	NO	YES	YES
N	300	300	300
Adj. R2	0.489	0.965	0.986

Note: The values in parentheses are robust standard errors. ***, **, and * represent significance at the 1%, 5%, and 10% levels, respectively.

provides a baseline regression for the impact of CSR on environmental pollution per unit of output. In column (1), neither control variables nor time province fixed effects are present. Column (2) controls for time and province fixed effects but does not include control variables. Column (3) contains both control variables and time-province fixed effects. Based on the regression results in columns (1)–(3), we find that the regression coefficient of LNCSR is negative and significant at the 1% level, suggesting that CSR contributes to reducing environmental pollution per unit of output, thereby improving the quality of economic growth. Thus, Proposition 1 is validated. The quality of economic growth is positively influenced by CSR, which enhances efficiency, fosters innovation, strengthens reputation, reduces risk, and attracts talent and investments. The awareness of CSR and active fulfillment of social responsibility encourage resource conservation, environmental protection, and social welfare, as well as enhance economic growth. The results in our study are consistent with previous research conducted by Yousefian et al. [37] and Inekwe et al. [57], which concluded that good CSR performance promotes green economic growth.

5.3. Robustness and Endogeneity Analysis

In this section, we analyze the robustness and endogeneity of the empirical study.

5.3.1. Robustness Test

Our study uses three methods to test the robustness of the empirical model. (1) Change the independent variables. We use the CSR score disclosed by RKS ESG (denoted by RCSR) to replace the CSR score disclosed by www.hexun.com (accessed on 1 July 2023). Subsequently, we examine the impact of RCSR on environmental pollution per unit of output based on the model (29), and the results are shown in column (1) of

Table 4. Robustness test regression results.

	(1)	(2)	(3)
	ENERGYPD	COALPD	Lag ENERGYPD
LNRCSR	−0.587 *** (0.141)
LNCSR		−0.790 *** (0.154)	−0.778 *** (0.174)
Controls	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
N	300	300	270
Adj. R2	0.983	0.984	0.983

Note: The values in parentheses are robust standard errors. *** represents significance at the 1% level.

. (2) Change the dependent variable. By adopting unit GDP coal consumption (COALPD) to replace the original dependent variable ENERGYPD, we estimate the impact of CSR on unit output environmental pollution, the results are shown in column (2) of Table 4. (3) Lagged variables. Lagging the dependent variables by one period (denoted by Lag ENERGYPD), the regression results are presented in column (3) of Table 4. Based on the regression results of the three robustness tests, CSR is highly effective at reducing environmental pollution per unit of output, consistent with the benchmark regression model.

5.3.2. Endogeneity Test

To address endogeneity issues, we use the Two-Stage Least Squares (2SLS) method to address factors like measurement errors, omitted variables, and mutual causation in empirical studies. Firstly, in the baseline regression model (29), the Dubin Wu-Hausman test is employed to test for endogeneity, and the result rejects the original exogenous hypothesis, indicating that CSR and environmental pollution per unit of output are endogenous. Secondly, the instrumental variables (IV) method is the most direct approach to addressing endogeneity. Considering the lagged effect of the previous period’s CSR performance on the environmental pollution of current unit output, where the impact of current CSR activities may not directly affect current environmental pollution, we employ a panel instrumental variable approach.

Drawing on the work of Zhang et al. [58], we use the first-order lag, second-order lag, and third-order lag of the CSR variable as instrumental variables to address endogeneity issues. The lagged values of the CSR variable (first-order, second-order, and third-order lags) are highly correlated with the current CSR value as past CSR activities influence current CSR decisions, thus fulfilling the relevance criterion for instrumental variables. Although these lagged values are correlated with the current CSR, they are past values and are not affected by the current error term. Since CSR activities indirectly influence current environmental pollution through current CSR actions, they satisfy the exogeneity condition for instrumental variables. The regression results in

Table 5. Instrumental Variable Regression Results.

	(1)	(2)	(3)
IV	First-Order Lag	Second-Order Lag	Third-Order Lag
	ENERGYPD	ENERGYPD	ENERGYPD
LNCSR	−0.884 *** (0.035)	−0.887 *** (0.038)	−0.892*** (0.040)
Constant	1.716 *** (0.308)	1.927 *** (0.327)	2.177 *** (0.318)
Controls	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
N	270	240	210
Adj. R2	0.934	0.936	0.938

Note: The values in parentheses are robust standard errors. *** represents significance at the 1% level.

show that the coefficients for CSR are −0.884, −0.887, and −0.892, all of which are statistically significant. These findings show that the conclusions drawn from the instrumental variable regression are consistent with the main regression.

5.4. Mechanism Analysis

5.4.1. Influencing Channels Test Results

To examine how CSR impacts environmental pollution per unit of output, we examine the mediation effect using models (30)–(32).

Table 6. Mediation effect regression results.

	(1)	(2)	(3)	(4)	(5)
	ENERGYPD	LNGIP	ENERGYPD	LNGUP	ENERGYPD
LNCSR	−0.794 *** (0.152)	1.031 *** (0.168)	−1.061 *** (0.171)	0.727 (0.785)	−0.815 *** (0.147)
LNGIP			−0.260 *** (0.055)
LNGUP					0.029 *** (0.009)
Constant	0.314 (1.203)	−7.148 *** (1.537)	2.171 * (1.259)	−13.687 (8.650)	0.708 (1.178)
Year FE	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES
N	300	300	300	300	300
Adj. R2	0.984	0.996	0.986	0.899	0.985

Note: The values in parentheses are robust standard errors. *** and * represent significance at the 1% and 10% levels, respectively.

presents the results of the influence channel tests. The results of the baseline regression estimation are presented in column (1). Green invention patents (GIP) are used as a proxy variable for substantive green innovation in columns (2) and (3). Green utility patents (GUP) are used as a proxy variable for strategic green innovation in columns (4) and (5). First, the results in column (1) of Table 6 demonstrate that CSR significantly reduces environmental pollution per unit of output. As shown in column (2), substantive green innovation (LNGIP) is significantly positively correlated with LNCSR, showing that a greater level of CSR can lead to more substantive green innovations (GIP). Second, in column (3) of Table 6, the coefficients for substantive green innovation (LNGIP) and LNCSR are both significantly negative. In conclusion, a comprehensive comparison of the results shown in columns (1)–(3), demonstrates that by supporting substantive green innovation in the green innovation sector, CSR reduces environmental pollution per unit of output, and improves economic growth quality.

Nevertheless, the coefficient of LNCSR in column (4) of Table 6 is not significant, which implies that CSR has no significant impact on the number of GUP. We can conclude that CSR does not affect the quality of economic growth by influencing the progress of strategic green innovation (GUP). In conclusion, our study finds that CSR mainly contributes to improving economic growth quality by promoting substantive green innovation (GIP). This study reveals that the test results of the mediating mechanism are consistent with our theoretical analysis. These findings highlight the importance of distinguishing between different types of green innovation when evaluating the impact of CSR on economic growth.

To verify the robustness of the results concerning the mediating effect of substantive green innovation, we also carried out the Sobel test, and the test result is $z=2.635,p<0.01$. According to this result, substantive green innovation serves as a mediator, consistent with the outcomes of the preceding stepwise regression analysis. In addition, the mechanism test is also analyzed using the R&D input from emission reduction technologies (denoted by LNTECH) as an alternative proxy variable of substantive green innovation. As shown in

Table 7. Substitution of mediator variables.

	(1)	(2)	(3)
	ENERGYPD	LNTECL	ENERGYPD
LNCSR	−0.794 *** (0.152)	1.025 *** (0.167)	−1.063 *** (0.171)
LNTECL			−0.263 *** (0.057)
Constant	0.315 (1.203)	−7.067 *** (1.536)	2.170 * (1.260)
Controls	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
N	300	300	300
Adj. R2	0.983	0.996	0.984

Note: The values in parentheses are robust standard errors. ***, and * represent significance at the 1% and 10% levels, respectively.

, the estimated results are consistent with the GIP indicator and demonstrate the robustness of the proxy indicator for substantive green innovation. Thus, Proposition 2 is validated.

5.4.2. The Results of Green Finance Moderating Effect

To further test Proposition 3 in the theoretical model, this empirical design is based on the empirical model (33).

Table 8. Moderation effect regression results.

	(1)	(2)	(3)	(4)	(5)
	Full Sample	Low Level	Med-Low Level	Med-High Level	High Level
LNCSR	−0.454 *** (0.126)	0.262 (0.383)	−0.157 (0.350)	−1.641 *** (0.298)	−0.452 *** (0.133)
LNCSR·GF	−0.583 *** (0.099)	−1.817 *** (0.473)	−1.667 *** (0.522)	−0.780 *** (0.146)	−0.757 *** (0.227)
Controls	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES
Province FE	YES	YES	YES	YES	YES
N	300	73	73	74	72
Adj. R2	0.988	0.987	0.996	0.997	0.998

Note: The values in parentheses are robust standard errors. *** represents significance at the 1% level.

shows the results of the moderating effect. Based on the result of the full sample in column (1) of Table 8, the coefficient of the interaction term between CSR and green finance is −0.583, significant at the 1% level. Therefore, the credit preference of financial institutions (green finance), plays a positive moderating role in CSR and the quality of economic growth. Thus, Proposition 3 is validated.

A grouping regression based on the quartiles of green finance levels was conducted to investigate the differentiated credit preferences of financial institutions (green finance indicators). The specific grouping procedure entails arranging the entire sample in ascending order of green finance level. The first 25% is categorized as the low-level group, the next 25–50% as the med-low-level group, the subsequent 50–75% as the med-high-level group, and the last 25% as the high-level group. The results of the grouped regression are shown in columns (2)–(5) of Table 8. In the grouped regression analysis, it was observed that after introducing the interaction term between CSR and green finance, the effect of CSR on enhancing environmental quality remained significant for both the med-high level and high-level green finance groups. Additionally, within these groups, green finance exhibited a notably significant positive moderating effect (coefficients of −0.780 with p < 0.01 and −0.757 with p < 0.01). Furthermore, the grouped regression results suggest that lower levels of green finance may have difficulty enhancing the positive moderating effect of CSR. Therefore, it is imperative for financial institutions to continuously raise awareness of environmental responsibility and facilitate the growth of green finance.

Finally, further robustness tests of the mediating and moderating effects are carried out using the CSR score provided by RKS ESG (denoted by RCSR) as a proxy variable of CSR, and the regression results are presented in

Table 9. Tests of mediating and moderating mechanisms for substitution variables.

	(1)	(2)	(3)	(4)
	ENERGYPD	LNGIP	ENERGYPD	ENERGYPD
LNRCSR	−0.587 *** (0.141)	0.798 *** (0.181)	−0.759 *** (0.181)	−0.270 ** (0.111)
LNGIP			0.216 *** (0.059)
LNRCSR·GF				−0.607 *** (0.102)
Constant	−1.488 (1.128)	−5.080 *** (1.694)	−0.392 (1.375)	−4.329 *** (1.006)
Controls	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
Province FE	YES	YES	YES	YES
N	300	300	300	300
Adj. R2	0.983	0.995	0.805	0.878

Note: The values in parentheses are robust standard errors. ***, and ** represent significance at the 1% and 5% levels, respectively.

. Based on the results in columns (1)–(3) of Table 9, we can find substantive green innovation still mediates between CSR and the reduction of environmental pollution per unit of output. According to column (4) of Table 9, green finance still plays a positive moderating role in the reduction of environmental pollution per unit of output through CSR. After replacing the explanatory variable, the regression results do not change significantly, indicating the robustness of the empirical results.

5.5. Heterogeneity Analysis

To further investigate the regional heterogeneity of the impact of CSR on the quality of economic growth (environmental pollution per unit of output), we divided 30 provinces in China into three geographical regions: Eastern, Central, and Western. The regression results are shown in

Table 10. Heterogeneity analysis.

	(1)	(2)	(3)
	Eastern	Central	Western
	ENERGYPD	ENERGYPD	ENERGYPD
LNCSR	−0.312 *** (0.071)	−0.278 *** (0.034)	−0.124 (0.094)
Constant	3.777 *** (0.921)	−3.202 ** (1.298)	2.368 *** (0.403)
Controls	YES	YES	YES
Year FE	YES	YES	YES
Province FE	YES	YES	YES
N	110	80	110
Adj. R2	0.878	0.805	0.836

Note: The values in parentheses are robust standard errors. ***, and ** represent significance at the 1% and 5% levels, respectively.

. The regression coefficients of CSR on environmental pollution per unit of output are −0.312, −0.278, and −0.124, respectively. Only the Eastern and Central regions are significant at the 1% level, with the regression coefficient in the Eastern region higher than that in the Central region, indicating that CSR has a stronger positive effect on the quality of economic growth in the Eastern region than in the Central region. The Western region did not pass the significance test. The regression results indicate significant regional differences in the impact of CSR on the quality of economic growth.

The reason may be that the Eastern region typically possesses more developed economic and social infrastructure, as well as higher levels of environmental governance, regulatory enforcement, and policy support. These factors can enhance CSR investments and practices in CSR, improve production efficiency, and foster innovation capabilities, thereby promoting the quality of economic growth. In contrast, the Western region, despite its abundant natural resources, faces challenges in balancing resource exploitation and environmental protection, leading to uneven resource allocation and insufficient CSR empowerment. Additionally, the Western region may lag behind the Eastern region in infrastructure development and technological innovation, resulting in relatively lower CSR investments and practices in CSR, and consequently, it is difficult to achieve the desired effect of enhancing the quality of economic growth.

6. Conclusions and Policy Implications

6.1. Conclusions

In the new development stage, accelerating the implementation of green development concepts to drive firms to achieve high-quality development has become an important research topic in the academic community. This paper constructs a general equilibrium model that integrates the impacts of CSR on substantive green innovation and strategic green innovation, exploring the changes in emission reduction technologies from firms fulfilling CSR and analyzing the impact on economic growth quality (i.e., environmental pollution per unit of output). The theoretical model not only illustrates the role of CSR in economic development through a comparison of economic growth paths but also captures the relationship between CSR and the quality of economic development. To empirically test our theoretical model, we use panel data from 30 provinces in China from 2013–2022. The results show that CSR has a significant negative effect on environmental pollution per unit of output. The robustness of our conclusions has been tested using a variety of methods. Based on the theoretical model and empirical analysis, we find that substantive green innovation acts as a transmission mechanism through which CSR promotes economic growth quality. Meanwhile, the credit preferences of financial institutions play a positive moderating role in CSR and economic growth quality. The theoretical model considers both pollution tax policy and green finance policy that incorporate CSR. These findings are valuable in guiding firms toward fulfilling their social responsibilities and improving their substantive green innovation capabilities.

Our research highlights the importance of CSR in driving substantive green innovation within firms, fostering a low-carbon industrial transition, and promoting affordable clean energy solutions. This advancement is crucial for sustainable societal development and aligns with the sub-goals of the United Nations Sustainable Development Goals (SDGs), highlighting its global relevance and impact. Compared to existing literature on CSR and economic growth at national or industry levels [37,38,59], this paper confirms CSR’s role in improving economic growth quality, consistent with prior studies. Additionally, it uniquely explores how CSR, particularly through green innovation—a category not previously addressed in the literature—impacts economic growth mechanisms. These findings provide valuable insights for policymakers and businesses aiming to integrate CSR into their strategies for sustainable development.

6.2. Policy Implications

In the context of sustainable development, our research has significant implications.

Firstly, at the firm level, firms need to actively embrace their social responsibilities by developing and implementing strategies to conserve energy and reduce emissions, which serves as a microcosm of national economic development. This involves not only creating comprehensive plans for green innovation but also implementing practical initiatives within their innovation and research processes. To effectively promote green innovation and capitalize on opportunities for sustainable development, firms should prioritize integrating CSR principles alongside green innovation as key initiatives that harmonize economic goals with social responsibility. This approach necessitates embedding CSR into core business strategies, allocating resources to green innovation initiatives, and fostering a corporate culture that emphasizes sustainability practices. Managers, as firm decision-makers and strategists should integrate CSR principles into the firm’s long-term strategic planning. They should prioritize allocating resources to support green technologies and sustainable development projects, establish and maintain effective monitoring systems, and regularly evaluate the effectiveness of CSR initiatives. This approach is essential for achieving sustainable development and enhancing the company’s growth prospects. In addition, firms implementing these policies may face initial high costs of green technologies and resistance to change from traditional business practices. Thus, firms should conduct strategic investment planning to manage costs effectively and provide employee training to facilitate adaptation to sustainable practices.

Secondly, at the governmental level, governments should enhance the development of practical policy systems that enable firms to effectively fulfill their social responsibilities. This includes establishing supportive regulatory environments, implementing incentives such as tax benefits and subsidies for sustainable practices, fostering collaboration among governmental agencies, businesses, and civil society through inclusive policy-making processes and resource sharing, and enhancing monitoring and evaluation mechanisms to ensure adherence to social responsibility standards and advance sustainable development goals. The government may face challenges in implementing these policies due to insufficient financial resources to support incentive measures, which can be addressed by optimizing fiscal budget allocation and enhancing fiscal transparency to ensure the effective implementation of these incentives. Additionally, the government should consider developing long-term strategic plans to ensure the sustained effectiveness and adaptability of these policies in addressing future challenges.

Finally, at the financial institution level, our study also demonstrates that the credit preferences of financial institutions play a positive moderating role in CSR and economic growth quality. Financial institutions identify substantive and strategic green innovation behaviors in innovation sectors by examining the level of CSR fulfillment by firms. They increase substantive R&D investment in innovation sectors, strengthening CSR’s effect on improving environmental quality. Hence, the provision of financial services is crucial for firms that upgrade their green innovation technologies to comply with CSR. Actively promoting the construction of green finance will contribute to a sustainable economy. However, green finance regulators need to strengthen their dynamic monitoring and integrate their planning in order to prevent “overcapacity” and unnecessary losses. Furthermore, financial institutions should enhance their assessment frameworks to better evaluate the environmental and social impacts of their investments, provide targeted support for firms engaging in green innovation, and collaborate with policymakers to ensure that green finance initiatives are effectively aligned with broader sustainability goals.

6.3. Research Limitations and Recommendations to Further Research

Though this paper explores the mechanisms by which CSR influences the quality of economic growth, there are limitations to conducting this research, despite our analysis, theoretical models, and empirical testing. The core explanatory variable of this study is CSR. However, this study uses CSR indicators at the provincial level, which are derived from aggregating and averaging CSR scores of micro-level individual firms within provinces. Such measurement introduces potential errors because it is difficult to obtain CSR scores from all firms at the provincial level, which may further limit the explanatory power of our conclusions. Future research can further address the issue of data incompleteness by leveraging machine learning and big data analysis to obtain more comprehensive CSR scores from a larger number of firms. Additionally, utilizing firms’ ESG scores as proxies for missing individual CSR scores can enhance the accuracy and explanatory power of CSR metrics. This approach would improve the robustness of the analysis and provide a more precise assessment of CSR’s impact on economic growth quality.

There are several ways in which this study can be extended in the future. Firstly, our research concentrates on exploring the influence of CSR on economic growth quality. Using provincial-level data in China, we conduct this analysis. Data on Chinese A-share listed companies that can be used in future research. Data from listed companies can facilitate a richer empirical analysis, such as classifying firms into state-owned and non-state-owned enterprises. Consequently, incorporating the data of all these listed firms would also be valuable for research purposes. Secondly, in the theoretical model, government subsidies are not taken into account. However, firms alone cannot achieve sustainable growth. To achieve optimal economic growth and a green emission reduction trajectory, government subsidy policies, innovation policies, and financial policies must be synergized effectively. Therefore, the field would benefit from incorporating government subsidies into account in future research. Thirdly, future research can explore the significant roles of different stakeholders in promoting CSR and substantive green innovation by studying consumers’ green preferences, investors’ focus on sustainable development, and the environmental regulations set by regulatory agencies. Besides, analyzing the impact of non-governmental organizations and community groups also can offer a deeper understanding of how societal pressures affect corporate behavior. This analysis can help assess the collective impact of these stakeholders on economic growth and emphasize the importance of collaboration in advancing sustainability initiatives.