Assessing the Influence of Open Innovation among Chinese Cities on Enterprise Carbon Emissions

Abstract

Currently, China is the largest carbon emitter and the pressure of carbon reduction in China is very severe. However, the lack of technological innovation momentum is a bottleneck factor that restricts carbon reduction in Chinese cities. In this context, open innovation is gradually replacing closed innovation and playing an increasingly important role in improving the technological innovation performance of enterprises. Analysis shows that a large amount of literature has explored the impacts of industry technological innovation and green technology innovation on carbon emissions, while there is little research on how open innovation affects carbon emissions. This study calculates cities’ open innovation indicators and the carbon emission intensity indicators of listed enterprises. Using a three-fixed-effects model, it examines the effect and mechanism of open innovation on carbon emissions of enterprises and verifies the heterogeneity effect. The research results indicate that open innovation can significantly reduce the carbon emission intensity of enterprises by reducing transaction costs and upgrading the industrial structure. Further heterogeneity analysis shows that open innovation has an obvious carbon emission reduction effect on non-state-owned, polluting, small- and medium-sized enterprises and enterprises in central cities.

1. Introduction

With the rapid development of the global economy and the gradual warming of the global climate, the increasingly serious greenhouse effect has become an environmental problem that threatens human survival and is the biggest challenge facing the world in the 21st century. Countries around the world are actively seeking effective ways to reduce carbon emissions and slow down the pace of global warming. From the signing of UNFCCC to the adoption of the Kyoto Protocol, and then to the introduction of the Paris Agreement in 2015, the current global climate governance pattern has gradually formed. Figure 1 shows that China’s annual CO₂ emissions in 1985 reached 2 billion tons. In 1986, China’s annual CO₂ emissions accounted for over 10% of the world for the first time. Since then, the proportion has increased year by year, peaking at 31.2% in 2020. At present, China has become the largest carbon emitter, and the pressure of Chinese carbon reduction is very severe. In this situation, Chinese President Xi Jinping proposed the “Dual Carbon” goals of “Carbon peaking” and “Carbon neutrality” at the general debate of the 75th United Nations General Assembly held in September 2020. China officially launched the national carbon market in July 2021. Shortly afterward, the State Council successively issued the “Action Plan for Carbon Dioxide Peaking Before 2030” and “Responding to Climate Change: China’s Policies and Actions”. Through a series of measures such as planning, legislation, and policy formulation, China is promoting the implementation of the “Dual Carbon” goals through practical actions. China’s position and determination in carbon reduction play a very important role all over the world.

At present, China has taken multiple measures to promote technological innovation and produced a marked effect. In the light of the Global Innovation Index 2023 report released by the WIPO, China’s innovation index ranking has jumped from 34th in 2012 to 12th in 2023. Up to now, technological innovation has become one of the main driving forces for reducing Chinese carbon emissions and the determinant to realize the “Dual Carbon” goals. However, the lack of technological innovation is also a bottleneck factor in restricting carbon reduction. In this context, it is of great significance to clarify the logical relationship between technological innovation and carbon emission efficiency and its influencing mechanism for China to achieve the goal of “dual carbon”.

In the era of the network economy, the speed of technological updates is accelerating, and the innovation cycle is constantly shortening. It is no longer appropriate for enterprises to restrain their competitors through traditional intellectual property protection. The situation forces enterprises to transform their competitive mindset to a cooperative competition approach. To win competitive advantages, enterprises must accelerate the flow and exchange between internal knowledge and external knowledge. In this context, open innovation is gradually replacing closed innovation and playing an increasingly vital role in improving the technological innovation performance of enterprises. “Open innovation” utilizes purposeful knowledge flows to expedite internal innovation and expand the market of external innovation [1]. From an angle of technology flow, open innovation is defined as engaging in technology acquisition and development through the dynamic capabilities of enterprises, combining internal resources of enterprises with external ones [2]. Referring to classification, open innovation in enterprises is divided into inward-oriented and outward-oriented innovation [3]. Based on the four quadrants of introversion, extroversion, money, and non-money [4], open innovation is divided into four categories: display, sale, absorption, and acquisition. On the measurement of open innovation, ref. [5] used accounting indicators to measure the degree of open innovation, and [6] used the volume of cooperative patent applications to characterize open innovation. Cooperative patents are an important achievement of open innovation and a significant representation of the effectiveness of open innovation. Therefore, this study selected the number of cooperative patents as a variable to focus on verifying the impact of open innovation between cities on the carbon emissions of enterprises in China.

Theoretically, the effects of open innovation and carbon emission reduction involve multiple economic and management theories, including the Knowledge-based Theory and the Dynamic Capability Theory. The authors of [7] proposed the Knowledge-based Theory, which states that enterprises are a collection of knowledge. By the acquisition, integration, absorption, and transfer of internal and external knowledge, knowledge is transformed into innovative achievements to achieve value output, which ultimately forms their core competitiveness. The core idea of the Dynamic Capability Theory is that enterprises can perceive and seize external opportunities when facing complex and changing external environments, and can effectively combine old resources and integrate new resources to enhance core competitiveness [8]. With the development of the times, the theoretical connotations above will be continually enriched. It can be preliminarily observed that open innovation can promote enterprises to optimize their production technology, help them explore green development paths, and is a vital channel for enterprises to reduce carbon emissions. In the context of the new economy and technology era, this study selects Chinese cities as research samples and focuses on examining the impact of open innovation between Chinese cities on the carbon emissions of enterprises.

In the literature, there are relatively few issues regarding open innovation. The existing studies are mainly related to research on technological innovation. Firstly, regarding the research on the effects of technological innovation, the impact of green technological innovation on economic growth has a single-income threshold effect [9], and the green technology innovation capability of enterprises can enhance their competitiveness [10]. In addition, open innovation can effectively enhance enterprise factor productivity [11]. Secondly, the study of the factors of carbon emissions shows that policies, economy, and innovation can influence carbon emissions. For example, the construction of the National Independent Innovation Demonstration Zone [12] and the policies of China’s innovative city pilots [13] have significantly reduced carbon emissions. Economic growth helps to reduce carbon emissions [14]. Moreover, the measures of China’s environmental innovation [15] and digital transformation of enterprises [16] have effectively reduced carbon emissions. Regarding studies on the carbon emission effect of technological innovation, refs. [17,18] both pointed out that compared with general technological innovation, green technological innovation can significantly affect carbon emission efficiency. The studies show that green innovation has a significant positive effect on carbon emission efficiency [18,19,20], and collaborative green innovation is more effective than independent green innovation [21]. Currently, there is very little research on the impact of open innovation on carbon emissions. This is especially true for China where the progress of related research is restricted because of the lack of data about the carbon emissions of enterprises. However, this study selected micro-enterprises as the research subject to further expand our understanding of open innovation and enrich the study of the mechanisms of open innovation and the pathways of carbon reduction.

This study calculates the indicators of open innovation and the indicators of carbon emission intensity of Chinese listed enterprises from 2012 to 2021 and examines the effect of open innovation between cities on the carbon emissions of enterprises. Meanwhile, this paper also considers the heterogeneity of micro-enterprises. The results of this study indicate that strengthening technological cooperation between cities can raise the efficiency of technological innovation for enterprises, thereby reducing carbon emissions. The results of heterogeneity analysis are that open innovation has an obvious carbon reduction effect on non-state-owned, pollution-type, small, and medium-sized enterprises in central cities. Under the new economy and technology era, this study empirically analyzes the carbon emissions reduction effect of open innovation between cities, and multi-dimensionally verifies the mechanism and heterogeneity effect around the research. This study provides theoretical support for building mechanisms of open innovation between Chinese cities.

To sum up, this study provides the following contributions: Firstly, from a research perspective, this study provides a more comprehensive research framework to analyze how open innovation affects the carbon emission of enterprises by selecting two different perspectives: transaction costs and industrial structure. On the one hand, applying transaction cost theory to the field of environmental economics, especially carbon emissions issues, provides a new theoretical perspective for understanding how enterprises choose open innovation when facing high transaction costs. On the other hand, exploring how industrial structure affects the relationship between open innovation and carbon emissions provides new insights into the synergistic effects between industrial structure and environmental policies. Secondly, based on the research content, this study reveals the function of open innovation on the carbon emissions of enterprises through theoretical and empirical analysis. This not only enriches existing theories but also provides specific operational paths for practice. Promoting open innovation to reduce carbon emissions of enterprises guides enterprises in optimizing environmental performance and strategic support for achieving urban green development. Thirdly, based on the research data, this study chooses the number of cooperative patents as the measured indicator for open innovation and uses tools such as Python to mine the data. The data on the carbon emission intensity of Chinese enterprises is calculated by using information such as enterprises’ main business costs and revenues, industrial main business costs, and industrial energy consumption [16]. The accurate data above provide the basis for the research in this paper.

The main content below includes a literature review and theoretical mechanism, model and variables, empirical analysis, mechanism verification, heterogeneity analysis, and conclusions and suggestions.

2. Literature Review and Theoretical Mechanism

2.1. Literature Review

Based on our review of the literature, studies related to open innovation can be summarized as follows:

The research on the effects of open innovation mainly focuses on the innovation itself, and the interpretation of “open” is insufficient. Most experts studied the macroeconomic effects or microeconomic effects of technological innovation. The authors of [9] found that the effect of green technological innovation on economic growth has a single-income threshold effect. The authors of [22] empirically tested the spatial spillover effect of low-carbon technological innovation on energy efficiency based on Data Envelopment Analysis (DEA) and machine learning methods. The green technology innovation capability of enterprises can enhance their competitiveness [10]. Low-carbon technology innovation exerts a dominant positive influence on the performance of manufacturing enterprises [23], while [24] believes the impact is bidirectional. Furthermore, ref. [11] empirically tested whether open innovation can effectively enhance enterprises’ factor productivity.

The study of the influence factors of carbon emissions has attracted global attention. Multiple research results have shown that factors such as policies and the economy have impacts on carbon emissions. Several studies of policies’ impacts on carbon emissions show that the construction of China’s National Independent Innovation Demonstration Zone (NIIDZ) [12] and the policies of China’s innovative city pilots [13] have significantly reduced carbon emissions. As the population size increases, the promotion effect of the policies of China’s innovative city pilots on carbon reduction efficiency is enhanced [25]. The low-carbon city pilot policy can improve the synergistic emission reduction efficiency of CO2 and PM2.5 [26]. From both production and consumption perspectives, ref. [27] found that China’s low-carbon consumption policy has effectively reduced carbon intensity and per capita carbon emissions. A study on the effect of the economy on carbon emissions [14] concluded through empirical research that economic growth helps to reduce carbon emissions. The authors of [28] elaborated on the significant carbon reduction effect of the urban agglomeration economy based on the agglomeration economy theory. The authors of [29] pointed out that the increase in financial risk not only directly reduces global carbon emissions, but also indirectly affects global carbon emissions by promoting technological innovation.

Regarding the carbon emissions reduction effects of open innovation, a large amount of the literature is still exploring the effect of industry technological innovation and green technological innovation on reducing the carbon emissions of enterprises. In the field of industry technology innovation, refs. [30,31] concluded that renewable energy technology innovation is beneficial for enhancing carbon emission efficiency. The authors of [32] found that the relationship between carbon emissions and technological innovation in the logistics industry is U-shaped, which indicates that technological innovation in the logistics industry has a rebound effect on carbon emissions. The authors of [33] analyzed the relationship between R&D investment in new energy vehicle enterprises and carbon emissions reduction using technological innovation as a threshold variable. In the field of green technology innovation, refs. [17,18] found that green technological innovation has a greater impact on carbon emission efficiency compared to general technological innovation. The authors of [18,19,20] have all shown that green technological innovation has a significant positive effect on urban carbon emission efficiency, and upgrading industrial structure has played an important role in improving urban carbon emission efficiency through green technological innovation. In terms of open innovation, only a small amount of the literature has been indirectly involved. For example, ref. [34] concluded that improving the efficiency of collaborative innovation between industry, academia, and research can reduce carbon emissions. The authors of [21] used the number of green patents obtained to measure collaborative green innovation and compared the impact of collaborative green innovation and independent green innovation on carbon emission reduction. But the measurement indicators need to be improved. In addition, collaborative innovation is different from open innovation, which focuses on utilizing external innovation resources for innovation.

To summarize, green technology innovation can promote economic growth and competitiveness of enterprises, while open innovation can effectively enhance the total factor productivity of enterprises. Meanwhile, economic growth and innovation of enterprise both contribute to reducing carbon emissions. Regarding the carbon emissions reduction effects of open innovation, a large amount of the literature explored the influence of industrial technological innovation and green technological innovation on carbon emissions reduction, but there is little research on how open innovation affects carbon emissions. Based on the status of the existing research, this study selects micro-enterprises as a research subject and uses Python to retrieve all patent information of Chinese cities from the Wanfang Patent Database from 2012 to 2021 (including abstract, address, application year, authorization year, etc.). This study uses patents with several applicants ≥ 2 as cooperative application patents. Moreover, this study uses the number of cooperative patents to measure open innovation between cities and uses a three-fixed-effects model to study the mechanism of open innovation on carbon emissions. This study will deepen our comprehension of open innovation and enrich future study of open innovation on carbon reduction.

2.2. Theoretical Mechanism

Open innovation, as an emerging innovation model, has a potentially positive effect on reducing transaction costs and upgrading industrial structure, thereby reducing carbon emissions. As shown in Figure 2. At the level of micro-enterprises, transaction costs can affect transaction activities and organizational forms of enterprise, thereby affecting the industrial structure. When transaction costs are high, enterprises may be more inclined towards internal integration and vertical integration of the industrial structure. When transaction costs decrease, enterprises may be more inclined to seek external resources and partners, which promotes the industrial structure to a state of horizontal differentiation and collaborative development. From an intermediate perspective, changes in industrial structure can also affect the transaction costs of the enterprise. With the development of technology and changes in the market environment, the industrial structure will face continuous adjustment and upgrading. In this process, the emergence of emerging industries will affect the transaction costs of the enterprise.

Open innovation can significantly reduce the costs of information acquisition, information processing, and information exchange by promoting technology sharing and cooperation, thereby incentivizing more carbon emissions reduction activities. In addition, open innovation can promote the development of industrial structures in the direction of low energy consumption and low carbon emissions. Therefore, this study conducts in-depth research from the two perspectives of transaction costs and industrial structure.

Transaction costs refer to various related costs incurred in completing a transaction. The concept of transaction cost was initially proposed by Ronald Coase, Nobel laureate in economics, and mainly used to explain the essence of the existence of enterprises. Transaction cost is a broad concept that covers all related costs from the beginning to the end of a transaction, mainly including search cost, bargaining decision cost, supervision cost, etc. Transaction cost can affect the investment decision-making behavior of enterprises [35], and its impact on the development of enterprises cannot be ignored.

The innovation activities of enterprises require a large amount of R&D costs. Under the traditional closed innovation model, enterprises mainly rely on their own research and development capabilities for technological innovation. Due to limited resources and high costs, enterprises take high R&D risks. However, through their collaborative innovation, enterprises can effectively reduce trial-and-error costs, and improve innovation efficiency by utilizing the diffusion effect of innovation. Open innovation can promote knowledge sharing and technology transfer and facilitate enterprises to exchange information and cooperate with their external partners more frequently. Especially, as artificial intelligence and digitization continue to advance, the efficiency of obtaining information for enterprises has improved. These technological advances help to reduce information asymmetry and lower search costs and negotiation costs during the transaction process [36]. Simultaneously, technological progress has changed commodity exchange and reduced unit transaction costs [37]. The widespread application of innovation can effectively accelerate the circulation, aggregation, and integration of various factors and reduce transaction costs. Moreover, innovation can optimize resource allocation and improve the efficiency of factor allocation [38].

The issue of carbon emissions involves the concept of externalities. Ronald Coase proposed that if the transaction cost is zero, regardless of how rights are defined, the optimal allocation of resources can be achieved through market transactions. Therefore, reducing transaction costs is the key to building an effective trading mechanism. Although carbon trading reduces the costs of mitigating climate change, transaction costs may lower its cost-effectiveness [39]. However, when transaction costs exceed transaction income, carbon trading hurts enterprise performance [40]. Reducing transaction costs helps to drive the activity and allocation efficiency of the carbon trading market. In the carbon trading market, the reasonable allocation of carbon emission rights can be achieved by buying and selling carbon emission rights [41]. Carbon trading allows enterprises that effectively reduce emissions to obtain more emission rights, while those with higher emission reduction costs need to purchase more emission rights. When transaction costs decrease, the process of allocating carbon emission rights will become smoother and more efficient, which helps to promote the transfer of carbon emission rights to efficient emission reduction enterprises. In addition, reducing transaction costs also helps to promote the research and application of carbon emissions reduction technologies and improve green total factors productivity [42]. In light of this, this study formulates its first hypothesis:

H1: 

Open innovation can reduce carbon emissions by reducing transaction costs.

Innovation is the foundation and necessary condition for upgrading industrial structures, which is triggered by technological change. Schumpeter’s “Innovation Theory” emphasizes the important role of technological innovation and provides a preliminary explanation of how technological innovation affects socio-economic activities. Malerbar proposed the theory of the “Industrial Innovation System” from the perspective of Schumpeter’s industrial innovation model and the origin of technological institutions. The industrial innovation system can break through the boundaries of industries and enable more open interaction among participants within the industry. Moreover, it can achieve common industrial development through resource complementarity and knowledge sharing. Open innovation requires enterprises to integrate internal and external innovation resources and technologies to enhance their technological innovation capabilities. Technological innovation can change the original production mode, reduce production costs, and attract more innovative factors to improve factor allocation efficiency [43]. Green technological innovation promotes the transformation of low-value-added and heavily polluting industries towards high-value-added and low-polluting industries [9]. Through the dissemination of information among related industries, inter-industry skill complementarity is promoted [44] and industrial upgrading is accelerated [45]. In addition, the reduction in sharing costs contributes to the establishment of a close cooperation mechanism, thus promoting the upgrading of industrial structure [46]. Under the framework of open innovation, such cooperation and exchange can help break down barriers between industries. Moreover, it can promote the integration and development of different industries and accelerate the optimization and upgrading of industrial structures.

The characteristics of industrial structure significantly influence the distribution pattern of energy consumption and emissions across industries. The rationalization of industrial structure contributes to the optimization of factor configuration, rational allocation, and effective utilization of energy resources [47]. Upgrading the industrial structure also contributes to optimizing energy structure, minimizing energy waste, and enhancing energy utilization efficiency, thereby reducing carbon emissions [48]. In light of this, this study formulates its second hypothesis:

H2: 

Open innovation can reduce carbon emissions by upgrading industrial structures.

3. Model and Variables

3.1. Setting a Model

This study constructs an econometric model with fixed effects as follows:

$$\mathrm{l}\mathrm{n}{intencoo2}_{it}={\alpha }_0+{\alpha }_1{opi}_{it}+\phi X_{it}+\delta Y_{it}+F{E}_{city}+F{E}_{year}+F{E}_{indu}+{\epsilon }_{it}$$

(1)

In this formula, i denotes “Industry” and t denotes “Year”. intencoo2, as the dependent variable, denotes “Carbon Emission Intensity of Enterprise” and opi, as the core explanatory variable, denotes “Open Innovation”. X represents “Enterprise Control Variable” and Y represents “City Control Variable”. FE_city, FE_year, and FE_indu, respectively, represent fixed effects of city, year, and industry. This study controls for these three fixed effects to absorb omitted effects as much as possible. α₀ represents the “constant term” and ε_it represents the “random error term”. Additionally, in all regression equations, this study employed t-statistics adjusted for robust standard errors, with clustering at the industry year level.

3.2. Variable Description

(1)

The dependent variable: Carbon emission intensity of enterprise (intecoo2). Constrained by data limitations, current studies lack continuous variable data to measure Chinese carbon emissions. To measure intecoo2, this study divides the carbon dioxide emissions of an enterprise by the main business revenue of the enterprise [49]. However, the vast majority of Chinese enterprises have not directly disclosed their carbon dioxide emissions, and this study estimates them approximately by industrial energy consumption. The calculation formula is as follows:

$${em}_{it}={\displaystyle \frac{{cost\_f}_{it}}{{cost\_i}_{jt}}}\times {TEC}_{jt}\times 2.493$$

(2)

$${intencoo2}_{it}={\displaystyle \frac{{em}_{it}}{{income\_f}_{it}}}\times {\displaystyle \frac{1}{1000000}}$$

(3)

where intencoo2 denotes “Carbon Emission Intensity of Enterprise”. If intencoo2 is larger, the intensity is greater. TEC denotes “Industrial Energy Consumption”. Referring to the standard provided by the Xiamen Energy Conservation Center, burning per ton of standard coal can approximately discharge 2.493 tons of carbon dioxide [50]. In the formula, j represents “industry”, income_f denotes “Main business revenue of enterprise”, cost_f denotes “Main business cost of enterprise”, and cost_i denotes “Industrial main business cost”.

(2)

Core explanatory variable: Open innovation (opi1). This study uses cooperative patent applications to measure the level of open innovation between cities. The specific identification process is as follows:

First, this study uses Python 3.10.2 software to capture all patent information (including abstract, address, application year, authorization year, etc.) in China from the Wanfang Patent Database from 2012 to 2021.

Second, to obtain cooperative patent data, patents with several applicants ≥ 2 are selected as patents with cooperative patent applications, and the screening function of Stata 17 software is used to screen out patents containing cooperative application information.

Third, the address information of each patent is used to identify the city to which each patent belongs, and the patent authorization date is used as the date of cooperative patent applications. Then, the annual number of cooperative patent applications in each city is obtained by summing up.

Then, to add up the word frequency of cooperative patent applications plus 1, take the natural logarithm as the characterization indicator of open innovation (lnopi1).

This study also draws a spatial distribution map of open innovation (Figure 3), which reveals the temporal growth trend and spatial distribution overview of open innovation. The results in Figure 3 indicate that from 2012 to 2021, the level of open innovation shows a sharp upward trend, with significant improvements observed across the country. From a spatial distribution perspective, the eastern region has the highest level of open innovation, especially in the Yangtze River Delta region. The central region ranks second, while the western region has the lowest level.

(3)

Control variables. This study has two control variables: enterprise and city. Here, there are 6 variables at the enterprise level and 4 variables at the city level, as shown in

Table 1. Variable definition and explanation.

Variable Type	Variable Type	Variable Representation Symbol	Explanation
The dependent variable	Carbon emission intensity of the enterprise	lnintencoo2	CO2 emissions of enterprise/main business income of the enterprise
The core explanatory variables	Level of open innovation	lnopi1	Natural logarithm of (total word frequency of joint patent authorization + 1)
The control variables	Enterprise scale	lnsize	Natural logarithm of total assets of enterprise
	Cash flow ratio of enterprise	cflow	Net cash flow generated from business activities of enterprise/total assets of enterprise
	Enterprise nature	govcon	If it is a state-owned enterprise, it is defined as 1, otherwise, it is defined as 0.
	Growth rate of enterprises	tagr	Growth rate of total assets of the enterprise
	Years of listing	lnage	Natural logarithm of (sample year of enterprise—year of listing)
	Concentration of enterprise equity	top10	Shareholding ratio of the top ten shareholders of the enterprise
	Level of foreign investment in cities	fdi	Foreign direct investment in cities/urban GDP
	Advancement of urban industrial structure	indu	Added value of urban tertiary industry/added value of secondary industry
	Urbanization rate	urban	Urban population/regional total population
	Level of urban economic development	lnrgdp	Natural logarithm of urban GDP

.

3.3. Data Sources and Descriptive Statistical Analysis

Based on the initial samples of this paper, the data sources are mainly from WIND and CSMAR databases. Referring to most of the literature, this study also excluded the sample of enterprises with missing variables and finally obtained 10,747 observations from 2240 listed enterprises between 2012 and 2021.

Table 2. Results of descriptive statistics.

Variable	N	Mean	S.D.	Min	Max
lnintenco	10,747	3.1635	1.1242	0.3874	5.9787
lnopi1	10,747	7.0079	1.3925	1.0986	10.3198
lnsize	10,747	22.1138	1.1946	16.1613	27.5470
cflow	10,747	0.0498	0.1306	−10.2162	2.2216
govcon	10,747	0.2711	0.4445	0.0000	1.0000
tagr	10,747	0.1424	0.4125	−0.9661	22.5489
lnage	10,747	2.1071	0.7599	0.6931	3.4657
top10	10,747	0.4170	0.1896	0.1011	0.9849
fdi	10,747	0.0212	0.0109	0.0001	0.0796
indus	10,747	1.4992	0.8824	0.6112	5.2440
urban	10,747	68.6936	10.9000	35.7300	89.6000
lnrgdp	10,747	11.6015	0.4249	9.4548	13.0557

lists the descriptive statistical results for all variables, which shows that the mean, standard deviation, minimum, and maximum are all scientifically reasonable. It needs to be explained that although opi1 is a discrete variable, its value reaches more than 10,000. Therefore, we performed logarithmic processing on it.

3.4. Correlation Analysis

This study also drew a heatmap for the correlation analysis between open innovation and the carbon emission intensity of enterprises (Figure 4). The heatmap reveals a significant negative correlation and signifies that open innovation presumably helps to lower the carbon emission intensity of enterprises. In addition, we drew a binscatter linear fitting diagram of open innovation and enterprise carbon emission intensity, as shown in Figure 5. It can also be seen that there is an obvious negative relationship between open innovation and enterprise carbon emission intensity, which preliminarily indicates that open innovation is likely to reduce enterprise carbon emission intensity. These results preliminarily confirm the basic results of hypothesis 1 and hypothesis 2, that is, open innovation can reduce carbon emissions. Of course, to more rigorously verify the significance of this negative relationship, in Section 4 we use econometric methods to prove it.

4. Empirical Analysis

4.1. Benchmark Regression

The benchmark regression results are listed in

Table 3. The results of benchmark regression.

	EV: lnintencoo2
	(1)	(2)	(3)	(4)	(5)
lnopi1	−0.0542 ***	−0.0698 ***	−0.0503 ***	−0.0414 ***	−0.0429 ***
	(0.0050)	(0.0055)	(0.0094)	(0.0090)	(0.0089)
lnsize				0.0141 ***	0.0142 ***
				(0.0039)	(0.0039)
cflow				−0.2030 *	−0.2036 *
				(0.1226)	(0.1228)
govcon				0.0684 ***	0.0682 ***
				(0.0124)	(0.0124)
tagr				−0.0245 **	−0.0249 **
				(0.0097)	(0.0099)
lnage				0.0622 ***	0.0616 ***
				(0.0071)	(0.0070)
top10_HHI				0.0268	0.0265
				(0.0231)	(0.0230)
fdi					−0.7671
					(0.5204)
indus					0.0541
					(0.0373)
urban					0.0104 ***
					(0.0035)
lnrgdp					−0.0192
					(0.0187)
Constant	3.5434 ***	3.6527 ***	3.5146 ***	2.9921 ***	2.4439 ***
	(0.0335)	(0.0370)	(0.0651)	(0.1002)	(0.3234)
FEindu	YES	YES	YES	YES	YES
FEyear	NO	YES	YES	YES	YES
FEcity	NO	NO	YES	YES	YES
Observations	10,747	10,747	10,733	10,733	10,733
R-squared	0.9466	0.9516	0.9537	0.9580	0.9581

Note: The asterisks *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively.

. By controlling only for industry-fixed effects, the regression coefficient result is significantly negative at 1%. The results are shown in column (1) of Table 3. The regression results of controlling for year and city-fixed effects are shown in columns (2) and (3). Columns (4) and (5) add control variables of enterprises and cities in sequence. It can be seen that the R² of the model is on the increase, and the interpretability is continuously rising. The regression coefficient has always been significantly negative at 1%. The results correspond to the expectations of this study. In addition, the control variable results are also basically consistent with the expectation. The higher the cash flow ratio and growth speed of the enterprise, the higher the level of urban economic development and foreign investment, and the easier it is to lower carbon emissions. On the contrary, state-owned enterprises, large enterprises, and enterprises with high equity concentration are more likely to increase their carbon emission intensity. The above results once again confirm the basic results of hypothesis 1 and hypothesis 2, that is, open innovation can reduce carbon emissions, but the specific mechanism needs further empirical testing.

4.2. Endogeneity Regression

To conduct a regression analysis to test for endogeneity, we first replaced the explanatory variables. Considering the bias caused by the variables, this study replaced opi1 with opi2, and opi2 denotes the number of cooperative patents calculated using the “application date as the year”. The results are shown in column (1) of

Table 4. The results of endogeneity treatment.

	Replacing Explanatory Variables	Controlling the Cooperative Effect	IV1 = L.lnopi1	IV2 = lnshc × year
	(1)	(2)	(3)	(4)
lnopi1		−0.0407 *** (0.0080)	−0.0785 *** (0.0219)	−0.4907 *** (0.1607)
lnopi2	−0.0337 *** (0.0102)
Controls	YES	YES	YES	YES
FEindu	YES	YES	YES	YES
FEyear	YES	YES	YES	YES
FEcity	YES	YES	YES	YES
N	10,733	10,721	7736	10,683
R2	0.9581	0.9650	0.0331	−0.1126
First stage regression
IV			0.286 *** (0.016)	0.011 *** (0.002)
F			338.61 ***	25.89 ***
Kleibergen–Paap rk LM statistic			475.83 ***	26.58 ***
Cragg–Donald Wald F statistic			849.06 ***	51.39 ***
Kleibergen–Paap rk Wald F statistic			338.61 ***	25.89 ***

Note: The asterisks *** denotes statistical significance at the 1% levels.

. It needs to be explained that although opi2 is a discrete variable, its value reaches more than 10,000. Therefore, we perform logarithmic processing on it. The conclusion that “open innovation can reduce the carbon emission intensity of enterprises” is unchanged. Second, we controlled the cooperative effect. This study tests the fixed effects of industry, year, and city, as well as the control variables at the dual levels of enterprises and cities by benchmark regression, but “omitted variables” have to be addressed. Therefore, this study further incorporates industry year effects for regression analysis. The results are shown in column (2). It can be observed that the core conclusion that “open innovation can reduce the carbon emission intensity of enterprises” has not changed either. Third, we conducted instrumental variable regression. The instrumental variable method is a commonly used method to solve endogeneity problems, so this study uses the core explanatory variable lagged one period as the instrumental variable (IV1 = L.lnopi1). The results are shown in column (3). Currently, many studies have begun to focus on the impact of culture on the behavior of enterprises. Through a survey of private enterprises in China, it is found that Confucian culture has a significant promoting effect on enterprises’ R&D investment [51]. Therefore, this study refers to the existing literature and uses the multiplicative terms “Number of Confucian schools in different regions” and “Year” as instrumental variables (IV2 = lnshc × year). The results are shown in column (4). The result of the first stage of instrumental variable regression is significantly positive, which shows that the greater the number of Confucian schools, the stronger the ability to develop open innovation. Confucian culture is an important symbol of traditional Chinese culture, deeply influencing the operation and management of Chinese enterprises. Confucian schools are important carriers for promulgating ideology and culture. The more schools there are the better the cultural environment and traditions in the region. Meanwhile, a society with a higher level of education and cultural literacy may place more emphasis on innovation and be more open and inclusive. And Confucian culture emphasizes innovation ideology, which helps to stimulate entrepreneurs’ passion for innovation. All of these are conducive to the formation of an open innovation environment. The number of Confucian schools, as historical data, has the advantage of being objective and stable, and its quantity has no impact on carbon emissions. The estimation results of the second stage are significantly negative. The conclusion is robust. The instrumental variables passed the significance test for identifying the problem, so the selected instrumental variables were effective. In summary, the regression results of this study were not affected by endogeneity issues.

4.3. Restricted Sample Regression

Considering the impact of sample extreme values on the results, this study further investigates the effect of open innovation on the carbon emissions of enterprises by restricting the sample. First, during the period of the pandemic of COVID-19 from 2020 to 2021, considering that a large number of Chinese enterprises stopped production and their carbon emissions declined significantly, this study excluded the sample data from 2020 and 2021. The regression results show that open innovation has reduced the carbon emissions of enterprises. Please refer to column (1) of

Table 5. The results of restricted sample regression.

	EV: lnintencoo2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
lnopi1	−0.0423 ***	−0.0390 ***	−0.0429 ***	−0.0429 ***	−0.0429 ***		0.0301
	(0.0096)	(0.0096)	(0.0138)	(0.0103)	(0.0118)		(0.0253)
lnopi1_w						−0.0290 ***
						(0.0084)
lnopi1 × lnopi1							−0.0053 ***
							(0.0018)
Constant	2.8980 ***	3.1841 ***	2.4439 ***	2.4439 ***	2.4439 ***	2.8250 ***	2.3991 ***
	(0.3138)	(0.3487)	(0.5858)	(0.2444)	(0.5073)	(0.0955)	(0.3262)
FEindu	YES	YES	YES	YES	YES	YES	YES
FEyear	YES	YES	YES	YES	YES	YES	YES
FEcity	YES	YES	YES	YES	YES	YES	YES
Observations	7261	8921	10,733	10,733	10,733	10,733	10,733
R-squared	0.9635	0.9646	0.9581	0.9581	0.9581	0.9640	0.9582

Note: The asterisks *** denotes statistical significance at the 1% levels.

for details. Second, Beijing, Shanghai, Tianjin, and Chongqing are the most developed and influential cities in China’s economy. Although the four municipalities have a small spatial scope, they gather a large number of listed enterprises in China. So, to eliminate the interference of the four municipalities on the sample of enterprises, this study deleted the samples of the four municipalities and conducted another regression analysis. The conclusion “open innovation reduces the carbon emission intensity of enterprises” did not change, as shown in column (2) of Table 5. Third, after adjusting the clustering standard error to the individual industry level, each year level, and individual city level, it was found that the core conclusion remains unchanged. Please refer to the results in columns (3)–(5). Fourth, to eliminate the interference of singular values on the results, this study truncates the maximum and minimum values of each indicator by 1%, and the conclusion that “open innovation can reduce the carbon emission intensity of enterprises” still holds. Please refer to the results in column (6). Finally, to eliminate nonlinear effects, this study introduces the quadratic term of open innovation for nonlinear regression. The estimated results show a linear relationship between open innovation and the carbon emissions intensity of enterprises. Please refer to the results in column (7). Overall, the conclusion that open innovation can reduce the carbon emission intensity of enterprises is robust all along.

5. Mechanism Verification

5.1. Transaction Costs

Open innovation can help enterprises reduce transaction costs and achieve resource integration and sharing. Reducing transaction costs can stimulate enterprises to innovate and apply low-carbon technologies and lower carbon emissions. In the two main categories of transaction costs, internal transaction costs mainly include the expenses borne by an enterprise to resolve conflicts between supply and demand. These costs are closely related to the daily operational activities of the enterprise, including production process management, employee training, resource allocation, collaborative work with other business units, etc. The external transaction costs of an enterprise refer to the costs incurred in conducting transactions in the open market, mainly including search costs, negotiation costs, and fulfillment costs. Therefore, this study uses the proportion of management expenses to total assets to characterize internal transaction costs (cost1) and uses the ratio of financial expenses to total liabilities to characterize external transaction costs (cost2). The results are shown in columns (1) and (2) of

Table 6. The results of mechanism verification.

	Cost1	Cost2	Cost2	Cost2	Ais	Ais	Ais	Ris1	Ris2
	All	All	Small E	Big E	All	Center C	NonCenter C	Center C	NonCenter C
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
lnopi1	0.0014	−0.0067 **	−0.0101 *	−0.0029 **	0.0239 ***	0.0380 ***	−0.0596 ***	−0.0049 **	0.0105 ***
	(0.0012)	(0.0029)	(0.0064)	(0.0014)	(0.0088)	(0.0111)	(0.0103)	(0.0022)	(0.0040)
Constant	0.2297 ***	0.2703 **	0.1344	0.2744	−1.1726 ***	−2.9419 ***	−0.4122 *	−0.2836 ***	−0.3754
	(0.0689)	(0.1136)	(0.1198)	(0.2107)	(0.2502)	(0.6153)	(0.2340)	(0.0665)	(0.2429)
Observations	10,733	10,733	4696	6015	10,623	4065	6558	6655	4073
R-squared	0.2869	0.0888	0.1055	0.1529	0.9835	0.8714	0.9867	0.8709	0.8924
FEindu	YES	YES	YES	YES	YES	YES	YES	YES	YES
FEyear	YES	YES	YES	YES	YES	YES	YES	YES	YES
FEcity	YES	YES	YES	YES	YES	YES	YES	YES	YES

Note: The asterisks *, **, and *** denote statistical significance at the 10%, 5%, and 1% levels, respectively.

. Open innovation significantly reduces the external transaction costs of enterprises, and thus hypothesis 1 has been validated. Open innovation can improve the efficiency of enterprises in utilizing external innovation resources and reduce the costs of finding suitable technologies and partners. Furthermore, it cannot be ignored that an open innovation model may lead to an increase in the internal transaction costs of an enterprise. This is because open innovation requires coordination of internal and external resources, which can increase internal transaction costs. Enterprises should seek a balance between reducing external transaction costs and controlling internal transaction costs to achieve optimal overall efficiency. In particular, we divided the sample into large and small enterprises, i.e., by the median of the total assets of the enterprises. The results are shown in columns (3) and (4) of Table 6. Open innovation is more conducive to reducing the financial costs of small enterprises. It shows that open innovation is beneficial to small enterprises and reduces the cost of innovation. This further validates the expectations of hypothesis 1.

5.2. Industrial Structure

Under the open innovation model, enterprises can better utilize external resources and achieve optimal allocation. This helps to share resources between industries, promote industrial structure optimization, and thus reduce carbon emissions. Previous studies have mainly considered the optimization of industrial structure from two dimensions: advancement and rationalization [52]. The advancement of industrial structure mainly refers to the development of industrial structure from low level to high level. The rationalization of industrial structure focuses more on adjusting the unreasonable industrial structure and promoting the coordinated development of various industries. Referring to existing research, this study matches data on the two dimensions of the industrial structure of the region. The ratio of the “output value of the tertiary industry to the output value of the secondary industry” is adopted to measure the advancement of industrial structure (ais). The “Thiel index” is adopted to measure industrial structure rationalization (ris). The specific calculation formula is as follows:

$${{ris}_{i,t}={\sum }_{m=1}^3{y}_{i,m,t}\mathrm{l}\mathrm{n}(y_{i,m,t}/l_{i,m,t})}_{i,t}\hspace{1em}m = 1,2,3$$

(4)

where y_i,m,t denotes the proportion of “output value (m-industry; t-period; i-region) to regional GDP (t-period; i-region)”. l_i,m,t denotes the proportion of “employment (m-industry; t-period; i-region) to total employment (t-period; i-region)”. The larger the rationalization value of the industrial structure, the more obvious the industrial structure deviation from the equilibrium state, and thus the more unreasonable the industrial structure is.

While analyzing the influence of open innovation on the advancement of industrial structure, this study also conducts empirical analysis on the rationalization data of central cities (ris1) and non-central cities (ris2). The results are shown in column (5) of Table 6. It can be seen that open innovation has significantly improved the advancement of industrial structures and contributed to optimizing and upgrading industrial structures. Therefore, hypothesis 2 has been validated. In particular, we divided the sample into central and non-central cities. Central cities include municipalities directly under the central government, provincial capitals, separately planned cities, and important node cities. All other cities are categorized as non-central cities. The results are shown in columns (6) and (7) of Table 6. Open innovation is more conducive to the upgrading of the central city’s financial industry structure. This further validates the expectation of hypothesis 2. The data in columns (8) and (9) indicate that open innovation contributes to central cities’ rationalization of industrial structure, but has the opposite effect in non-central cities. The probable reasons for this phenomenon are that enterprises in central cities can more conveniently utilize the benefits of resource aggregation and knowledge spillover. The vast economic scale and diversified market demand of central cities provide enterprises with abundant opportunities for product and service innovation. Relatively speaking, non-central cities face a more severe situation. The innovation mechanism in non-central cities is relatively inadequate. Moreover, non-central cities lack sufficient research institutions and enterprise R&D departments to form effective synergies. Moreover, non-central cities are often limited by the scarcity of innovative resources, talent loss, and limited market demand, which makes it difficult for them to fully absorb the dividends brought by open innovation. These factors above work together to limit the potential of non-central cities to achieve industrial structure rationalization through open innovation.

6. Heterogeneity Analysis

6.1. Equity Nature

Depending on equity nature, enterprises can be divided into two categories: state-owned and non-state-owned. To analyze the heterogeneity of equity nature, this study introduces an interaction term (lnopi1 × govcon) between open innovation and enterprise nature based on the benchmark model in Formula (1). If the enterprise is state-owned, govcon = 1, and vice versa govcon = 0. The regression results of the heterogeneity of equity nature are reported in column (1) of

Table 7. The results of the heterogeneity test.

	Equity Nature	Enterprise Size	Polluting Industrial Enterprises	Geographical Location
	(1)	(2)	(3)	(4)
lnopi1	−0.0572 ***	−0.3191 ***	−0.0276 ***	−0.0345 ***
	(0.0098)	(0.0570)	(0.0074)	(0.0061)
govcon	−0.1838 ***	0.0667 ***	0.0566 ***	0.0589 ***
	(0.0459)	(0.0123)	(0.0116)	(0.0118)
lnopi1 × govcon	0.0358 ***
	(0.0065)
lnsize	0.0130 ***	−0.0760 ***	0.0147 ***	0.0145 ***
	(0.0039)	(0.0178)	(0.0037)	(0.0038)
lnopi1 × lnsize		0.0124 ***
		(0.0025)
lnopi1 × pollute			−0.0384 ***
			(0.0112)
lnopi1 × central				−0.0057 ***
				(0.0020)
Constant	2.7523 ***	4.4204 ***	2.2676 ***	2.3007 ***
	(0.3265)	(0.4970)	(0.2961)	(0.2988)
FEindu	YES	YES	YES	YES
FEyear	YES	YES	YES	YES
FEcity	YES	YES	YES	YES
Observations	10,733	10,733	10,747	10,747
R-squared	0.9582	0.9583	0.9572	0.9572

Note: The asterisks *** denotes statistical significance at the 1% levels.

. The interaction coefficient is significantly positive, which is opposite to the coefficient of the open innovation variable. Open innovation is not conducive to reducing the carbon emissions of state-owned enterprises but is more conducive to non-state-owned enterprises. What is the possible reason? Non-state-owned enterprises are more sensitive to facing market changes and pay more attention to cost control and efficiency improvement. To enhance market competitiveness, non-state-owned enterprises are more motivated to adopt carbon emission reduction measures. By contrast, state-owned enterprises may face more government control and regulation, and the processes of decision-making may be more cumbersome. Moreover, the motivation of state-owned enterprises to change existing operation patterns and production modes may be insufficient. Meanwhile, state-owned enterprises may focus more on economic benefits, while emission reduction investments may not bring direct economic returns in the short term. These factors may all hinder state-owned enterprises from actively utilizing open innovation to reduce carbon emissions.

6.2. Enterprise Size

Depending on the enterprise size (size) defined in this paper, an interaction term (lnopi1 × lnsize) between open innovation and enterprise size is introduced. The regression results of heterogeneity of enterprise size are reported in column (2) of Table 7. The coefficient of the interaction term is significantly positive, which is opposite to the coefficient of the open innovation variable. It can be seen that open innovation is instrumental in reducing the carbon emission intensity for small- and medium-sized enterprises. The possible reasons are as follows: firstly, small- and medium-sized enterprises (SMEs) have more flexible organizational structures and decision-making mechanisms, which can respond to market changes more quickly. Secondly, small- and medium-sized enterprises often lack sufficient research and development resources and capabilities. Through open innovation, they can more effectively utilize external resources to promote their low-carbon transformation. Finally, small- and medium-sized enterprises may pay more attention to cost control and green image during market competition, thus proactively taking measures for carbon emissions reduction.

6.3. Polluting Industrial Enterprises

Based on the benchmark model in Formula (1), this study sequentially introduces the categorical variable of the polluting industrial enterprise (pollute) and the interaction term between “open innovation” and “polluting industrial enterprise classification” (lnopi1 × pollute). Among them, if the enterprise is a pollution type, then pollute = 1, and vice versa, pollute = 0. The regression results of the heterogeneity of polluting industrial enterprises are shown in column (3) of Table 7. The interaction coefficient is significantly negative, which indicates that open innovation is more conducive to reducing carbon emissions for pollution-type enterprises. The main reason is that polluting enterprises often face stricter environmental regulations and standards. They need to constantly update their technology to reduce carbon emissions in order to avoid penalties for violating emissions regulations. In addition, open innovation is more conducive to meeting market demand through technological updates and knowledge sharing, thus reducing the carbon emissions of polluting enterprises.

6.4. Geographical Location

The effects of open innovation on the carbon emissions of enterprises may vary while the geographical locations of enterprises are different. To explore the characteristics of carbon emissions of enterprises in central cities and non-central cities, this study introduces the geographic location variable (central) and an interaction term between “open innovation” and “geographical locations of enterprise” (lnopi1 × central) based on the benchmark model in Formula (1), where, if the enterprise belongs to the central city, central = 1, and vice versa, central = 0. The regression results are reported in column (4) of Table 7. The interaction coefficient is significantly negative, which indicates that open innovation is instrumental in reducing the carbon emissions of enterprises in central cities. The central cities have good geographical advantages, more innovative resources, and can fully absorb the dividends of open innovation. The central cities can continuously improve urban innovation capabilities and promote the research and application of low-carbon technologies through open innovation.

7. Conclusions and Suggestions

The increasingly serious global greenhouse effect has become an environmental problem that threatens human survival and poses a serious menace to the development of human society. China is taking multiple measures to promote technological innovation and implementing the “Dual Carbon” goals through practical actions. Here, this study investigates the effects of open innovation between cities on the carbon emissions of enterprises in China. This study selected the data of Chinese A-share-listed enterprises in Shanghai and Shenzhen from 2012 to 2021 as the initial sample. This study constructed characterization indicators for open innovation by using Python software to capture all patent information from the Wanfang Patent Database and utilized the filtering function of Stata software to screen out patents containing cooperative application information. From both theoretical and empirical perspectives, this study has tested the impact of open innovation between cities on the carbon emissions intensity of enterprises. Research has found that, firstly, open innovation between cities significantly reduces the carbon emission intensity of enterprises. This conclusion still holds after a series of robustness tests, such as replacing explanatory variables, controlling for cooperative effects, instrumental variable regression, and restricted samples. Secondly, mechanism testing indicates that open innovation can significantly reduce carbon emissions by reducing transaction costs and upgrading industrial structures. Thirdly, there is heterogeneity in the impact of open innovation on carbon emissions reduction in enterprises. Open innovation has an obvious carbon emission reduction effect on non-state-owned, polluting, small- and medium-sized enterprises and enterprises in central cities. The main reasons for the carbon emission reduction effect of open innovation are as follows: Open innovation has significantly promoted the advancement and rationalization of the industrial structure of central cities. Non-state-owned enterprises are more sensitive to facing market changes and pay more attention to cost control and efficiency improvement. Small- and medium-sized enterprises place greater emphasis on environmental image in market competition.

In the era of the network economy, technological innovation is accelerating, and the situation is forcing enterprises to shift their competition mindset to a cooperative competition approach. Open innovation has brought new ideas for enterprises to carry out technological innovation to promote carbon emissions reduction. Open innovation enhances the technological innovation capabilities of enterprises through cooperation between cities, and it is the main path to effectively reduce transaction costs and upgrade industrial structure. The results of this study provide useful insights for the study of open innovation and carbon emissions-related issues. In future studies, the question of how to accelerate the development of open innovation between cities to raise carbon emissions reduction efficiency remains. This study presents the following suggestions:

Firstly, the government should lead the construction of an open innovation ecosystem. Different cities may have different technological needs and advantages in the field of carbon emissions reduction. Building an open innovation platform between cities can promote cooperation among enterprises, research institutions, and social organizations. The platform can reduce the research and application costs of carbon reduction technologies through resource sharing. Furthermore, strengthening intellectual property protection can help increase the enthusiasm of enterprises towards open innovation.

Secondly, upgrading of the urban industrial structure should be continuously promoted. There may be opportunities for industrial complementarity and cooperation between different cities, so it is necessary to encourage industrial coordinated development. The promotion of collaborative innovation for the application and promotion of carbon reduction technologies is profitable throughout the entire industry chain. The establishment of inter-city industrial alliances and the promotion of industrial docking cooperation can maximize the benefits of carbon emission reduction.

Thirdly, a good combination of policy, economy, technology, and talents should be utilized. The government should increase policy support for open innovation in non-central cities’ enterprises and state-owned enterprises, including financial support, tax incentives, and innovation funds, to encourage non-central cities to engage in economic cooperation with surrounding areas or central cities, and cooperatively bear the costs and risks of open innovation. Local governments should guide state-owned enterprises to balance economic benefits and environmental responsibilities and incorporate emission reduction into their long-term development strategies. At the same time, local governments should increase support for technological innovation of state-owned enterprises and stimulate the vitality of open innovation in state-owned enterprises. By cultivating innovative talents in the field of carbon emission reduction, the application of carbon emission reduction technology between cities can be promoted.

This study examines the impact, mechanism, and heterogeneity of open innovation between cities on carbon emission reduction. Open innovation can reduce the carbon emissions intensity of enterprises by diminishing transaction costs and optimizing industrial structures, but there may be undiscovered paths that require further exploration. In addition, due to data acquisition being limited, this paper only uses Chinese A-share-listed enterprises as research samples. If the research sample is expanded to cover all enterprises, the research conclusions may be more accurate.