Research on the Relationship Between Structural Characteristics of Corporate Social Networks and Risk-Taking Levels: Evidence from China

Abstract

The complex market environment places unprecedented pressure on business decision-making processes. Effectively utilizing existing social resources to establish risk prevention mechanisms and accurately assess an enterprise’s risk-taking capacity has become a core issue for corporate survival and development. This paper examines 1810 listed companies on the Shanghai and Shenzhen A-shares markets from 2010 to 2022, constructing comprehensive social networks based on multiple corporate governance entities. It investigates the influence and transmission mechanisms of corporate social networks on risk-taking levels. The results reveal that (1) enhanced corporate social network centrality, structural holes, and connectivity significantly and positively affect corporate risk-taking levels; (2) information transparency and corporate governance quality serve as important mediating mechanisms through which social networks influence corporate risk-taking; (3) significant heterogeneity exists regarding executives’ backgrounds and industry attributes—specifically, in firms with executives possessing financial backgrounds and in high-tech industry enterprises, network characteristics play a more pronounced role in promoting risk-taking. This research not only enriches the literature on factors influencing enterprise risk-taking but also provides theoretical foundations and practical insights for improving corporate risk management capabilities through optimized social network structures.

1. Introduction

At present, the focus of China’s economic growth is shifting from quantitative catching-up to high-quality development. At the same time, the international situation is characterized by the prevalence of unilateral trade doctrines and a rising wave of anti-globalization. As a result, the market environment faced by enterprises has become increasingly complex [1,2]. With the continuous advancement of capital market reform, the unique market structure, regulatory environment, and corporate governance characteristics of the A-share market in China determine the nature of risks faced by enterprises, and their risk management methods have distinctive Chinese characteristics [3,4]. As most enterprises are state-controlled or have local government backgrounds, the behavior of these enterprises in the capital market is subject to changes in government policies, industrial policies, and macro-economic situations [5,6]. In addition, the investor structure of the A-share market in China is also distinctive, with retail investors as the majority. The market is highly volatile, and information asymmetry is serious [7,8,9]. These factors make the risks faced by A-share enterprises highly uncertain and complex.

Corporate risk-taking generally refers to an enterprise’s attitude and acceptance of risks in the course of its operations [10,11]. As an important indicator of an enterprise’s ability to cope with changes in the external environment and internal operating pressure, the level of enterprise risk-taking has always been a key variable in academic research and enterprise strategic decision-making [12,13]. How to maintain competitiveness in an uncertain market environment and take necessary risk-taking actions at appropriate times has become a core issue of concern for many enterprise managers and scholars. For different enterprises, risk-taking levels may vary due to factors such as industry characteristics, management attitudes, and capital structure [14,15,16]. Risk-taking includes not only an enterprise’s ability to cope with operating uncertainties but also its risk tolerance in capital market financing decisions, investment decisions, and product innovation decisions.

In this context, how to use social resources, especially social network resources, to affect the risk-taking ability of enterprises has become a topic worthy of further discussion [17,18,19]. Social network theory is an important field for studying the relationship and interaction between individuals or organizations, emphasizing the key role of social relations in resource acquisition, information dissemination, and behavior selection [20]. Since the 1990s, scholars have gradually applied it to the field of enterprise management. In the field of enterprise management, social networks involve not only the internal organizational structure of the enterprise and the relationship between employees, but also the relationship between the enterprise and external stakeholders (such as suppliers, customers, investors, etc.). A large number of scholars have studied the effect of social networks on enterprise efficiency [21,22,23], performance [24,25], innovation [26,27,28], access to resources [29,30], and other aspects.

However, research on the impact of corporate social networks on risk-taking levels is still relatively scarce, especially in China’s A-share enterprises; how social networks affect the risk-taking level of enterprises remains an important topic worth exploring. At the same time, studies on the construction of social networks are often limited to analyzing the role of certain types of governance subjects, such as the board of directors [31,32] or shareholders [33,34]. A few scholars have expanded the network structure to build a “shareholder–director” network [26] and the “senior executives and directors” network [27]. However, few studies build social networks based on complete corporate governance subjects. Based on this, this research focuses on A-share enterprises in China, using data from governance bodies including shareholders’ meetings, boards of directors, management teams, and boards of supervisors to construct a comprehensive social network. It explores the impact mechanism of corporate social networks on corporate risk-taking levels, aiming to fill the research gap in this field.

The purpose of this study is to explore the impact of corporate social networks on risk-taking levels in China’s A-share market. This study examines Shanghai and Shenzhen A-share listed companies from 2010 to 2022 as samples, with governance data and financial data obtained from the CSMAR, iFind, and Wind databases. Building upon the “Board of Directors Supervision and Administration” network, this research expands and constructs a comprehensive enterprise social network that includes all relevant corporate governance entities such as the board of directors, board of supervisors, management team, and shareholders involved in corporate governance, operations, and decision-making. Two methods are employed to measure social networks: first, calculating the centrality and structural holes in the network from the perspective of individual governance subjects, and then measuring enterprise network conditions according to individual data indicators; second, directly from the enterprise perspective, calculating the number of other enterprises connected through governance bodies such as directors, supervisors, and shareholders to represent the enterprise network in which the company is situated. Based on social network indicators extracted by these two methods and corporate risk-taking levels, this paper conducts an empirical study, uses information transparency and corporate governance quality as mediating variables for mechanism analysis, and combines executives’ financial backgrounds and industry high-tech attributes for heterogeneity analysis.

Marginal contribution: (1) Although many studies have explored the impact of social networks on corporate efficiency, innovation, performance, resource acquisition, and other aspects, research on the impact of social networks on corporate risk-taking levels is relatively limited. Therefore, this research applies social network theory to the study of corporate risk-taking, filling an academic gap in this field. (2) Most existing research on corporate risk-taking focuses on internal and external environments, capital structure, governance mechanisms, and management characteristics, while paying little attention to the micro-level impact of social networks. This study specifically examines the micro-characteristics of social network structures. (3) Currently, when building social networks, most of the literature focuses on a single type of corporate governance subject, while few studies integrate different governance subjects into the network simultaneously. Additionally, there are few empirical studies on this topic in China’s A-share market. This study addresses these research gaps.

2. Literature Review

2.1. Relevant Research on Corporate Risk-Taking Level

The level of corporate risk-taking is a critical component of strategic decision-making, reflecting the enterprise’s—particularly its management’s—risk preference and risk management capabilities [35,36,37]. By summarizing the literature, it is found that the factors affecting the level of corporate risk-taking are mainly concentrated at the social, corporate, and individual levels.

The social impact on the level of corporate risk-taking mainly includes the macro-environment, institutional systems, and cultural background. In the macro-environment, Mclean R.D. and Zhao M. [38] and Huang Liang [39] found that the macro-economic environment and the policies implemented by the government have a direct impact on enterprises’ risk-taking ability. When local governments increase economic growth targets, the risk-taking level of enterprises significantly increases. In terms of institutional factors, John et al. [40] proved that investor protection systems can effectively alleviate the principal–agent problem and prevent managers from pursuing personal interests at the expense of the enterprise’s overall interests, thus improving the enterprise’s risk-taking level. In terms of cultural background, Li et al. [41] believed that culture affects the risk-taking level of an enterprise by influencing management decisions and the formal institutional system.

At the enterprise level, we can study the impact on the level of risk commitment from the perspective of internal and external governance of the enterprise. In terms of equity structure, Koerniadi et al. [42] found that equity checks and balances can promote enterprises to invest in high-risk and high-return projects and improve the level of enterprise risk-taking. In terms of senior management incentives, effective senior management incentives can also improve the level of risk-taking. Kini and Williams [43] believed that internal incentives can promote executives to choose high-risk and high-return projects for promotion. In terms of external supervision, Liao et al. [44] found that as an independent external supervision mechanism, an independent audit can alleviate the principal–agent problem and thus improve the risk-taking level of the enterprise. In terms of board size, Harjoto et al. [45] and Ma Ning [46] found that in a larger board size, the stronger the heterogeneity among the directors, which helps make decisions from different backgrounds and perspectives, making decisions tend to be stable, thus reducing the risk-taking level of the enterprise.

At the individual level, the impact on the risk-taking level is mainly studied from the characteristics of management, including management’s character, background, age, and gender. Zhang et al. [47] proved that overconfident CEOs can improve an enterprise’s risk-taking ability. However, due to defensive motivations, these overconfident CEOs tend to avoid complex risk control measures, which consequently weakens the enterprise’s risk resistance. Wang and Zhang [14] demonstrated that when senior executives have an overseas background, exposure to a more open international capital environment can influence their cognitive abilities and behavioral choices, thereby enhancing the enterprise’s risk-taking level. When executives have financial background, enterprises can effectively ease financing constraints and optimize resource allocation, thus improving the level of risk-taking of enterprises.

2.2. Relevant Research on Corporate Social Networks

Social network theory is an important field for studying the relationship and interaction between individuals or organizations, emphasizing the key role of social relations in resource acquisition, information dissemination, and behavior selection [48,49,50]. With the development of sociology and anthropology, scholars began to explore how network structure affects individual behavior and group dynamics. White stressed that the structure of social relations is the key to understanding social behavior. The relationship between individuals is not isolated, but embedded in a larger network, which affects individual behavior, information flow and resource allocation. As research deepened, Burt [51] first proposed the theory of a “structural hole” in 1992, emphasizing that an individual’s position in a social network and the structure of their relationships significantly impact their access to information and resources. Burt believes that individuals in a “structural hole” can connect different social groups, obtain and control unique information and resources, and thus gain a competitive advantage. In addition to the “embeddedness” theory and the “structural hole” theory, social network-related theories include social capital theory and reputation hypothesis theory. Based on the above theory, scholars have studied the effect of social networks on enterprise efficiency [21,23], innovation [26,27,52], and risk-taking [24,53,54].

When building corporate social networks, the existing literature is often limited to analyzing the role of certain types of governance subjects, such as the board of directors [21,55] and shareholders [23,33]. Extending this approach, a few scholars expanded the network structure to build a “shareholder–director” network [26] and a “senior executives and directors” network [27].

In order to further study the impact mechanism of real corporate social network on corporate risk-taking level, the network structure needs to be further expanded. This paper will expand and construct an enterprise social network on the basis of the “directors’ supervision and management” network and include all the relevant corporate governance subjects such as corporate governance, operation, and decision-making, including the board of directors, the board of supervisors, the management, and the board of shareholders. At present, there are two common methods to measure the location of social networks: One is to calculate the number of individuals who work with other enterprises and the distance of individuals who do not work together in the network, starting from the individuals who are the subjects of governance, and then measure the enterprise network of the enterprise according to the personal data [21,32]. The second is to directly calculate the number of other enterprises connected by the governance bodies such as directors, supervisors, and shareholders from the perspective of the enterprise to represent the enterprise network in which the enterprise is located [27,56]. The two measurement methods have their own advantages and disadvantages. Therefore, this paper uses two measurement methods to calculate the corporate social network at the same time to ensure the stability and reliability of the results.

2.3. Relevant Research on Corporate Social Networks and Risk-Taking

The mechanism by which corporate social networks influence the level of corporate risk-taking has been a subject of both positive and negative scholarly investigation. From the perspective of negative effects, Dai Juanping and Zheng Xianlong [57] observed that while political connections provide various benefits to enterprises—such as easier access to financing and financial subsidies—the costs associated with maintaining these connections often outweigh the benefits, thereby suppressing corporate risk-taking levels. Similarly, He Linjie et al. [58] argued that occupying a structural hole position within a network makes it difficult to achieve collective control due to the dispersion of network members. In such cases, enterprises may lose effective mechanisms to curb opportunistic behaviors among partners.

On the other hand, the positive effects of corporate social networks on risk-taking have been recognized by a greater number of scholars. For instance, Chan et al. [59] demonstrated that when banks and enterprises maintain strong relationships, reduced information asymmetry between the two parties encourages enterprises to engage in greater risk-taking. Additionally, Du Shanchong et al. [34] found that in shareholder chain networks, the supervisory, resource-sharing, and informational benefits significantly enhance corporate risk-taking capabilities, particularly in highly competitive environments or under conditions of policy uncertainty.

Despite these studies, there remain two divergent views on the mechanism through which corporate social networks affect risk-taking levels. As such, further research is required to examine the specific impact of corporate social networks on risk-taking, particularly using a sample of listed companies. Moreover, most existing studies focus solely on the direct effects of social networks on corporate risk-taking, neglecting the exploration of potential underlying mechanisms. While the literature frequently highlights the resource heterogeneity that social networks bring to enterprises, it rarely delves into which specific resources or effects influence risk-taking levels. Furthermore, there is a lack of research examining the interplay between internal and external factors, as well as the mediating mechanisms that may influence risk-taking. To address these gaps, this paper aims to comprehensively analyze the impact of corporate social network structures on risk-taking levels and to uncover the internal relationships between these networks and corporate risk-taking behavior.

2.4. Commentary and Assessment

This chapter reviews the literature on key research elements, including corporate risk-taking, social networks, and the impact of social networks on corporate risk-taking. In summary, social network theory offers a novel perspective, emphasizing that enterprises can more effectively acquire information, share resources, and mitigate risks through interactions and connections with other organizations or individuals. These interactions enhance their ability to navigate uncertainty and improve their risk-taking capacity. The limitations of current research can be summarized as follows:

(1)

Construction of Corporate Social Networks: Existing studies on corporate social networks often focus on analyzing the role of a single type of governance entity, such as the board of directors or shareholders. This paper broadens the scope by constructing a “board–supervisor–executive” network, which incorporates the board of directors, board of supervisors, management team, and shareholders’ meetings to provide a more comprehensive framework for analyzing corporate social networks.

(2)

Theoretical Divergences in Mechanisms: There are ongoing theoretical debates regarding the mechanisms through which corporate social networks influence risk-taking levels. To address these gaps, this study first investigates the direct impact of social network characteristics on risk-taking levels. Secondly, it examines mediating mechanisms, such as the information transparency effect and the corporate governance effect. Through empirical analysis, this research aims to provide a more comprehensive understanding of the intrinsic mechanisms and economic consequences that link corporate social networks with risk-taking behavior.

3. Research Hypothesis

3.1. Corporate Social Network Centrality and Corporate Risk-Taking

Risk-taking in an enterprise is a resource-intensive activity that relies heavily on significant resource dependencies [60]. These required resources encompass various aspects, including investment objectives, technology, capital, land, and product sales channels [53,61]. Corporate social networks, which are composed of shareholders, directors, and high-level governance bodies, play a crucial role in supporting these activities. According to stakeholder theory, stakeholders such as corporate governance bodies are integral to the growth and development of enterprises. Within this context, corporate social networks serve as carriers of rich social capital and represent a vital resource that contributes to the healthy development of enterprises [62]. Through social networks, enterprises can acquire scarce resources at lower costs. A broader network increases the likelihood of gaining access to capital, information, resources, and business opportunities, thereby enhancing competitive advantages and promoting efficient production and operations.

The centrality of a corporate social network is a key metric used to evaluate the importance of an enterprise within the network. This metric reflects whether an enterprise occupies a “core” position in the network structure. From the perspective of the “collaborative governance view”, enterprises with high centrality typically possess strong capabilities to acquire and control various resources [17]. Analyzing this through the lens of “social capital theory”, higher centrality provides enterprises with significant advantages in enhancing their risk-taking ability through three primary effects: resource allocation, trust-based cooperation, and reputation incentives [23]. Enterprises at the core of the corporate social network are better positioned to access information and resources efficiently. Simultaneously, leveraging their advantages in information, resources, reputation, and control, these enterprises can effectively reduce information asymmetry among corporate governance bodies. This reduction in asymmetry strengthens corporate governance effectiveness, thereby improving the level of corporate risk-taking. Based on these insights, this paper proposes the following:

H1. 

The higher the centrality of corporate social network, the higher the risk-taking level.

3.2. Corporate Social Network Structural Holes and Corporate Risk-Taking

The structural hole in a corporate social network refers to a third party that acts as a bridge between individuals who are not directly connected within the network. Overall, this structure resembles a “hole” in the network [63]. In social network analysis, structural hole theory highlights the critical role of key positions within the network. According to the theory of weak tie advantage, structural holes provide network members with opportunities to establish weak connections with other members [64]. These weak connections are considered “bridges”, enabling the third party that links two unconnected network members to gain advantages in terms of information resources and control.

The informational advantages derived from structural holes can be categorized into three dimensions: access, preemption, and referral [22,65]. Access refers to the ability to obtain effective information through specific channels while minimizing costs. Preemption reflects the timeliness advantage that a network node possesses over others in acquiring information. Referral involves opportunities to be recommended for additional information. Consequently, network members positioned at these “bridge” locations can gather rich and diverse information from various circles. The more structural holes they occupy, the greater their advantages in information access and control, thereby enhancing the enterprise’s risk-taking capacity. Based on these insights, this paper proposes the following:

H2. 

The more abundant the structural holes occupied by enterprises in the social network, the higher their risk-taking propensity.

3.3. Corporate Social Network Connectivity and Corporate Risk-Taking

The linkage index constructed in this paper represents the number of other enterprises that a given enterprise connects to through governance bodies such as directors, supervisors, and shareholders. It reflects whether the enterprise occupies a “core” position within the corporate social network. A high degree of connectivity indicates significant overlaps in these roles across enterprises, influencing corporate governance, business relationships, and industry competition. Such enterprises serve as important nodes, enabling access to more social resources.

High connectivity impacts risk-taking through information resource advantages and competitive advantages. Overlapping governance roles interconnect enterprises, facilitating the flow of information and resources, which promotes knowledge transfer [25,66] and expands access to critical resources. Moreover, these enterprises are better positioned to obtain timely and accurate information for investment decisions, including insights into risks and opportunities. This enables them to seize opportunities, improve investment efficiency [67], and overcome resource constraints. By leveraging social capital within networks, such enterprises reduce uncertainties and enhance their risk-taking capacity. Based on these insights, this paper proposes the following:

H3. 

The higher the corporate social network connectivity, the higher the risk-taking level.

3.4. Intermediation of Information Transparency

Information transparency refers to the availability, relevance, and reliability of information during dissemination and sharing [68]. It ensures that relevant parties can easily access, understand, and evaluate information, enhancing trust, facilitating decision-making, and reducing uncertainties. Corporate social networks, as channels for acquiring and transmitting information, enable enterprises to share information in a timely and reliable manner, supported by the reputation mechanism. These networks improve transparency through information exchange [69] and by strengthening external stakeholder oversight [30]. This dual effect enhances trust, reduces uncertainty, and ultimately boosts the enterprise’s risk-taking capacity. Based on these insights, this paper proposes the following:

H4a. 

The centrality of corporate social networks can improve the transparency of corporate information and thus the level of risk-taking.

H4b. 

Abundance of corporate social network structural holes can improve the transparency of corporate information, thus improving the level of risk-taking.

H4c. 

Corporate social network connectivity can improve the transparency of corporate information and thus the level of risk-taking.

3.5. Intermediation of Corporate Governance Level

Corporate governance refers to the relationships and institutional arrangements among a company’s stakeholders, including the board of directors, management, and shareholders. Its core objective is to ensure transparency, effective risk management, and sustainable development [70]. Enterprises at the core of social networks can strengthen oversight, enhance investor trust [71], reduce information asymmetry, and improve corporate governance and risk-taking levels [72,73]. Internally, enterprises in central positions benefit from social capital and weak tie advantages, gaining access to informational, financial, and human capital resources. Externally, their strong engagement with stakeholders improves information exchange and oversight mechanisms, enhancing transparency and trust. This fosters effective supervision by investors and regulators [74], reduces non-compliance risks, and increases risk-taking capacity.

In summary, enterprises central to corporate social networks improve governance by reducing information asymmetry and moral hazards while strengthening supervision, thereby enhancing their risk-taking levels. Based on these insights, this paper proposes the following:

H5a. 

Corporate social network centrality can improve the level of corporate governance and thus the level of risk-taking.

H5b. 

An abundance of corporate social network structural holes can improve the level of corporate governance and thus the level of risk-taking.

H5c. 

Corporate social network connectivity can improve the level of corporate governance and hence the level of risk-taking.

3.6. Heterogeneity Analysis—Executives’ Background

According to the senior echelon theory, executives with a financial background typically demonstrate a stronger capacity for risk-taking, which grants them greater tolerance for risk in decision-making processes [75]. Their expertise in financial theories and extensive experience in investment and financing enable them to effectively access and utilize resources within corporate social networks, thereby significantly enhancing the enterprise’s level of risk-taking [76]. Specifically, their financial background equips them with superior risk identification and assessment skills, allowing for them to optimize risk management strategies. Moreover, they possess extensive financial network resources, which increase the likelihood of successful financing approvals and reduce financing constraints [77]. In addition, their deep understanding of laws and regulations ensures compliance during risk-taking activities, helping to mitigate legal risks [78]. Therefore, this paper argues that the financial background of executives plays a heterogeneous role in shaping the effectiveness of corporate social networks. Based on these insights, this paper proposes the following:

H6a. 

The improvement in corporate social network centrality on the level of corporate risk-taking will have heterogeneity characteristics due to the financial background of senior executives.

H6b. 

The improvement in corporate risk-taking level due to the enhancement of the richness of corporate social network structural holes will have heterogeneous characteristics due to the financial background of senior executives.

H6c. 

The improvement in corporate risk-taking level by the improvement in corporate social network connectivity will have heterogeneous characteristics due to the financial background of senior executives.

3.7. Heterogeneity Analysis—High-Tech Industry

High-tech enterprises are characterized by being knowledge-intensive, technology-intensive, and talent-intensive, with innovation serving as the core driver of their operations. Innovation and human capital hold significant importance for these enterprises. Positioned at the forefront of technological innovation, high-tech enterprises often operate in environments marked by high levels of uncertainty and market volatility in their products and services [79]. Consequently, compared to traditional enterprises, high-tech enterprises exhibit a stronger inclination to leverage social networks in order to access critical information and resources, thereby enhancing their level of risk-taking. Based on these insights, this paper proposes the following:

H7a. 

The improvement in corporate risk-taking level due to the enhancement of corporate social network centrality will be different depending on whether it is a high-tech enterprise.

H7b. 

The improvement in corporate risk-taking level due to the enhancement of the richness of the corporate social network structure will be different depending on whether it is a high-tech enterprise.

H7c. 

The improvement in corporate risk-taking level due to the improvement in corporate social network connectivity will be different depending on whether it is a high-tech enterprise.

4. Research and Design

4.1. Model Construction

In order to verify the effect of corporate social network structure on the level of risk-taking, this paper constructs the following benchmark regression model:

$$Ris{k}_{i,t}={\alpha }_0+{\alpha }_1\times Cen{1}_{i,t}+Control{s}_{i,t}+{\epsilon }_{i,t}$$

(1)

$$Ris{k}_{i,t}={\alpha }_0+{\alpha }_1\times C{I}_{i,t}+Control{s}_{i,t}+{\epsilon }_{i,t}$$

(2)

$$Ris{k}_{i,t}={\alpha }_0+{\alpha }_1\times Connec{t}_{i,t}+Control{s}_{i,t}+{\epsilon }_{i,t}$$

(3)

Among them, the dependent variables $Cen{1}_{i,t}$, $C{I}_{i,t}, \mathrm{and} Connec{t}_{i,t}$ represent the centrality index, structural holes index, and connectivity index of enterprise i in the social network in t year. $Control{s}_{i,t}$ are control variables, including characteristic variables of the enterprise (see

Table 1. Variable definition table.

Variable Type	Variable Name	Variable Symbol	Variable Definition
Dependent variable	Level of corporate risk commitment	Risk	Measured by the volatility of corporate return (Roa)
Independent variable	Centrality index	Cen1	Taking the average value of all directors’, supervisors’, and shareholders’ centrality indexes of each enterprise, and, using the principal component analysis method, constructing the network centrality index
	Structural holes index	CI	The difference between 1 and the limit is used to measure the richness of the structural holes.
	Linkage index	Connect	The number of other enterprises that the enterprise connects through governance bodies such as directors, supervisors, and shareholders, divided by 1000
Mediator variable	Transparency of information	KV	KV index, which is the influence coefficient of trading volume on the yield
	Level of corporate governance	Govern	The principal component analysis is carried out by selecting nine indicators from three levels: shareholders, board of directors, and incentive mechanism
Control variable	Company size	Size	Natural logarithm of annual total assets
	Asset/liability ratio	Lev	Total liabilities at year end/total assets at year end
	Net profit margin on total assets	ROA	Average net profit/total assets balance
	Cash flow ratio	Cashflow	Net cash flows from operating activities/total assets
	Operating income growth rate	Growth	Current year’s operating income/previous year’s operating income − 1
	Number of directors	Board	The number of board members is taken as a natural number.
	The largest shareholder holds the shares proportion	Top1	Number of shares held by the largest shareholder/total shares
	Equity balance degree	Balance1	Second largest shareholder/first largest shareholder
	Book-to-market ratio	BM	Book value/total market value
	Tobin q value	TobinQ	(Market value of outstanding shares + number of non-outstanding shares * net assets per share + carrying amount of liabilities)/total assets
	Length of incorporation of the company	FirmAge	Ln (Year of the Year–Year of Incorporation + 1)

for details). Each variable is calculated and explained in detail below.

In view of the problems such as overuse and endogenous bias in the traditional intermediate effect step-by-step test, this paper is based on the research recommendations of Jiangya [80], focusing on the core explanatory variables to be interpreted variables causal identification credibility and use the same method to identify the core explanatory variables and intermediate variables causal relationship, so as to correctly identify the impact mechanism. Based on the above analysis, it can be seen that the corporate social network structure is more and more core and compact, which will significantly improve the level of corporate risk-taking, and this conclusion is supported by a series of endogenous tests and robustness tests in the following discussion. Therefore, on the basis that Xue Jiao and Tian Gaoliang have confirmed the relationship between information transparency and corporate risk-taking [81,82], if the same identification framework can be used to detect that the structural characteristics of corporate social network will also affect information transparency, it can be verified that the impact channel proposed in this paper is effective. Similarly, on the basis that Li Hao and Hong Jinming have confirmed the relationship between corporate governance and risk-taking [83,84], we intend to verify whether corporate social network structure characteristics affect corporate governance. Based on the above analysis, the following mediation effect model is constructed:

$$K{v}_{i,t}={\alpha }_0+{\alpha }_1\times X_{i,t}+Control{s}_{i,t}+{\epsilon }_{i,t}$$

(4)

$$Gover{n}_{i,t}={\alpha }_0+{\alpha }_1\times X_{i,t}+Control{s}_{i,t}+{\epsilon }_{i,t}$$

(5)

4.2. Index Construction

4.2.1. Dependent Variable—Risk-Taking Level

The variable to be explained in this paper is the level of enterprise risk commitment. At present, we mainly use yield volatility, leverage, volatility of stock returns, and indicators that have significant impact on corporate decision-making to assess the level of corporate risk-taking. Considering the characteristics of China’s A-share market, such as high volatility, large number of retail investors, strong speculation, and great influence from government policies, this paper chooses the volatility of corporate return rate (Roa) as a measure of corporate risk-taking level; refer to John et al., as shown in Equation (6) [40,85]. In order to mitigate the impact of industry and cycle, the enterprise Roa is subtracted from the annual industry average to obtain Adj_Roa. Taking every 3 years (T years to t + 2 years) as an observation period, the standard deviation of Roa (Adj_Roa) after industry adjustment is calculated separately on a rolling basis.

$$Risk{1}_{i,t}=\sqrt{{\displaystyle \frac{1}{T-1}}{\displaystyle \sum }_{t=1}^T{\left(Adj\_R{o}_{i,t}-{\displaystyle \frac{1}{T}}{\displaystyle \sum }_{t=1}^{T}Adj\_Ro{a}_{i,t}\right)}^2丨\mathrm{T}=3}$$

(6)

The volatility of stock returns is used as a robustness test, specifically referring to Bernile et al. [20], as shown in Equation (2), using the logarithm of the annual standard deviation of daily individual stock returns to assess the level of corporate risk exposure.

$$Risk{2}_{i,t}=ln\left(250\times \sqrt{var\left(R_{i,t}\right)}\right)$$

(7)

4.2.2. Explanatory Variables—Network Indicators

At present, there are two common methods to measure the location of social networks: One is to calculate the number of individuals who work with other enterprises and the distance of individuals who do not work together in the network, starting from the individuals who are the subjects of governance, and then measure the enterprise network of the enterprise according to the personal data [21,32]. The second is to directly calculate the number of other enterprises connected by the governance bodies such as directors, supervisors, and shareholders from the perspective of the enterprise to represent the enterprise network in which the enterprise is located [27,56].

According to research by Xie Deren and Chen Yunsen [86], the corporate social network is defined as follows: If shareholders/directors/supervisors/senior management work in an enterprise at the same time, they are directly connected. When one or more shareholders/directors/supervisors/senior management of an enterprise work in other enterprises, these enterprises are connected through these people. This study involves both of the aforementioned approaches, and the corporate social network discussed is formed through the direct or indirect connections among shareholders, directors, supervisors, and senior management.

In this paper, the first method is used to measure the characteristics of corporate social networks with centrality and structural holes. The second method is used to directly measure the characteristics of the degree of connection between enterprises through the social network.

Centrality Index

In social networks, centrality can be divided into degree centrality, intermediate centrality, proximity centrality, and feature vector centrality. Referencing the works of Xie Deren and Chen Yunsen [87] as well as Peng Huatao [24], the method for calculating centrality is further elaborated. Degree centrality is used to measure the local breadth of the network position of an enterprise. A higher value indicates a stronger circle effect and richer accessible resources. Specifically, degree centrality is defined as

$$Degre{e}_i={\displaystyle \frac{{{\displaystyle \sum }}_j{X}_{ji}}{g-1}}$$

(8)

Among them, i represents a certain director, supervisor, or shareholder; j represents other directors, supervisors, or shareholders other than i in the current year; and $X_{ji}$ represents a network connection relationship. If i and j serve on at least one company’s board of directors, the value is 1; otherwise, it is 0. g is the total number of members of the board of directors, management, meeting shareholders, and supervisory board of a listed company in the current year. Since the number of members may vary from year to year, (g − 1) is used to eliminate scale differences.

Proximity centrality reflects the global depth of a node in a social network’s connected subnetwork, i.e., it indicates whether the node is closer to the center position in the subnetwork. The larger the index of proximity centrality, the more important and “core” the node is to the subnetwork, and the closer it is to other members of the subnetwork, so that it can obtain information and resources efficiently. The proximity center is

$$Closenes{s}_i={\displaystyle \frac{g-1}{{{\displaystyle \sum }}_{j=1}^{g}d\left(i,j\right)}}$$

(9)

Among them, $d\left(i,j\right)$ represents the shortest path length between directors, supervisors, or shareholders i and j.

Intermediary centrality is the degree of nodes’ intermediary in the social network, which reflects the “bridge” role of directors, supervisors, and shareholders in the network. The greater the degree of intermediary center, the stronger the intermediary role of the node in the network and the stronger the ability to transfer and control the flow of information. The degree of intermediary center is

$$Betweennes{s}_i={\displaystyle \frac{{{\displaystyle \sum }}_{j<k}{g}_{jk\left(n_i\right)}/g_{jk}}{\left(g-1\right)\left(g-2\right)/2}}$$

(10)

where $g_{jk}$ is the number of shortcuts (shortest paths) that must be taken when connecting directors, supervisors, senior executives, or shareholders i and k. $g_{jk\left(n_i\right)}$ is the number of i included in the shortcuts of directors, supervisors, senior executives, or shareholders j and k. ${\displaystyle \sum }_{j<k}{g}_{jk\left(n_i\right)}/g_{jk}$ represents the degree to which there is i in the shortcut of connecting all other nodes in the entire enterprise social network, and $\left(g-1\right)\left(g-2\right)/2$ is used to eliminate scale differences caused by years [88].

Feature vector centrality is an index to measure the influence of nodes in the network; it reflects the quality of the network relationship of nodes. A higher value indicates that the more “core” the node is in the network, the greater the resource advantage it obtains. The feature vector center degree is

$$Eigenvecto{r}_i={\displaystyle \frac{1}{\lambda }}{\displaystyle \sum }_j{b}_{ij}{E}_j$$

(11)

Eigenvector centrality is obtained by solving the “eigenvalue-eigenvector” equation $BE=\lambda E$. $b_{ij}$ represents the adjacency matrix. If directors, supervisors, senior management, or shareholders$i$ and $j$ serve in at least one same enterprise, then the value of $b_{ij}$ is 1; otherwise, it is 0. $\lambda$ is the largest eigenvalue of B, and $E_j$ is the eigenvalue of the centrality of director, supervisor, senior management, or shareholder j.

According to the above algorithm, the obtained centrality indicators belong to the individual level of the corporate governance entities. Referring to the methods of Zhang Yuming and Zhang Xinyue [32], as well as Chen Yunsen and Xie Deren [21], the mean and maximum values of the centrality indicators of all directors, supervisors, senior management personnel, and shareholders of each enterprise are taken as the measurement indicators at the enterprise level. Then, using the principal component analysis method, a comprehensive indicator of the four centrality indicators is constructed as the network centrality indicator [64]: Cen1 and Cen2 (for robustness tests).

Structural Holes Index

This paper draws lessons from Lazega and Burt [51], Chen Yunsen [22], Zhang Yuming, and Zhang Xinyue [32]. In this way, the difference between 1 and the limit is used to measure the richness of the structural holes. The limits are

$$C{I}_i=1-{\displaystyle \sum }_j{\left(p_{ij}+{\displaystyle \sum }_q{p}_{iq}{p}_{qj}\right)}^2$$

(12)

Among them, $p_{ij}$ represents the strength of the direct relationship between directors, supervisors, senior executives or shareholders i and j, and ${\displaystyle \sum }_q{p}_{iq}{p}_{qj}$ represents the strength of all connection relationships between directors, supervisors, senior executives, or shareholders i and j. The larger the CI, the richer the structural holes in the corporate social network. Then, the average value of the richness indicators of all directors, supervisors, senior executives, and shareholders of each enterprise is taken as the measurement indicator at the enterprise level.

Connectivity Index

This article refers to Yu Chenyang and He Liu [27]; regarding the approach, the construction of the connectivity index (Con) is the number of other enterprises connected by the corporate governance bodies such as directors, supervisors, and shareholders, and then divided by 1000; the dimensions are treated uniformly. In addition, from an enterprise perspective, the number of other enterprises that the enterprise connects through governance bodies such as directors, supervisors, and shareholders is calculated to represent the enterprise network in which the enterprise is located, and Connect2 is recalculated.

4.2.3. Others

Heterogeneity Variable

This paper selects FinBack and HighTech as the classification variables. FinBack is valued at 1 when someone from the Board High School has a financial background; otherwise, it is 0. Referring to the practices of Peng Hongxing et al. (2017) and He Xiaoyu and Qin Yong (2018), the hi-tech industry has finalized the hi-tech listed companies based on the provisions of the Organization for Economic Cooperation and Development (OECD) and in combination with the industry classification code of the Industry Classification Guide for Listed Companies [89,90]. High-tech industry enterprise value is 1; otherwise, it is 0.

Intermediate Variable

In this paper, information transparency (KV) and corporate governance level (Govern) are selected as intermediate variables. The measurement index of information transparency is the enterprise’s KV index, which is the influence coefficient of transaction volume on the yield rate. The higher the KV index, the lower the quality of information disclosure, i.e., the lower the information transparency [32]. By constructing a comprehensive evaluation index of corporate governance quality to measure the level of corporate governance, eight corporate governance indicators are selected: separation of duties between the chairman and the general manager, proportion of independent directors, proportion of board of directors, proportion of senior management, proportion of the largest shareholder, size of the board of directors and the board of supervisors, and the sum of the top three senior management salaries [90]; this is achieved using the principal component analysis method to construct the corporate governance level index Govern.

Control Variable

This paper refers to the research of Wang Li and Chen Yunsen [21,22,23]; concerning the enterprise level, we selected control variables that may have an impact on the level of risk-taking, including company size, asset/liability ratio, net profit margin on total assets, cash flow ratio, growth rate of operating income, number of directors, shareholding ratio of the largest shareholder, equity checks and balances, book-to-market ratio, Tobin Q value, and the company’s age of incorporation.

4.3. Dataset

The research sample of this paper is from the listed companies in Shanghai and Shenzhen A shares in China from 2010 to 2022. As the risk-taking level of the explained variable requires a three-year window data in the calculation, data from 2010 to 2022 were actually used. The corporate governance data and financial data were obtained from the databases CSMAR, iFind, and Wind, and the samples were filtered according to the following principles: (1) Enterprises with S, ST, and PT marks were excluded. (2) Enterprises listed in the financial industry were excluded. (3) Samples with obvious missing or obvious abnormal data were removed. (4) Data of directors, supervisors, senior management, and shareholders were manually processed; the name differences caused by different symbols, letter case, and corporate record habits were adjusted; and different individuals with the same name and surname were distinguished between. (5) In order to eliminate the influence of extreme values on the results, 1% Winsor tail reduction was applied to the continuous variables. The specific operation involves replacing the lowest 1% of values in the dataset with the value at the 1st percentile and replacing the highest 1% of values with the value at the 99th percentile. After screening, valid sample observations of 1810 enterprises were obtained.

First of all, the average value of Risk is 0.021 and the standard deviation is 0.022, which indicates that there is significant differentiation in risk preference among enterprises. The choice of different degrees of risk-taking by enterprises may be closely related to their position in the social network and their ability to access resources. Corresponding to this is the core feature of social network: the average value of the centrality (Cen1) is −0.026, which indicates that some enterprises are in the core position of resource flow and information exchange, while others are relatively marginal. The mean value of intermediate centrality (CI) is 0.136, which indicates that some enterprises act as a bridge to connect each node in the network. The average value of Connect is 0.163, which indicates that there is a significant difference in connection strength between enterprises. The specific procedure can be observed in

Table 2. Descriptive statistics.

VarName	Obs	Mean	SD	Min	Max	Skewness	Kurtosis
Risk	11,529	0.021	0.022	0.001	0.150	2.803	13.568
Cen1	11,529	−0.026	1.786	−2.549	5.541	1.171	3.534
CI	11,529	0.136	0.084	0.004	0.379	0.671	3.000
Connect	11,529	0.163	0.202	0.000	0.596	0.907	2.072
Size	11,529	22.814	1.375	20.287	26.739	0.558	2.991
Lev	11,529	0.485	0.198	0.069	0.893	−0.113	2.218
ROA	11,529	0.042	0.049	−0.151	0.215	0.065	5.763
Cashflow	11,529	0.050	0.066	−0.146	0.241	0.012	3.855
Growth	11,529	0.159	0.344	−0.487	2.062	2.622	14.382
Board	11,529	2.221	0.173	1.792	2.708	0.341	4.019
Top1	11,529	35.733	15.401	8.734	75.002	0.414	2.468
Balance1	11,529	0.326	0.287	0.008	0.986	0.761	2.328
BM	11,529	0.682	0.260	0.143	1.225	−0.085	2.181
TobinQ	11,529	1.818	1.094	0.816	7.009	2.447	10.103
FirmAge	11,529	2.918	0.340	1.609	3.526	−1.116	4.848
KV	11,529	0.490	0.199	0.017	8.065	2.719	11.289
Govern	11,529	0.598	0.800	−2.254	2.610	−0.647	3.193

.

5. Empirical Analysis

5.1. Descriptive Statistics

The distribution characteristics of the main variables are summarized using descriptive statistical analysis (see Table 2). The sample comprises 11,529 observations, encompassing multidimensional data from A-share listed companies in China.

First, the average value of Risk is 0.021, with a standard deviation of 0.022, indicating notable variation in enterprises’ risk preferences. The varying degrees of risk-taking among enterprises appear to be closely associated with their position within the social network and their ability to access resources. Consistent with this, the key features of social networks exhibit significant differentiation. Specifically, the average value of centrality (Cen1) is −0.026, suggesting that some enterprises occupy core positions in resource flow and information exchange, while others remain relatively peripheral. The mean intermediate centrality (CI) is 0.136, highlighting the bridging roles some enterprises play in connecting nodes within the network. Meanwhile, the average value of Connect stands at 0.163, revealing substantial differences in the degree of connection strength among enterprises.

Turning to the control variables, the fundamental characteristics of the enterprises are also markedly heterogeneous. The average enterprise size (Size) is 22.814, indicating that while the sample consists predominantly of large enterprises, there remains considerable variation in firm size. The mean asset/liability ratio (Lev) is 0.485, paired with an average return on assets (ROA) of 0.042, reflecting differences in firms’ capital structures and profitability. The mean cashflow is 0.050 and the average growth is 0.159, pointing to significant disparities in liquidity and growth rates across firms.

Additionally, the distributions of corporate governance variables show substantial variation. The board size (BOD), equity concentration (Top1), and equity balance (Balance1) demonstrate diverse corporate governance structures. Meanwhile, the book-to-market ratio (BM) and Tobin Q values illustrate varying levels of market performance. Finally, FirmAge indicates the participation of both newly established firms and long-standing companies, encompassing a wide range of corporate lifespans in the sample. The variable KV has a mean value of 0.490, with a standard deviation of 0.199, indicating moderate variation among enterprises. Similarly, the mean value of Govern is 0.598, with a standard deviation of 0.800, reflecting noticeable differences in governance levels across firms.

5.2. Multiple Regression Analysis

According to the information provided in

Table 3. Corporate social network structure and risk-taking level: benchmark estimates.

	(1)	(2)	(3)	(4)	(5)	(6)
	Risk	Risk	Risk	Risk	Risk	Risk
Cen1	0.001 **	0.001 ***
	(0.000)	(0.000)
CI			0.012 **	0.015 ***
			(0.006)	(0.005)
Connect					0.004 **	0.005 ***
					(0.002)	(0.002)
Size		−0.002 **		−0.002 **		−0.002 **
		(0.001)		(0.001)		(0.001)
Lev		0.013 ***		0.013 ***		0.013 ***
		(0.004)		(0.004)		(0.004)
ROA		−0.093 ***		−0.093 ***		−0.093 ***
		(0.012)		(0.012)		(0.012)
Cashflow		0.011 **		0.011 **		0.011 **
		(0.004)		(0.005)		(0.004)
Growth		−0.000		−0.000		−0.000
		(0.001)		(0.001)		(0.001)
Board		−0.001		−0.001		−0.001
		(0.003)		(0.003)		(0.003)
Top1		−0.000 **		−0.000 **		−0.000 **
		(0.000)		(0.000)		(0.000)
Balance1		0.002		0.002		0.002
		(0.002)		(0.002)		(0.002)
BM		−0.008 ***		−0.008 ***		−0.008 ***
		(0.003)		(0.003)		(0.003)
TobinQ		0.000		0.000		0.000
		(0.001)		(0.001)		(0.001)
FirmAge		−0.008 *		−0.007		−0.008
		(0.005)		(0.005)		(0.005)
_cons	0.021 ***	0.096 ***	0.020 ***	0.093 ***	0.021 ***	0.096 ***
	(0.000)	(0.026)	(0.001)	(0.026)	(0.000)	(0.026)
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
N	11,815	11,529	11,815	11,529	11,815	11,529
r2_a	0.287	0.320	0.287	0.320	0.287	0.320

Note: Robust standard errors are reported in parentheses. *, **, and *** indicate significance at the 1%, 5%, and 10% levels, respectively. The same applies below.

, columns (2), (4), and (6) in the table take the Risk of the enterprise risk-taking level as the dependent variable, and the estimation coefficients of the core explanatory variables, namely, Cen1, ci, and Connect, are 0.001, 0.015, and 0.005 respectively, which all pass the significance level test of 1%, indicating that the centrality index, structural holes index, and linkage index of the enterprise in the enterprise social network have significant positive promotion effect on the enterprise risk-taking level, which is in line with the expectations of hypotheses 1, 2, and 3. This paper argues that corporate social networks provide enterprises with access to resources, information, and support by connecting with other organizations or individuals. Centrality, structural holes, and connection degree, as important indicators of network location, promote the risk-taking ability of enterprises from different dimensions. Higher enterprise centrality reflects that the enterprise occupies the core position in the network, can obtain market information and resources more quickly, reduce the uncertainty in decision-making, and at the same time improve confidence in risk-taking by enhancing trust and cooperation. Enterprises with a high structural holes index act as a bridge in the network and can obtain heterogeneous information and control the flow of information, thus occupying an active position in the competition and enhancing their ability to cope with risks. Enterprises with high connectivity have a wide range of network connectivity nodes, which can help enterprises spread risks and reduce the cost of risk exposure through resource sharing.

5.3. Endogeneity Test

This paper adopts the following two methods to deal with potential endogenous problems: First, this paper selects the Cen1 and Connect of other companies in the same industry in the same year as the tool variables for 2SLS regression to solve the endogenous processing. This selection is based on the following theoretical foundations: From a correlation perspective, companies in the same industry face similar market environments, technological characteristics, and regulatory policies, resulting in significant correlation in their social network characteristics, which has been verified through empirical testing. The regression results are shown in

Table 4. Endogeneity test: tool variables—industry average.

	(1)	(2)	(1)	(2)
	Cen1	Risk	Connect	Risk
Cen1		0.004 ***
		(0.001)
Cen1_mean	0.708 ***
	(0.052)
Connect				0.037 ***
				(0.011)
Connect_mean			0.801 ***
			(0.061)
Controls	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
N	10,041	10,041	10,041	10,041
Cragg–Donald Wald F statistic	182.852		172.820

. From columns (2) and (4) of Table 4, the Cen1_mean sum is significantly positive at the level of 1%, thus meeting the relevance principle. Moreover, there is no evidence suggesting that the internal composition of other companies in the industry would affect the risk-taking level of the company in question, thus satisfying the exogeneity principle.

Secondly, the network centrality index, structural holes index, and connection degree index, which lag one stage, are taken as the tool variables, respectively. Lagged variables typically satisfy the correlation and exogeneity conditions. The lagged dependent variable is usually highly correlated with the current dependent variable because executive networks between companies cannot change frequently in the short term, exhibiting autocorrelation. Additionally, data lag typically occurs before the current time period. If the current period’s error term is not correlated with the previous period’s variables, then the lagged variable can be considered exogenous. According to the information provided in

Table 5. Endogeneity Test: Tool Variables–Lag Variables.

	(1)	(2)	(1)	(2)	(3)	(4)
	Cen1	Risk	CI	Risk	Connect	Risk
Cen1		0.001 ***
		(0.000)
L. Cen1	0.606 ***
	(0.009)
CI				0.031 ***
				(0.010)
L.CI			0.434 ***
			(0.009)
Connect						0.007 ***
						(0.002)
L. Connect					0.604 ***
					(0.009)
Controls	YES	YES	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
N	10,302	10,302	10,302	10,302	10,302	10,302
Cragg–Donald Wald F statistic	4803.643		4429.435		4803.643

, in the first-stage regression of instrumental variable method, the F values are all greater than 10. In the second-stage regression of the instrumental variable, the estimated coefficients of Cen1, CI, and Connect are significantly positive at the 1% level, which is consistent with the results of the benchmark regression.

5.4. Robustness Test

In order to verify the robustness of the model conclusion, the robustness test of variable substitution is carried out on the model, i.e., whether the estimation result is consistent with the main regression conclusion is examined by changing the definition of independent variable or dependent variable. We replaced the core independent variables with other variables with similar meanings or alternative indicators, learnt from the practice of Bernile et al. [90]. In particular, the volatility of stock returns as an alternative cause variable was selected as the robustness test; specifically, we used the logarithmic value of the annual standard deviation of daily stock returns to assess the level of corporate risk commitment. At the same time, the independent variables were changed, the maximum value of all directors’ height and shareholders’ centrality index of each enterprise was taken as the measurement index at the enterprise level, and then the principal component analysis method was used to construct a comprehensive index of four centrality indexes as the network centrality index Cen2. At the same time, we used the enterprise perspective to calculate the number of other enterprises that the enterprise connects through governance bodies such as directors, supervisors, and shareholders to represent the enterprise network in which the enterprise is located, and recalculated Connect2 and CI2. Looking at

Table 6. Robustness test: replace y.

	(1)	(2)	(3)
	total_risk_day	total_risk_day	total_risk_day
Cen1	0.006 ***
	(0.002)
CI		0.081 *
		(0.046)
Connect			0.034 **
			(0.015)
Controls
Firm FE	YES	YES	YES
Year FE	YES	YES	YES
N	11,592	11,592	11,592
r2_a	0.624	0.624	0.624

and

Table 7. Robustness test: replace x.

	(1)	(2)	(3)
	Risk	Risk	Risk
Cen2	0.001 ***
	(0.000)
Connect2		0.078 *
		(0.042)
CI2			0.005 ***
			(0.002)
Controls	YES	YES	YES
Firm FE	YES	YES	YES
Year FE	YES	YES	YES
N	11,529	11,529	11,529
r2_a	0.320	0.317	0.320

, it can be seen that the conclusion is consistent with the benchmark regression, and the conclusion remains stable.

5.5. Mechanism Analysis

5.5.1. Information Transparency Effect

As shown in

Table 8. Mechanism analysis: information transparency effect.

	(1)	(2)	(3)
	KV	KV	KV
Cen1	−0.020 ***
	(0.001)
CI		−0.077 **
		(0.034)
Connect			−0.134 ***
			(0.011)
Controls	YES	YES	YES
Firm FE	YES	YES	YES
Year FE	YES	YES	YES
N	10,997	10,997	10,997
r2_a	0.470	0.457	0.466

, the information transparency index in columns (1)–(3) and the coefficient of the characteristic value of the corporate social network structure are both significantly negative. This empirical result indicates that the more central the corporate position in the social network is, the better the connectivity is, the higher the information transparency is, and the more adequate the information disclosure is. Specifically, when an enterprise is at the core of the social network and has high connectivity, it can significantly improve its information disclosure quality. This is mainly reflected in the following aspects: First, enterprises with core position and high connectivity occupy key nodes in the social network, which can not only efficiently obtain diversified information, but also reduce distortion and delay in the information transmission process, thus improving the accuracy and timeliness of information disclosure. Secondly, these enterprises usually have high reputation capital and social trust. In order to maintain their reputation in the network, they will pay more attention to the quality of information disclosure and take the initiative to ensure the authenticity and transparency of information. Finally, external regulators and stakeholders have higher expectations for the information transparency of the core enterprises. This external pressure further pushes the enterprises to continuously improve the information disclosure standards. Through the above mechanism, the core enterprise has finally achieved the improvement of risk management level and optimization of market efficiency.

5.5.2. Corporate Governance Effect

As shown in

Table 9. Mechanism analysis: corporate governance effect.

	(1)	(2)	(3)
	Govern	Govern	Govern
Cen1	0.007 ***
	(0.003)
CI		0.605 ***
		(0.068)
Connect			0.034 *
			(0.020)
Controls	YES	YES	YES
Firm FE	YES	YES	YES
Year FE	YES	YES	YES
N	11,045	11,045	11,045
r2_a	0.952	0.953	0.952

, the coefficients of corporate governance level and corporate social network structure characteristic values in columns (1)–(3) are both significantly positive. This empirical result indicates that the more centrality and better connectivity an enterprise has in the social network, the more perfect its corporate governance mechanism will be. Through in-depth analysis, it was found that the positive impact of network core location on corporate governance is mainly achieved through the following ways: Firstly, core location gives enterprises a strong ability to acquire and allocate resources, which enables them to integrate various resources more efficiently and provide solid support for improving the governance mechanism. Secondly, high-centrality enterprises, relying on their network advantages, are more likely to build solid trust ties and strategic cooperation relationships, which not only effectively reduce the degree of information asymmetry, but also significantly inhibit the generation of moral hazard, thus improving the overall governance efficiency. Thirdly, in order to safeguard the reputation capital and market position accumulated in the network, core enterprises will often actively strengthen the awareness of compliance; strictly abide by laws, regulations, and industry standards; and continuously improve the quality and transparency of information disclosure. In addition, the high standard expectation and supervision pressure from external regulators and investors, as well as the enterprise’s own sound internal governance structure and efficient information processing mechanism, promote the core enterprises to continuously optimize governance decisions. Through the synergy of the above-mentioned multiple mechanisms, the core enterprise finally achieved a significant improvement in the governance level and promoted the governance efficiency and market efficiency to a higher level.

5.6. Heterogeneity Analysis

5.6.1. Management Team Characteristics

In this paper, the financial background of the directors is used as the basis for the classification of senior management characteristics, and the corresponding regression results are reported in

Table 10. Heterogeneity analysis: top management background.

	(1)	(2)	(3)	(4)	(5)	(6)
	FinBack = 1 Risk	FinBack = 0 Risk	FinBack = 1 Risk	FinBack = 0 Risk	FinBack = 1 Risk	FinBack = 0 Risk
Cen1	0.001 ***	0.000
	(0.000)	(0.001)
CI			0.014 **	0.012
			(0.006)	(0.011)
Connect					0.005 **	0.004
					(0.002)	(0.004)
Controls	YES	YES	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
N	8480	2891	8480	2891	8480	2891
r2_a	0.337	0.438	0.336	0.439	0.336	0.439

. Among them, (1), (3), and (5) are listed as samples of enterprises with financial background and (2), (4), and (6) are listed as samples of enterprises without financial background. The empirical results show that among the companies with a strong financial background, the estimation coefficients of Cen1, CI, and Connect are all significantly positive. In contrast, these estimates are not statistically significant in the sample of firms with no financial background.

This result reveals that executives with financial backgrounds play a key regulatory role between corporate social networks and risk-taking, which is manifested in the following aspects: firstly, executives with financial backgrounds can identify and assess risks more accurately by virtue of their professional knowledge advantages, thus formulating more scientific risk management strategies and playing an active role in the information acquisition process of social networks; secondly, such executives usually have extensive financial network, and have a deep understanding of the information demand characteristics of financial institutions such as banks, which will help enterprises to improve financing efficiency and ease financing constraints; finally, their professional understanding of financial laws and regulations can help enterprises to better grasp the opportunities in the risk-taking process and realize the optimal balance of risks and benefits. Therefore, when an enterprise has directors with financial background, it can make more effective use of its position advantages in the social network and give full play to the value of network resources, so as to make the network characteristics have a significant positive impact on the level of enterprise risk-taking.

5.6.2. Industry Characteristics

Based on whether the industry the enterprise belongs to is a high-tech industry, the corresponding regression results are reported in

Table 11. Heterogeneity analysis: industry attributes.

	(1)	(2)	(3)	(4)	(5)	(6)
	HighTech = 1 Risk	HighTech = 0 Risk	HighTech = 1 Risk	HighTech = 0 Risk	HighTech = 1 Risk	HighTech = 0 Risk
Cen1	0.001 **	0.000
	(0.000)	(0.000)
Connect			0.007 **	0.001
			(0.003)	(0.003)
CI					0.017 **	0.013 **
					(0.008)	(0.007)
Controls	YES	YES	YES	YES	YES	YES
Firm FE	YES	YES	YES	YES	YES	YES
Year FE	YES	YES	YES	YES	YES	YES
N	5708	5800	5708	5800	5708	5800
r2_a	0.314	0.354	0.314	0.353	0.313	0.354

. Among them, items (1), (3), and (5) are listed as samples of enterprises in high-tech industries and items (2), (4), and (6) are listed as samples of enterprises in non-high-tech industries. The empirical results show that in the high-tech enterprises, the estimation coefficients of Cen1, CI, and Connect are all significantly positive. In the sample of non-high-tech enterprises, only Connect shows significance, and the estimation results of other network characteristic indexes do not have statistical significance.

This differentiated result reveals the moderating effect of industry attributes in the relationship between corporate social network and risk-taking, and its internal mechanism can be understood from the following aspects: First, high-tech enterprises have obvious knowledge-intensive, technology-intensive, and talent-intensive characteristics, with technological innovation at the core competitiveness, which makes them face a more complex business environment and higher uncertainty. Second, as they are at the forefront of technological innovation, high-tech enterprises often need to continue to invest in large-scale research and development, taking on higher risks of innovation and market fluctuations. Finally, in order to meet these challenges, high-tech enterprises rely more on access to key information, technological resources, and market opportunities through social networks than traditional enterprises, thus enhancing their risk-taking capacity and innovation efficiency. Therefore, high-tech enterprises can make full use of their advantageous position in the social network to transform network resources into risk management capabilities, which explains why network characteristics have a more significant impact on the level of high-tech enterprises’ risk commitment.

6. Conclusions and Recommendations

6.1. Conclusions

The complicated market environment brings unprecedented pressure to the business decisions of enterprises. How to effectively utilize existing social resources to establish a risk prevention mechanism and accurately assess an enterprise’s risk-taking ability has become a core issue related to the survival and development of enterprises. Based on a rich sample of data from 1810 listed companies on the Shanghai and Shenzhen A-shares market in China from 2010 to 2022, this study constructs a comprehensive corporate social network from multiple perspectives, overcoming the limitations of previous research that focused only on a single governance entity. Specifically, this study builds an all-encompassing corporate social network from the perspective of personnel connections among governance entities such as the shareholders’ meeting, board of directors, board of supervisors, and management. Various indicators are used to measure network characteristics: centrality reflects the core position of a company in the network, structural holes represent the company’s ability to act as a bridge between unconnected individuals in the network, and connectivity measures the breadth of connections a company has with other companies through governance entities such as directors, supervisors, and shareholders. Through rigorous empirical analysis, this study thoroughly examines the impact pathways and transmission mechanisms of these network characteristic values on corporate risk-taking levels. Additionally, heterogeneity analyses are conducted by incorporating executives’ financial backgrounds and the high-tech attributes of industries, revealing the intrinsic patterns of how corporate social networks influence risk-taking.

This study found that the enhancement of corporate social network centrality, structural holes, and connection degree can promote the improvement of risk-taking level, and this effect still exists after endogeneity and robustness tests. The impact is heterogeneous in terms of senior executives’ financial background and industry attributes. The mechanism test results show that enhancing the transparency of corporate information and improving corporate governance are two important transmission paths for social network connections to promote enterprises to improve their risk tolerance.

6.2. Recommendations

Based on the above research findings, this paper puts forward the following policy recommendations:

First, enterprises should fully realize the strategic value of social network location and formulate network development strategies accordingly. Enterprises located at the edge of the network should actively expand their ties with other enterprises and stride towards the core position of the network through various channels. Enterprises that are already at the core of the network should give full play to their pivotal role in the network, effectively integrate and utilize network resources, and promote their own risk management capabilities while driving the coordinated development of the entire network.

Second, enterprises should attach importance to the professional background composition of the senior management team. In view of the fact that executives with financial background can better play the advantages of network resources, enterprises should pay attention to the formation of diversified management teams when selecting and appointing talents, especially the introduction and training of executives with financial professional background. At the same time, we should give full play to the professional advantages of these senior executives in risk identification, resource acquisition, and strategic decision-making to provide intellectual support for the steady development of the enterprise.

Third, policy-making should fully consider the differences in industry attributes. This study found that network structure plays a more significant role in promoting the risk-taking of high-tech enterprises, which indicates that the regulatory authorities should implement policies in accordance with their own industry when formulating relevant policies. For high-tech enterprises, policy guidance can be used to support them in building a broader social network and help them to access innovative resources and risk management experience. For enterprises in traditional industries, it is necessary to guide them to make rational use of network resources and moderately increase the level of risk-taking on the basis of sound operation.

Fourth, efforts should be made to improve the information disclosure mechanism and corporate governance system. Enterprises should make full use of the information advantages brought by social networks to continuously improve the quality of information disclosure and enhance the transparency of operations. At the same time, it is necessary to transform network resources into governance effectiveness, continuously improve the corporate governance structure and enhance the governance level by absorbing advanced governance experience and introducing diversified supervision mechanisms.

6.3. Discussion and Limitation

The limitations of this article and directions for further research are summarized as follows: (1) This study is limited to a single perspective, focusing on Chinese stock market. If data comparability allows, it would be beneficial to include comparisons with developed countries and other emerging economies to further explore the driving factors behind the impact. Thereby, this would verify the applicability of the conclusions under different regulatory, ownership, and investor characteristics. (2) The current research primarily relies on ROA volatility and stock return volatility to measure risk-taking. Risk-taking is a series of processes involving enterprise capabilities, behaviors, willingness, and decision-making. When measuring the level of corporate risk-taking, it needs to be examined from a comprehensive perspective, especially by selecting different indicators for different types of enterprises. Future research could incorporate additional risk proxy variables to provide a more comprehensive view of corporate risk behavior, such as investment aggressiveness, R&D expenditure, or leverage-based risk indicators. (3) In constructing the corporate social network, this study only selects data from China A-share enterprises over the past 13 years. Future research could extend the time period and expand the sample size. Furthermore, social networks could be constructed separately according to different enterprise types to more comprehensively explore the role of social networks on enterprises.