The Impact of Corporate Social Network on Innovation: A Mediation Analysis of Agency Costs and Financial Constraints

This paper examines the impacts of corporate social network on innovation activities (both R&D inputs and patent outputs) and explores the potential influence path of social networks driving corporate innovation. Using Chinese share markets’ listed companies in the period from 2009 to 2016 as the sample, we establish a new type of social network based on the connections of the core management team (including directors, supervisors and executives) and find that a corporate social network based on the core management team has a positive causal effect on enhancing R&D investments and patent performance. Empirical results are robust after controlling for endogeneity and the causal relationship between social networks and corporate innovation is robustly monotonic. We perform a mediation analysis following the bootstrapping procedures, and find that alleviating financial constraints plays a mediating role in the process of a social network of this new type improving corporate innovation, while mitigating agency conflicts does not. Our conclusions demonstrate that, in the promotion of corporate innovation by the corporate social network, financial constraints have a more profound impact than agency costs.


Introduction
Corporate innovation ability is an important factor affecting a firm's value and rectors, supervisors and executives from the overall perspective of the corporate core management team, includes all directors, supervisors and executives in the social network construction, extends the "director/CEO network" to the "core management team network", and attempts to explore the impact of a larger and wider social network on corporate innovation (both R&D inputs and patent outputs). Because R&D investment is the material base of innovation activities, we first examine the relationship between social networks and R&D intensity.
Patent outputs are the direct results of R&D investment, so we further study the effects of social networks on patent performance. Social network centrality is the most important proxy used to measure the characteristics of a corporate social network. The greater the company's social network centrality is, the more numerous the company's social connections, the greater the network power, and the stronger the ability to benefit from the network. Therefore, we expect that companies with high social network centrality will have more active innovation practices, such as higher R&D investment and better patent performance. We hypothesize the following: H1: The corporate social network based on the core management team enhances corporate innovation.
H1a: Social network centrality is positively associated with R&D investment. H1b: Social network centrality is positively associated with patent performance.

Influence Path between the Social Network and Corporate Innovation
Corporate innovation is a long-term process and is filled with uncertainty. High R&D costs, continued risk-taking and uncertain future returns are the three pressures that companies must bear while trying to innovate. Compared with the other corporate practices, indeed, innovation activities need a great deal of money, have long development cycles and result in benefits after a few years. Liu et al. (2015) also emphasize that innovation practices are restricted by internal agency costs and external financial constraints [33].

Mediating Effects of Agency Costs
Due to the uncertainty of innovation results, R&D investment may cause agency problems between company owners and managers [1]. Holmstrom (1989) argues that larger firms are at a comparative disadvantage in conducting highly innovative research [34]. When venture capitalists eventually withdraw, their monitoring services are no longer as valuable and the company needs to bear its own monitoring costs, which is not good for corporate innovation. Jensen  and residual loss; to a certain extent [35], Holmstrom's (1989) research can be understood to imply that agency costs impede corporate innovation [34]. Using engineering employee data of high-technology companies, Zenger (1994) observes that small firms more efficiently offer performance contracts to attract and retain engineers with higher abilities and skills and that small firms more efficiently resolve severe agency problems in R&D, thus encouraging R&D activities [36]. Hall (2002) argues that in the principal-agent relationship, managers pay more attention to short-term benefits than do shareholders and avoid investing more resources in R&D projects [37]. Consequently, a firm with significant agency conflicts would invest less in R&D activities. Other studies also show that management ownership, concentrated ownership and effective monitoring can reduce the agency costs associated with innovative activities, and alleviating agency conflicts is beneficial to improving R&D inputs and patent outputs [1] [38]. Therefore, we argue that solving agency problems and reducing agency costs are effective channels of promoting corporate innovation.
A company's social network relationship has the effect of optimizing corporate governance. Ferris et al. (2017) note that the social network provides mechanisms for information sharing and punishment for reputation loss, and these mechanisms alleviate information asymmetry and agency problems [11]. Because managers fear the threat of reputation loss, social ties weaken managerial incentives to expropriate, and reduce agency costs. Chen (2012) focuses on the social network of independent directors and observes that network characteristics alleviate the type-I agency conflicts between executives and shareholders and the type-II agency conflicts between large shareholders and minority shareholders [39]. The phenomenon is that the higher the social network centrality is, the lower the two types of agency costs.
In summary, a corporate social network can help mitigate agency conflicts, and solving agency problems can effectively enhance corporate innovation. Therefore, we hypothesize that reducing agency costs is the influence path of the social network affecting corporate innovation, and agency costs play a mediating role between social network centrality and corporate innovation. Thus, we propose the following hypothesis: H2: The corporate social network based on the core management team enhances corporate innovation by means of the mediating effect of agency costs.
H2a: Agency costs mediate the relationship between social network centrality and R&D investment.
H2b: Agency costs mediate the relationship between social network centrality and patent performance. of R&D investment [40]. Due to the lack of collateral value, asymmetric information problems, the uncertainty of innovation outcomes and executives' adverse selection and moral hazard, it is difficult to obtain sufficient financial support from external financing channels for corporate R&D investment [2] [5] [37] [41] [42]. Consequently, firms' innovative practices often face severe financial constraints that cause a lack of R&D expenditures and impede innovation and its outputs. Guariglia and Liu (2014) indicate that in transition economies (e.g. China), financial constraints have long been regarded as the major obstacle for corporate innovation [43]. Zhang et al. (2017) empirically confirm that financial constraints significantly impede R&D investments by Chinese companies and that credit rent-seeking aggravates the restrictions of financial constraints on R&D investment [44]. The negative effect of financial constraints on R&D investment is especially prominent in smaller, younger, and low-payout firms. Li (2011) suggests that R&D-intensive firms are subject to more financial constraints because information asymmetry and agency problems are more severe for these firms, and the author confirms that a financially constrained R&Dintensive firm is more likely to suspend or discontinue R&D projects [4]. Financial market development would increase R&D investment by mitigating financial constraints in growing firms, which should spur innovation and lead to higher overall economic growth [2] [5]. Hsu et al. (2014) claim that a well-developed financial market contributes to reducing the financing costs, improving resource allocation efficiency, and then enhancing patent outputs [45]. In a capital market, equity financing and debt financing are the two main financing channels for companies. Based on this, Hsu et al. (2014) further examine the different impacts of equity market development and credit market development on innovation [45]. The researchers observe that companies that are more dependent on external finance exhibit a higher innovation level in countries with more developed equity markets. However, development of the credit market appears to discourage corporate innovation.
These studies document that alleviating the financial constraints of the equity market and optimizing the financing environment have significant positive effects on corporate innovation. In addition, Howell (2017) points out that receiving early-stage government R&D grants has a signaling effect that could ease a firm's financial constraints through a certification mechanism and have a positive impact on patenting [46]. Therefore, we believe that alleviating financial constraints is another effective channel of promoting corporate innovation.
The social network studies show that corporate social connections not only optimize corporate governance but also improve the corporate financing environment. A social network constructed by executives is a type of informal institutional arrangement and can effectively promote information exchange and resource sharing between the firms in the same network. Jin and Yu (2018) observe that the size of an executive network is negatively associated with financial constraints, and the executive network plays an important role in mitigating ex-  [10]. The authors suggest that the executive network can effectively alleviate financial constraints because executives can promote the sharing of resources by communicating and coordinating with members of the network, which reduces the transaction costs of financing and provides a guarantee for loans. Xu and Cao (2016) indicate that an independent director network helps firms obtain more external environment information, and directors with central network positions play better advisory roles; thus, social networks loosen the firms' financial constraints [47]. It is clear that whether a social network is an executive network or a director network, both rich social ties and central network position are beneficial to optimizing the corporate financing environment. Following the logic that the social network alleviates financial constraints and the notion that financial constraints impede corporate innovation, we hypothesize that alleviating financial constraints is an important influence path of social networks enhancing corporate innovation, and financial constraints play a mediating role between social network centrality and corporate innovation. Thus, we hypothesize the following: H3: The corporate social network based on the core management team enhances corporate innovation by means of the mediating effect of financial constraints.
H3a: Financial constraints mediate the relationship between social network centrality and R&D investment.
H3b: Financial constraints mediate the relationship between social network centrality and patent performance.
Our conceptual framework is shown in Figure 1

Data Sample
We use all Chinese share markets' listed companies in the period from 2009 to 2016 as our sample, and establish a social network based on the core management team's connections, which covers the whole share market in China. To address the question of whether the core management team network affects corporate innovation, we need the following four types of data: corporate management team data, corporate financial data, corporate patent data and corporate social network data. We collect data of the first three types from the China Stock Market and Accounting Research (CSMAR) database. If management team data is missing from the CSMAR database, we supplement it by using the CNINF website (http://www.cninfo.com.cn/) and the SINA Finance website (https://finance.sina.com.cn). Our sample consists of 19,757 firm-year observations, and we collect a total of 378,540 items of core management members' occupational information. The corporate social network data are calculated on the basis of corporate management team data. We manually match, collate and code the corporate management team data and then calculate the corporate social network data (i.e. network centrality) using the Ucinet 6 software.

Social Network Construction
In this paper, we define the core management members as all directors, supervisors and executives in the company. The social network connections are based on multiple appointments of corporate core management members. If a core management's member serves in two companies at the same time, we conclude that there is a social connection (also called a social tie or a social interlock) between the two companies [13] [28] [48]. The social network is composed of many unique social connections. Cai and Sevilir (2012) classify board connections into "first-degree connections" and "second-degree connections" [8]. The researchers indicate that a first-degree connection occurs if two firms share a common director, and a second-degree connection is present if one director of a firm and one director of another firm have been serving on the board of a third firm. Therefore, all social connections in this paper refer specifically to "first-degree connections". We extend the "director/CEO network" commonly considered in the existing literature to the "core management team network", which integrates all the social connections between directors, supervisors and executives among different companies and includes them in one network; thus, we can provide a wider social network analysis for Chinese listed companies' activities.

Dependent Variables: Corporate Innovation Measures
The prior studies of corporate innovation mainly develop two types of proxies: innovation input and innovation output. The former measures the firms' support for research and development, and R&D expenditure is a representative variable [30]. Innovation output is a measure of the research and development achievements, and captures the quality of innovation more accurately. The main proxy of innovation output is patenting activities, including patent applications and patent citations [49]. Because companies need to go through a strict review process to apply for and receive patents, the number of patent applications and citations are more effective variables for measuring the success of corporate innovation. Pakes (1985) emphasizes that patents are an output of R&D investments and R&D activities, and there is a high correlation between R&D inputs and patent outputs [50]. Much empirical evidence proves that R&D intensity is

Independent Variables: Social Network Centrality
Social network centralities (SocialNetwork) are important proxies used to measure the characteristics of a corporate social network, and mainly include degree centrality (Degree), closeness centrality (Closeness), betweenness centrality (Betweenness) and eigenvector centrality (Eigenvector). These four variables comprehensively measure the embeddedness characteristics of firms' nodes in a social network, and can better capture the firms' network status, power and importance. Degree is the number of links of each company divided by the number of companies in the network. Betweenness is the number of shortest paths linking any two companies in the network that pass through the firm's node. Closeness is the inverse of the average distance between the firm and all other firms in the network. Eigenvector is a measure of the relative importance of a firm's node in the network; this variable is the dominant eigenvector of the sociomatrix and is used in the network literature to measure the prestige of a company [17]. To calculate social network centralities, we manually match the corporate core management team data year-by-year and construct the annual "firm-firm" adjacency matrix (i.e. an element in the adjacency matrix equals to 1 if there is a social network connection between two firms, and equals to 0 otherwise) accord-  first, and then define the connection size as the number of connections plus 1 to take the natural logarithm (e.g. ln(1 + PLink) and ln(1 + FLink)). Finally, we also use ln(1 + PLink) and ln(1 + FLink) for the robustness checks. [58]. The management expenses ratio is positively associated with agency costs.

Mediator Variables
Following prior studies, we define MER as the ratio of management expenses to operating income.
Hadlock and Pierce (2010) construct a measure of financial constraints that is based solely on firm size and age, and is called the SA index [59]. This SA index is calculated as (−0.737 × Size) + (0.043 × Size 2 ) − (0.04 × Age), where Size equals the natural logarithm of total assets, and Age is the number of years the firm has been listed. Hao [62]. Hence, we also use the SA index (SA Index) to measure corporate financial constraints. It is important to note that the SA Index is always a negative number. A larger SA Index (in absolute value is smaller) indicates that a firm is faced with looser financial constraints and has a better financing environment. In contrast, a smaller SA Index (in absolute value is larger) indicates that a firm is faced with more stringent financial constraints and

Control Variables
Our selection of control variables follows the prior literature [30] [49] [63]. We introduce the following control variables: 1) Size is defined as the natural logarithm of total assets; 2) Lev is defined as the ratio of total liabilities to total assets; 3) ROE is defined as return on equity; 4) CapitalExp is defined as the ratio of capital expenditures to total assets; 5) TAR is defined as the ratio of tangible assets to total assets; 6) FCF is defined as free cash flow per share; 7) Sales is defined as operating income per share; 8) SGR is defined as the growth rate of operating income; 9) Age is defined as the number of years the firm has been listed, and we use ln(1 + Age) as the control variable.

Empirical Models
To address the impact of the corporate social network on corporate innovation, we estimate the following: where i denotes firm, k denotes industry, and t denotes time. The dependent variable (Innovation i,k,t ) represents firm i's corporate innovation in industry k in year t. We use the investment intensity of R&D (R&D) and patent performance (ln(1 + Patent), which consists of ln(1 + Apply), ln(1 + Grant) and ln(1 + Valid)) as proxies for corporate innovation. The independent variable (SocialNetwork i,t−1 ) denotes firm i's network centrality in year t − 1, which includes Degree, Closeness, Betweenness, Eigenvector, ln(1 + PLink) and ln(1 + FLink). Control variables (Ctrls i,t−1 ) contain firm characteristics that could affect a firm's innovation, as discussed in Section 3.3.4, and all control variables are lagged by one year. ν t and μ k represent year fixed effects and industry fixed effects, respectively, and ε i,k,t is the random error term. Following common practices [10]    R&D investment intensity (R&D) and patent performance (ln(1 + Apply)) are significantly positive, which suggests that a high level of social network centrality can improve the firms' R&D inputs and patent outputs. These findings confirm the hypothesis H1 that a company with more network connections and higher network centrality has more funds to invest in R&D activities and has better patent performance.

Endogenous Issues
For the regression of social network measures on corporate innovation, there may be some endogenous issues, such as reverse causality and correlated omitted variables. To address the causal relationship of whether the corporate social network affects innovation investment and innovation performance, we use a lagged variable regression and two-stage instrumental variable least squares regression (IV-2SLS) to control for potential endogeneity.   There is a possibility that a firm with better innovation performance can attract other firms to establish social network connections with it. In other words, there is a reverse causality issue between social network centrality and corporate innovation. In the benchmark regression, considering the time lag of corporate innovation, we examine the impact of the social network on R&D inputs and patent outputs by using one year lagged network centrality. In theory, the subsequent events cannot affect the previous events, and then considering a multiperiod lagged variable regression can control for potential reverse causality [28].
To further explore the multiperiod impact of social network centrality on cor-  Table 4 show that Degree, which is lagged by two or three years, is still significantly positively associated with innovation investment (R&D) and innovation performance (ln(1 + Apply)), which indicates that the impact of social network connections on corporate innovation is not changed by reverse causality.   firms with older management groups may be more conservative in their investment decisions [17]. Older executives are primarily interested in short-term performance instead of long-term R&D projects, and have some passive sentiment in their innovation strategy. Therefore, the average age of the core management team's members may be negatively associated with corporate innovation. In other words, the core management team's characteristics could be the omitted variables in innovation regression. We define the number of core management members (ManagerNum) as the total number of all directors, supervisors and executives in the company, and define the average age of core management members (ManagerAvgAge) as the average age of all directors, supervisors and executives. Because the number and average age of core management members have skewed distributions, we add the natural logarithm of ManagerNum and ManagerAvgAge (e.g. ln(ManagerNum) and ln(ManagerAvgAge)) into our regression. Furthermore, in China each region has different industrial support policies for local firms. These regional characteristics are difficult to observe accurately in general. To address the potential endogeneity arising from unobservable regional heterogeneity, we control for region fixed effects in our benchmark model. Regressions with region fixed effects and core management team's characteristics are reported in Table 5. These results show that the coefficients indicating the influence of social network centrality (Degree) on R&D investment intensity (R&D) and patent performance (ln(1 + Apply)) are still positive and significant after controlling for the region fixed effects and core management team's characteristics. As expected, ln(ManagerNum) is positively associated with R&D and ln(1 + Apply), and ln(ManagerAvgAge) is negatively associated with R&D and ln(1 + Apply). Our empirical results have not changed because of regional features and core management team's characteristics.
To address the potential endogeneity arising from other unobservable hetero-   The IV-2SLS regression results of social network centrality on corporate innovation are shown in Table 6. Instrumental variables avgRegionDegree and ln(FirmNumRegion) have passed the underidentification test, weak identification test and overidentification test, suggesting that our instrumental variables are both relevant and effective. In the 1-stage regression, avgRegionDegree and ln(FirmNumRegion) are positively associated with Degree, which means that a firm with higher other firms' average social network centrality and a larger number of firms in the same region has a larger social network centrality. In the 2-stage regression, the instrumented Degree has a significant positive impact on R&D inputs (R&D) and patent outputs (ln(1 + Apply)), as before. The IV-2SLS regression results are consistent with the benchmark regression (Table 3). Therefore, we believe that there is a causal relationship between social networks and corporate innovation. Specifically, a company with a central position in a social network has higher innovation investment and better innovation performance.

Mediation Analyses
To examine the mediating effects of agency costs and financial constraints in the process of social networks promoting corporate innovation, we perform a mediation analysis following the bootstrapping procedures, as elucidated by Preacher and Hayes (2008) Table 7. In Row 1 of Table 7, the indirect effect of the social network (Degree) improving R&D investment (R&D) through agency costs (MER) is not significant, with a point estimate of −0.0004 and two 95% confidence intervals including 0 (both the normal distribution-based confidence interval and the bias-corrected confidence interval). This indicates that agency costs do not mediate the relationship between the social network and R&D investment. In Row 2 of Table 7, the indirect effect of the social network (Degree) enhancing patent performance (ln(1 + Apply)) through agency costs (MER) is not robustly significant, with a point estimate of 0.0026 and a 95% normal distribution-based confidence interval including 0. Although the 95% bias-corrected confidence interval excludes 0, we judge that this result is not robust and agency costs do not play a significant mediating role between the social network and patent performance. Therefore, our empirical tests of the hypothesis H2 have failed. Agency costs do not mediate the relationship between the corporate social network and corporate innovation. Alleviating agency conflicts is not an influence path of the social network driving corporate innovation.
In Row 3 and Row 4 of Table 7, the indirect effects of social network (Degree) promoting corporate innovation (both R&D and ln(1 + Apply)) through financial constraints (SA Index) are robustly significant, with two 95% confidence intervals excluding 0 (both the normal distribution-based confidence interval and the bias-corrected confidence interval). These results indicate that financial constraints mediate the relationship between the social network and corporate innovation (both R&D inputs and patent outputs) significantly. To summarize, financial constraints have a real mediating effect in the process of the corporate social network enhancing corporate innovation, and mitigation of financial constraints is an important influence path of the social network promoting corporate innovation. Our empirical evidence supports the hypothesis H3 completely.

Are the Effects of Social Networks on Corporate Innovation
Monotonic? Fich and Shivdasani (2006) note that a busy board is formed when the majority of outside directors hold three or more directorships [69]. Because the directors' attention is dispersed, a busy board cannot execute the monitoring function very well. A firm with a busy board is correlated with a weak corporate governance and displays a lower firm value. The social network relationship in this paper is established by the core management members' multioverlap employment. It is worth further discussing whether too many corporate social connections impede corporate innovation and whether the effects of social networks on corporate innovation are monotonic.
We examine the inverted U-shaped relationship between the social network and corporate innovation following the method of Lind and Mehlum (2010) and  Table 8. First, we show that the data range of Degree is 0 to 0.6461 in our sample. Second, we test the significance of the direct and squared terms of Degree. Regardless of whether we consider the R&D regression or the ln(1 + Apply) regression, Degree is significantly positive and Degree 2 is significantly negative. Third, we test whether the slope at both ends of the data range is sufficiently steep. The slope at the minimum value of Degree is positive and significant in both the R&D regression and the ln(1 + Apply) regression, but the slope at the maximum value of Degree is negative and not significant. This indicates that the true relationship between the social network and corporate innovation may be merely one half of an inverted U-shape. Fourth, we test whether the extreme point (also called the turning point) is located well within the data range. The extreme point estimates of Degree in the R&D regression and the ln(1 + Apply) regression are 0.4774 and 0.4587, respectively. Then, we calculate the extreme point's confidence interval; results show that the upper bound of the 95% confidence interval of the extreme point based on the Fieller method, in both the R&D regression and the ln(1 + Apply) regression, is outside the Degree range. The confidence intervals confirm that only one half of the inverted U-shaped curve is revealed in our data range.
In summary, there is no significant inverted U-shaped relationship between the social network and corporate innovation. In other words, we have the left half of the inverted U-shaped curve in our data range, and the relationship between social network and corporate innovation is robustly monotonic. This means that we indirectly capture the robust positive impact of the social network on corporate innovation. Having too many social connections will not impede corporate innovation. The corporate social network has a monotonic positive effect in enhancing R&D investment and patent performance.

Robustness Checks of Changing Samples or Variables
We use four methods to perform robustness checks. First, we change the sample to re-estimate the above models. Considering the different attitudes of various

Conclusions and Implications
Under Our empirical results suggest that the corporate social network based on the core management team has a positive causal effect on enhancing R&D inputs and patent outputs, and the relationship between social networks and corporate innovation is robustly monotonic. A firm that has more social ties, higher network centrality, better connection quality, a shorter connection path and a stronger ability to control the shortest path, makes a greater innovation investment and has a stronger innovation performance. These results are robust after controlling for endogenous issues (including reverse causality and correlated omitted variables). The mediation analyses show that our newly constructed core management team network is beneficial to loosening financial constraints, and optimizing the financing environment is the key to improving corporate innova- The director network focuses on the "director-director" connections between different companies, and the CEO network focuses on the "CEO-CEO" connections. However, the relationship between a director of one company and a CEO of another company also contains incremental information. Oh and Barker (2018) find that when a CEO serves as an outside director in other companies, CEO imitates the R&D intensity of tied-to firms in their own firm's R&D decisions [29]. This indicates that the "CEO-outside director" connection provides  [74]. In our opinion, Yu and He (2019) ignore the connections between independent director and executive or supervisor, or the connections between supervisor and director or executive, so they do not draw a significant empirical conclusion. Consequently, we believe that integrating all the social connections between directors, supervisors and executives among different companies has theoretical and practical value for researches on corporate innovation from the perspective of social networks. In addition to the connections of "director-director" and "CEO-CEO", the connections of "director-executive", "director-supervisor", "executive-executive", "executive-supervisor" and "supervisor-supervisor" among different companies can also carry lots of information related to innovation. This paper constructs social network based on the above six kinds of network carriers ("CEO-CEO" connection is a kind of "executive-executive" connections) and includes all connections between directors, supervisors and executives among different companies. This new type of social network can dig and exhibit more valuable information between company connections, and provides a wider and deeper network structure analysis for Chinese listed companies' innovation activities.
Second, we explore two specific influence channels of how the social network based on the core management team drives corporate innovation, and successfully confirm a significant mediating path. To keep competitive advantages, enterprises spend a lot of human resources, material resources and financial resources to carry out innovation activities. Due to the opacity of R&D process between shareholders and managers, and the information asymmetry between inside and outside the companies, the internal agency problem and external financial constraints impede the promotion of innovation ability and corporate  [11]. Therefore, we integrate social networks, corporate innovation, agency costs and financial constraints into a unified analysis framework to empirically test whether social networks can influence corporate innovation through "agency costs channel" or "financial constraints channel". The results show that the corporate social network based on the core management team can significantly enhance corporate innovation, and financial constraints play a mediating role in the process of social networks driving innovation. By contrast, agency costs have no significant mediating effects between social networks and corporate innovation. Our mediation analyses identify a clear path of social networks influencing innovation, that is, there is a "financial constraints channel" for social networks to drive corporate innovation. Firms with rich social network connections can alleviate the external financial constraints by reducing the information asymmetry between inside and outside the companies. The financing environment with weak financial constraints is helpful for firms to raise funds for R&D activities, thus promoting the corporate innovation level. Compared with financial constraints, agency costs do not build a bridge between social networks and corporate innovation. That is to say, the promotion of social networks on innovation ability is not through improving corporate governance. There is no "agency costs channel" for social network to affect corporate innovation. In summary, for the improvement of corporate innovation ability, the external financing environment, financing convenience and financial support are more important than the internal principal-agent efficiency. Alleviating financial constraints through the core management team network is an effective governance path to improving corporate innovation.