Access to this full-text is provided by Frontiers.
Content available from Frontiers in Psychology
This content is subject to copyright.
Frontiers in Psychology 01 frontiersin.org
How does independent director
aect tunneling?—Evidence
from social networks
Hanxiu Cheng , Jie Wang * and Mu Xing
School of Economics and Management, Nanjing University of Science and Technology, Nanjing,
China
Given the influence of controlling shareholders on the company, it is important
to analyze how independent directors can protect minority shareholders’
interests using the information and resources obtained from social
networks. This paper studies the impact of director networks on controlling
shareholders’ tunneling behavior in China over the period 2002–2020.
Using social network analysis, this paper finds that controlling shareholders’
appropriation to minority shareholders is mitigated in companies with well-
connected independent directors. These results remain consistent after a
series of robustness and endogeneity tests. This study also reveals that internal
controls play a mediating role between director networks and tunneling
behavior. In addition, the study indicates that the restraining eect of director
networks on controlling shareholders’ tunneling behavior is more pronounced
in companies with weaker audit monitoring and poorer transparency. In
conclusion, the results reveal that well-connected independent directors play
an important role in protecting minority shareholders’ interests.
KEYWORDS
social networks, tunneling, corporate governance, independent director, controlling
shareholders
Introduction
is paper examines how director networks impact controlling shareholders’ tunneling
behavior. Director networks provide an informal channel for independent directors to
communicate with each other among dierent companies. By serving on multiple
companies’ boards, independent directors can not only access the latest trends related to
corporate governance and development but also pass on the information to the companies
located in director networks. Previous studies have shown that companies with well-
connected independent directors typically exhibit higher quality nancial reporting (Omer
etal., 2020), higher stock returns (Larcker etal., 2013), and lower stock price crash risk
(Fang etal., 2021). However, relatively few studies have considered how the network
connectivity of independent directors aects their corporate governance capabilities. is
paper examines the relationship between independent directors’ network connectedness
and controlling shareholders’ tunneling behavior.
TYPE Original Research
PUBLISHED 10 November 2022
DOI 10.3389/fpsyg.2022.1011761
OPEN ACCESS
EDITED BY
Wuke Zhang,
Ningbo University,
China
REVIEWED BY
Rajat Deb,
Tripura University, India
Faisal Alfordy,
University of Hail,
SaudiArabia
*CORRESPONDENCE
Jie Wang
218107010185@njust.edu.cn
SPECIALTY SECTION
This article was submitted to
Organizational Psychology,
a section of the journal
Frontiers in Psychology
RECEIVED 04 August 2022
ACCEPTED 13 October 2022
PUBLISHED 10 November 2022
CITATION
Cheng H, Wang J and Xing M (2022) How
does independent director aect
tunneling?—Evidence from social networks.
Front. Psychol. 13:1011761.
doi: 10.3389/fpsyg.2022.1011761
COPYRIGHT
© 2022 Cheng, Wang and Xing. This is an
open-access article distributed under the
terms of the Creative Commons Attribution
License (CC BY). The use, distribution or
reproduction in other forums is permitted,
provided the original author(s) and the
copyright owner(s) are credited and that
the original publication in this journal is
cited, in accordance with accepted
academic practice. No use, distribution or
reproduction is permitted which does not
comply with these terms.
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 02 frontiersin.org
It is important to clarify how independent directors with
connections to other rms’ boards can impact controlling
shareholders’ tunneling behavior. Tunneling is an opportunistic
behavior in which controlling shareholders use their controlling
power to transfer the company’s assets and prots to their own
(Johnson etal., 2000) which seriously undermines the minority
shareholder’s interests as well as exacerbates the Type II agency
conict. Especially in countries with weaker investor protection,
tunneling behavior is more prominent (Friedman etal., 2003).
Independent directors are generally considered to bear greater
responsibility for protecting the interests of minority shareholders.
In 2004, the China Securities Regulatory Commission issued
“Certain Provisions on Enhancing the Protection of Public
Shareholders’ Rights and Interests,” which emphasized that
independent directors should beindependent in performing their
duties, free from any inuence by the actual controlling
shareholders. Moreover, for material related party transactions,
the approval of all independent directors is required. While there
is considerable doubt about independent directors’ eectiveness
in performing their duties (Morck, 2004; Ye, 2014; Wang, 2015),
they do have an indispensable role in corporate governance.
Tunneling behavior severely restricts the companies’ development
and harms minority shareholders’ interests. Well-connected
independent directors may eectively monitor controlling
shareholders’ tunneling behavior. Typically, tunneling behavior is
extremely insidious. Independent directors use the information
gained from director networks and governance experience
accumulated through serving in several companies to identify the
self-interested behavior of controlling shareholders. Moreover,
well-connected independent directors usually have a higher
reputational cost (Ferris etal., 2003; Renneboog and Zhao, 2011).
Independent directors who are at the center of the network show
higher independence and are more able to say “No” to controlling
shareholders’ tunneling behavior. Accordingly, the advantage of
obtaining well-informed is essential for independent directors to
perform their corporate governance duties.
While rich networks of directors contribute to the information
exchange, it may also adversely aect corporate governance. On
the one hand, according to the director busyness theory,
independent directors who are in several companies may beless
eective in monitoring the company due to busyness (Fich and
Shivdasani, 2006; Falato et al., 2014). On the other hand,
independent directors’ extensive exchange of information between
dierent companies also carries a risk of disclosing rms’ sensitive
information (Akbas etal., 2016; Ahern, 2017; Berkman et al.,
2020). erefore, it is unclear whether well-connected
independent directors have a positive or negative impact on
controlling shareholders’ tunneling behavior.
Chinese capital market is an ideal research context for
studying how well-connected independent directors aect
controlling shareholders’ tunneling behavior. First, China
is an emerging capital market with a relatively weak legal
system that protects investors, which facilitates controlling
shareholders to expropriate minority shareholders’ interests.
Second, shareholdings in Chinese listed companies are very
concentrated, and controlling shareholders hold a great deal of
power in the company. However, controlling shareholdings are
restricted greatly in trading, which increases the possibility of
controlling shareholders beneting from tunneling behavior.
Scholars such as Cheung etal. (2009), Jian and Wong (2010), and
Jiang et al. (2010) have provided evidence that controlling
shareholders widely engage in tunneling to prot among Chinese
listed companies. ird, China is a humanitarian society.
Networking and maintaining relationships profoundly aect
individual behavioral decisions. Independent directors develop
broad relationships that have the potential both to promote greater
independence in corporate governance and to avoid disapproving
behavior by maintaining relationships. erefore, it is an
interesting and worthwhile topic to investigate how the dierent
degrees of relationships established by independent directors
aect their decision-making in corporate governance when facing
the common legal system environment.
Drawing on the sociological network centrality analysis, this
paper constructs director networks for each year from 2002 to
2020 using a sample of Chinese A-share listed companies and
calculates independent directors’ network centrality. e paper
constructs four centrality metrics (degree centrality, closeness
centrality, betweenness centrality, and eigenvector centrality) and
measures the composite score via principal component analysis.
e composite score is used in this paper to measure the rm-level
director networks’ connectedness. Firstly, this paper examines the
relationship between director networks and controlling
shareholders’ tunneling behavior. It nds that the appropriation of
minority shareholders by controlling shareholders decreases as
director network connectedness increases, in line with predictions.
Aer a series of robustness and endogenous tests, the conclusions
remain the same. Second, the paper also nds the mediating role
of internal controls between director networks and controlling
shareholders’ tunneling behavior. To test whether the negative
relationship between director networks and tunneling behavior is
inuenced by the corporate governance environment, the
following tests are conducted: (1) examine the variation of main
regressions under the dierent extent of audit monitoring; (2)
examine the eect of rm transparency on main regressions. e
results show that the negative relationship between director
networks and controlling shareholders’ tunneling behavior is
more pronounced in rms with low audit monitoring and poor
transparency. In conclusion, this study provides evidence that,
with the information and resources obtained through the network,
independent directors can eectively monitor controlling
shareholders’ opportunistic behavior and protect minority
shareholders’ interests.
is paper makes several contributions to existing literature.
First, this paper contributes to the literature on the determinants
of controlling shareholders’ tunneling behavior. Previous literature
has examined the impact of independent directors on tunneling
behavior. For example, Liu etal. (2016) found that independent
directors’ attendance at board meetings helped mitigate tunneling
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 03 frontiersin.org
behavior. Gao and Kling (2008) and Gong etal. (2021) provided
evidence that the proportion of independent directors was
associated with less tunneling behavior. is study demonstrates
how well-connected independent directors play an important role
in mitigating controlling shareholders’ tunneling behavior. e
study diers from Chen et al. (2014), as this paper provides
evidence of the mechanisms by which director networks inuence
tunneling rather than just demonstrating the relationship between
director networks and tunneling behavior. at is, this paper
proves that well-connected independent directors improve
internal control quality, which in turn inhibits controlling
shareholder tunneling behavior. Second, this study contributes to
the corporate governance literature using social network theory.
Previous research has found that external connections of
independent directors can bring good information to a company
and improve the quality of nancial reporting (Omer etal., 2020).
It could also potentially disclose sensitive company information
and result in more insider trading (Akbas etal., 2016). is paper
provides evidence that well-connected independent directors play
a positive role in monitoring controlling shareholders’
opportunistic behavior. Moreover, the governance role of director
networks is more pronounced in the context of weak external
audit monitoring and low corporate transparency.
is paper is arranged as follows. In Section 2, the paper
reviews prior literature and proposes the hypothesis. Section 3
describes the data sample and research design. Section 4 performs
the main hypothesis on the relationship between director
networks and controlling shareholders’ tunneling behaviors.
Section 5 provides a mediating eects analysis. Cross-sectional
analysis is performed in Section 6. e conclusions are
summarized in Section 7.
Literature review and hypotheses
development
Controlling shareholders’ tunneling
behavior
Controlling shareholders could appropriate minority
shareholders’ interests in various ways, such as transferring
company assets through related-party transactions (Aharony
etal., 2010; Jian and Wong, 2010), providing loan guarantees for
controlling shareholders (Jiang etal., 2010), malicious dividends
(Chen etal., 2009; Atanassov and Mandell, 2018), etc. Johnson
etal. (2000) called them “Tunneling,” which referred to the self-
interested behavior in which controlling shareholders transfer
assets and prots from companies. Especially in countries with
underdeveloped capital markets (Friedman etal., 2003), the weak
legal protection system for investors gives controlling shareholders
more opportunities to manipulate minority shareholders’ interests
(Cheung etal., 2006). e tunneling behavior has serious negative
eects, both on the interests of minority shareholders and the
performance of the company. Jiang etal. (2010) observed that
controlling shareholders extensively used company loans to
transfer assets from listed companies, which seriously undermined
the company’s operating performance. Chen etal. (2011) revealed
that controlling shareholders’ tunneling behavior had a long-term
negative impact on the companies’ stock return. Chan etal. (2016)
provided evidence that although previous studies have found that
tax avoidance activities increased rm value when tax avoidance
activities were associated with tunneling, the rm value was
discounted. Although controlling shareholders’ opportunistic
behavior seriously undermines companies’ and minority
shareholders’ interests, Cheung etal. (2006) and Aharony etal.
(2010) found it extremely secretive and dicult to predict in
advance. Atanassov and Mandell (2018) stated that good corporate
governance would help to alleviate controlling shareholders’
appropriation. Eective internal controls (Ge et al., 2021),
strengthened internal governance by independent directors (Chen
etal., 2014), and outside audit monitoring (Jiang etal., 2010)
could also mitigate controlling shareholders’ tunneling behavior.
Director networks
Previous studies supported the view that independent
directors with extensive networks are better at monitoring
companies. Carpenter and Westphal (2001) observed that
independent directors’ social networks contribute to the rms’
strategic decisions. Using the high-quality resources provided by
director networks, companies can more easily pursue new
growth opportunities (Larcker etal., 2013; Singh and Delios,
2017). Well-connected independent directors accessed
additional information and resources by serving on multiple
company boards, which not only helped to reduce the likelihood
of misstatements in nancial reporting (Omer etal., 2020) but
also helped to improve management’s earnings forecasting
accuracy (Schabus, 2022). In addition, Li etal. (2019), Field etal.
(2013) and Feng et al. (2019) have shown that independent
directors located at the central location of the director networks
played an important role in enhancing the eciency of nancing
and improving IPO valuation.
While serving on multiple boards increases the likelihood that
independent directors provide information and resources to the
company, the opposing views argue that it can also divert their
eorts and reduce the eciency of monitoring (Core etal., 1999;
Falato et al., 2014; Liu et al., 2022). For example, Fich and
Shivdasani (2006) provided evidence that independent directors
were busy due to multiple directorships and that the companies
they serve exhibit lower protability. Both Core etal. (1999) and
Fich and White (2003) found evidence that there was a positive
relationship between independent directors occupying multiple
board positions and CEO compensation. On the other hand, the
concern about director networks is that well-connected
independent directors may inadvertently leak sensitive company
information. Akbas etal. (2016) revealed that for companies with
well-connected directors, investors were more informed. Ahern
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 04 frontiersin.org
(2017) further proved that social networks provided useful
information for insider traders. In addition, Berkman etal. (2020)
observed that directors also beneted from the information
advantage in the director networks and had more shareholdings
and transactions in companies with interlocking relationships. In
conclusion, there is no consistent agreement on the eectiveness
of director networks. While the resources and information that
directors access through networks can enable them to bestrict
monitors, there are also potential negative consequences for the
company due to their busy directorships.
Taken together, the existing studies have not reached a
consistent conclusion on whether external connections of
independent directors can improve corporate governance quality.
Independent directors’ extensive external connections reect
both resource accessibility and work busyness. erefore, it is
dicult to distinguish the impact of director networks on
independent directors’ corporate governance capabilities.
Controlling shareholders enjoy greater power in the company
and are likely to inuence independent directors’ appointments.
For this reason, independent directors may succumb to
controlling shareholders’ pressure and ignore tunneling behavior.
However, well-connected independent directors, driven by
reputation, are likely to maintain a high level of independence
and actively perform monitoring duties. is paper attempts to
advance the evidence on director networks and corporate
governance capacity by examining how director networks aect
controlling shareholders’ tunneling behavior.
Hypothesis
In the Chinese capital market, a high concentration of
shareholdings in listed companies leads to increasingly
prominent conicts between controlling shareholders and
minority shareholders. Controlling shareholders gain private
interests by transferring listed companies’ assets through related
party transactions, earnings management, and other
inappropriate means, described as “tunneling” (Johnson etal.,
2000). It seriously damages minority shareholders’ interests (La
Porta etal., 2000; Cheung etal., 2009; Jian and Wong, 2010; Jiang
etal., 2010). As the primary monitoring mechanism, regulatory
authorities hold high expectations for an independent director
to perform eective monitoring functions to protect the
minority shareholders’ interests. In 2001, the China Securities
Regulatory Commission (CSRC) issued the “Guiding Opinions
on the Establishment of Independent Director System in Listed
Companies” which stipulated that independent directors should
express an independent opinion on “Whether shareholders,
actual controllers, and their aliates of the listed company have
existing or newly incurred loans or other nancial transactions
to the listed company with a total amount higher than 3 million
RMB or higher than 5% of the latest audited net asset values of
the listed company, and whether the company has taken eective
measures to collect the debts.” In 2004, the CSRC issued
“Regulations on Strengthening the Protection of Public
Shareholders’ Interests,” which emphasized again that
signicantly related party transactions should beapproved by
more than half of the independent directors. Based on resource
dependency theory, independent directors are an important
channel to connect the company with the external environment,
and their human and relational capital acquired via network
connections is an important mechanism for the company to deal
with external uncertainty (Hillman et al., 2000). e more
central the independent director is in director networks, the
more information they obtain about the industry, strategy, risk,
etc. As a result, well-connected independent directors contribute
to the quality of board decisions, which in turn restrain
controlling shareholders’ tunneling behavior. As discussed
above, although the evidence on the eectiveness of director
networks is mixed, this paper conjectures that well-connected
directors are eective in restraining controlling shareholders’
tunneling behavior. e discussion above is formalized into
hypotheses, as follows:
Hypothesis: Director networks are negatively related to
controlling shareholders’ tunneling behavior.
Research design
Data and sample
e sample consists of A-share listed companies on the
Chinese market during the period 2002–2020. e sample period
begins in 2002 as the independent directorship in China is
mandatory from 2002, and ends in 2020, the latest year for which
full data are available to measure the one-year lagged variable. is
paper mainly obtains data from the China Stock Market and
Accounting Research (CSMAR) database. e internal control
data are from the DIB Internal Control and Risk Management
Database. In line with prior studies, companies in the nancial
industry and observations with missing data are excluded. Table1
shows the selection process of the sample. Ultimately, this paper
obtains 31,826 rm-year sample observations.
Empirical model and control variables
To test hypothesis, this paper estimates the following
multivariate regression model to test the eect of director
networks on dierent forms of tunneling:
TunnelNetworksControls
Year Industry
it it it
i
,,
,
,
=+
++
å+å+
bb
e
01
tt (1)
where the dependent variable, Tunn el, is measured by other
receivables divided by total assets (Jiang etal., 2010; Liu etal.,
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 05 frontiersin.org
2016). e independent variable, Networks, is the comprehensive
measure of director networks. e specic calculation steps are
described in subsection 3.3. e median of independent director
centrality indicators is used to measure rm-level centrality
indicators and conduct principal component analysis to construct
rm-level network connectedness (Networks_media). e mean
(Networks_ mean) and maximum values (Networks_ max) are
used for robustness analysis.
is study controls for a set of variables in all specications:
the natural logarithm of the total assets at the end of the year
(Size); the net income divided by total assets at the end of the
year (ROA); growth in sales from the last scal year to the
current scal year (Growth); an indicator variable equal to one
if company report income less than zero (Loss); total debt
divided by total assets at the end of the year (Lev); current assets
divided by current liabilities (CurrentRatio); an indicator
variable that takes the value of 1 if the board chairman and CEO
are the same people (Dual); the percentage of independent
directors on the board (Ratio); the book value of equity divided
by the market value of equity (BTM); an indicator variable equal
to one if a company is state owned (SOE); shareholding of
institutional investors (InvestorShare); an indicator variable
equal to one if the company is audited by a Big 4 audit rm
(Big4). All variables are winsorized at 1 and 99%. e industry-
xed eects and year-xed eects are also controlled. Detailed
denitions of the variables used in this paper are provided in
Appendix A.
Measurement of director networks
Following previous research on director networks (Larcker
etal., 2013; Omer etal., 2014, 2020), this paper uses four centrality
measures, degree centrality (Degree), closeness centrality
(Closeness), betweenness centrality (Betweenness), and eigenvector
centrality (Eigenvector), to measure dierent aspects of director
networks and use principal component analysis to form a
composite measure.
Degree centrality indicates that one director is directly
connected to other board members in the director networks. If
two independent directors work for the same board in year t, the
companies are linked together through this director, forming an
inter-company network. e degree of centrality measures how
important the independent directors are in the network. e
formula for degree centrality is as follows,
Degree
ij
n
ij
=
()
-
¹
å
d
,
1
where i indicates an independent director, j is a board member
other than i in the board networks.
d
ij,
()
equals to 1 if board i
and board j work together on at least one corporate board, and 0
otherwise. n indicates the number of board members in the entire
board network. e number of people in board networks varies
every year, and
n-1
is introduced to eliminate the eect of
network size (Freeman 1978).
Closeness centrality represents the close relationship among
individuals. If a director can quickly connect with others in the
network, which means that the director has faster access to
information or resources (Larcker etal., 2013), then the director
has a high degree of closeness centrality. e formula for closeness
centrality is as follows,
Closeness
n
ij
ij
=
-
()
¹
å
1
m
,
where
m
ij,
()
is the shortest distance from director i to
director j.
Betweenness centrality measures the ‘bridging’ role that a
director plays in the director networks. If a director is located in
the paths connecting various directors, it means that the director
has informational or relational importance in the director
networks. e specic calculation of betweenness centrality is
as follows,
Betweenness nn
gjk
gjk
ijk
i
=-
()
-
()
()
()
¹¹
å
2
12
,
,
where gjk
i,
()
is the number of shortest paths that director j
connects to director k through director i.
gjk,
()
refers to the
number of shortest paths connecting director j to director k.
Eigenvector centrality describes the quality of network
relationships. e degree of centrality measures the direct
connection of the director in the network. Whereas, eigenvector
centrality measures whether the neighbors of that director are
well-connected. Following Bonacich (1987), eigenvector centrality
is calculated as follows,
TABLE1 Sample description.
Panel A: Sample selection for constructing director networks (2002–2020)
Director-year observations
Observations of personal characteristics of directors, supervisors
and executives
897,333
Less observations of non-board members (476,008)
Less directors appearing twice in the same rm and year (25,113)
Sample used to construct the director networks 396,212
Less observations of non-independent directors (253,441)
Sample used to calculate company-level director networks 142,771
Panel B: Sample for regression analysis
Firm-year observations
Firm-year observations for the sample aer merging with other data 40,676
Less observations in the nancial industry and missing values (8,850)
Firm-year observations used for regression analysis 31,826
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 06 frontiersin.org
Eigenvector
gE
i
j
ij j
=å
1
l
where
g
ij
is the adjacency matrix, and
g
ij
equals 1 if director
i and director j work on at least one board, and 0 otherwise.
E
j
is
the eigenvalue of the centrality of director j.
l
is the maximum
eigenvalue of the adjacency matrix. e matrix form is as follows,
l
EGE=
Higher eigenvector centrality of independent directors means
they are more prestigious in the network and more advantageous
in accessing information and resources in the network (Larcker
etal., 2013).
For each company, the study calculates the median of all
director centrality indicators as the rm-level centrality.
Meanwhile, robustness analysis is performed using the mean and
maximum values. Each centrality indicator measures a dierent
aspect of the director networks, and it is unclear which particular
indicator better captures their economic value (Larcker etal.,
2013; Omer etal., 2020). Moreover, as shown in Panel A of Table2,
the four centrality indicators are highly correlated. erefore, this
research cannot simply use one indicator to measure director
networks’ connectedness.
To comprehensively measure the connectedness of
independent directors, this paper uses principal component
analysis to construct a composite score as the rm-level measure
of director networks (Networks). Panel B of Table2 reports the
principal component analysis for the four centrality degrees. e
factor score from the rst principal component with eigenvalues
greater than 1 is used to measure the director networks’
connectedness.
Empirical result
Descriptive statistics
Table3 provides descriptive statistics for the 31,826 rm-year
observations. As shown in Panel A, the 25th and 75th quartiles
of Tun n el range from 0.004 to 0.022, indicating that tunneling is
prevalent in the sample companies. Also, the minimum value of
Tunnel is 0.000 and the maximum value is 0.246, which suggests
that the degree of appropriation by controlling shareholders
varies considerably among companies. In Panel B, the sample is
divided into the high-connected group and the low-connected
group based on the median of Networks, and the dierences in
key variables between the two groups are compared. As shown in
Panel B, in the low-connected group, the mean value of Tunn el
was 0.023. In the high-connected group, the mean value of Tunnel
was 0.021. e mean dierence test between the two groups was
signicant at the 1% level. is is consistent with the previous
analysis, where controlling shareholders’ tunneling behavior
exhibits greater variation among companies with dierent
degrees of director network connectedness. In the high connected
group, controlling shareholders exhibit lower tunneling behavior,
which implies that director networks inhibit tunneling behavior
to some extent, Hypothesis is initially proved. In the correlation
analysis of Panel C, the three proxy variables of Networks,
Networks_media, Networks_mean, and Networks_max all present
signicantly negative correlations with Tunnel , consistent with
the Hypothesis.
Main analyses
is paper rst examines the relationship between director
networks and tunneling. Table4 reports the regression results. In
column (1), the result shows a negative and signicant coecient
for Networks_media (−2.61, p < 0.01). When the regression
analysis is re-run using Networks_mean (−3.60, p < 0.01) and
Networks_max (−4.22, p < 0.01) as independent variables, the
results are robust. ese results suggest that rms with well-
connected director networks have less tunneling behavior.
Following Omer et al. (2020), this paper also calculates the
economic signicance. For one standard deviation change in
Networks
1
, the odds of a decrease in tunneling behavior are 12.6,
12.8, and 13.1%, respectively. e impact of the director networks
on tunneling behavior is also economically signicant. In
1 In the untabulated analysis, the standard deviation
d
of Networks_
media, Networks_mean and Networks_max were 1.263, 1.284 and 1.309,
respectively. According to the equation,
100 1
*-
()
*
ek
bd
, the economic
magnitude of the three coecients are −0.126, −0.128 and − 0.131,
respectively.
TABLE2 Correlation analysis between centrality indicators.
Panel A: Correlation analysis of the four centrality indicators
Degree Closeness Betweenness Eigenvector
Degree 1.000
Closeness 0.257*** 1.000
Betweenness 0.710*** 0.367*** 1.000
Eigenvector 0.092*** 0.064*** 0.080*** 1.000
Panel B: Principal component analysis
Comp1 Comp2 Comp3 Comp4
Degree 0.6167 −0.0857 −0.4003 0.6723
Closeness 0.4289 −0.0433 0.8925 0.1325
Betweenness 0.6442 −0.1104 −0.2068 −0.7281
Eigenvector 0.1440 0.9892 −0.0187 −0.0172
Eigenvalue 1.942 0.980 0.798 0.279
Proportion
(%)
0.486 0.245 0.200 0.069
Cumulative
(%)
0.486 0.731 0.931 1.000
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 07 frontiersin.org
TABLE3 Descriptive statistics.
Panel A: Descriptive statistics of the full sample
Var i a b l e NMean Min P25 P50 P75 Max SD
Tunne l 31,826 0.022 0.000 0.004 0.009 0.022 0.246 0.038
Size 31,826 22.069 19.310 21.160 21.911 22.810 26.027 1.294
ROA 31,826 0.030 −0.358 0.011 0.033 0.062 0.197 0.073
Growth 31,826 0.174 −0.650 −0.033 0.105 0.271 2.969 0.459
Loss 31,826 0.117 0.000 0.000 0.000 0.000 1.000 0.322
Lev 31,826 0.463 0.062 0.301 0.460 0.616 1.010 0.211
CurrentRatio 31,826 2.061 0.224 1.020 1.463 2.279 12.913 1.982
Dual 31,826 0.230 0.000 0.000 0.000 0.000 1.000 0.421
Ratio 31,826 0.371 0.286 0.333 0.333 0.400 0.571 0.053
BTM 31,826 0.324 −0.005 0.208 0.304 0.421 0.787 0.162
SOE 31,826 0.203 0.000 0.000 0.000 0.000 1.000 0.402
InvestorShare 31,826 0.465 0.005 0.299 0.489 0.643 0.907 0.232
Big4 31,826 0.059 0.000 0.000 0.000 0.000 1.000 0.236
Panel B: Dierence analysis between groups
Low-connected High-connected Dierence
NMean NMean
Tunne l 15,919 0.023 15,907 0.021 0.003***
Size 15,919 21.873 15,907 22.265 −0.393***
ROA 15,919 0.027 15,907 0.033 −0.007***
Growth 15,919 0.171 15,907 0.177 −0.006
Loss 15,919 0.134 15,907 0.101 0.033***
Lev 15,919 0.453 15,907 0.473 −0.020***
CurrentRatio 15,919 2.175 15,907 1.947 0.229***
Dual 15,919 0.247 15,907 0.213 0.033***
Ratio 15,919 0.372 15,907 0.369 0.004***
BTM 15,919 0.320 15,907 0.328 −0.008***
SOE 15,919 0.183 15,907 0.222 −0.039***
InvestorShare 15,919 0.444 15,907 0.485 −0.041***
Big4 15,919 0.047 15,907 0.072 −0.024***
Panel C: Correlation analysis
(1) (2) (3) (4) (5) (6) (7) (8)
(1) Tunne l 1
(2) Networks_media −0.017*** 1
(3) Networks_mean −0.019*** 0.923*** 1
(4) Networks_max −0.031*** 0.805*** 0.962*** 1
(5) Size −0.126*** 0.142*** 0.152*** 0.164*** 1
(6) ROA −0.270*** 0.063*** 0.074*** 0.076*** 0.099*** 1
(7) Growth −0.041*** 0.024*** 0.029*** 0.027*** 0.033*** 0.228*** 1
(8) Loss 0.196*** −0.061*** −0.069*** −0.069*** −0.119*** −0.690*** −0.199*** 1
(9) Lev 0.226*** 0.077*** 0.086*** 0.083*** 0.336*** −0.375*** 0.025*** 0.224***
(10) CurrentRatio −0.092*** −0.074*** −0.085*** −0.082*** −0.255*** 0.212*** −0.030*** −0.116***
(11) Dual −0.020*** −0.074*** −0.086*** −0.084*** −0.123*** 0.007 0.006 0.007
(12) Ratio −0.016*** −0.092*** −0.091*** −0.048*** 0.038*** −0.020*** −0.006 0.011**
(13) BTM −0.139*** −0.006 −0.007 −0.009 0.072*** 0.152*** −0.052*** −0.165***
(14) SOE 0.056*** 0.116*** 0.148*** 0.143*** 0.059*** −0.007 0.048*** −0.017***
(15) InvestorShare −0.008 0.146*** 0.170*** 0.168*** 0.366*** 0.114*** 0.054*** −0.088***
(16) Big4 −0.034*** 0.074*** 0.093*** 0.106*** 0.333*** 0.059*** −0.015*** −0.042***
(Continued)
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 08 frontiersin.org
summary, Table4 supports the view that the degree of independent
director connectedness is associated with lower controlling
shareholders’ tunneling behavior.
Consistent with previous research (Gao and Kling, 2008; Liu
etal., 2016), controlling shareholders’ tunneling behavior increases
with poor company performance (Loss), nancial leverage (Lev),
and current ratio (CurrentRatio). is paper nds negative and
signicant coecients for Size, ROA, Growth, BTM, SOE, and
InvestorShare, indicating that rms with a larger size, higher
protability, and better growth opportunities are less likely to
experience controlling shareholders’ expropriation. In addition,
state-owned enterprises and rms with higher shareholdings of
institutional investors have fewer controlling shareholders’
tunneling behavior.
Robustness tests
To further investigate the robustness of the results, this paper
conducts a series of tests. Firstly, following Liu etal. (2016), the
paper performs the regression analysis by using ln(1 +Tun nel ) as
the dependent variable. e results are shown in Panel A of
Table 5, where the coecients of Networks_media, Networks_
mean, and Networks_max are still signicantly negative aer
replacing the measurement of the dependent variable. As a result,
the conclusions remain consistent aer changing the dependent
variable’s measurements. Second, clustering standard errors at the
rm level are used to settle possible clustering eects in the main
regression analysis (Liu etal., 2016). e results are shown in
Table5. e coecient of Networks_media, Networks_mean, and
Networks_max are signicantly negative, which is consistent with
the results in Table4. In conclusion, the results remain robust aer
adjusting for clustering eects. ird, the nancial crisis severely
undermines rms’ performance, making it dicult for controlling
shareholders to appropriate rms’ interests. Sometimes, the
controlling shareholders may berequired to provide support to
the company. To avoid the impact of the nancial crisis, this study
restricts the sample to the period before the nancial crisis, that
is, before 2006. In panel C of Table5, the results are consistent
with the full sample, which means that the sample contains
nancial crisis years that do not aect the main results. Finally, as
How etal. (2008) revealed, compared to larger companies, smaller
companies have more limitations in acquiring resources and suer
from worse corporate governance, which makes them easier to
expropriate by controlling shareholders. If the results are
inuenced by rm size, then it is expected that more signicant
results on director networks and tunneling behavior will
beobserved in the small rm. To further test whether the results
are driven by rm size, this paper divides the sample into two
groups based on rm size quartiles. If the rm size is in the rst
quartile, it is in the small rm group, otherwise it is in the large
rm group. As shown in Panel D of Table5, the coecients of
Networks_media, Networks_mean and Networks_max are
signicantly negative in both sample groups. is alleviates
concerns about the eect of rm size.
Endogeneity tests
In the above analysis, this paper has demonstrated that well-
connected independent directors play an important role in
mitigating controlling shareholders’ tunneling behavior. However,
some omitted variables may lead to a spurious negative correlation
between director networks and tunneling behavior. To mitigate
this concern, this paper rst controls for rm xed eects in the
baseline regressions to control for the impact of time-invariant
rm characteristics on the results. e results are shown in Panel
A of Table6. e coecients of Networks_media, Networks_mean,
and Networks_max are signicantly negative. e results support
the ndings that director networks inhibit controlling
shareholders’ tunneling behavior.
Secondly, controlling shareholders retain considerable
control over the company, they may interfere with the
appointment of independent directors and choose the “obedient”
independent directors to join the company to successfully
conduct tunneling. In addition, as mentioned above, well-
connected independent directors usually possess a higher
reputation, which makes them more likely to accept oers from
companies in good standing. To alleviate the potential
endogeneity issues aecting the results, this paper addresses the
impact of director networks on tunneling behavior in an
exogenous shock context. On 19 October 2013, the Organization
(9) (10) (11) (12) (13) (14) (15) (16)
(9) Lev 1
(10) CurrentRatio −0.646*** 1
(11) Dual −0.118*** 0.118*** 1
(12) Ratio −0.019*** 0.035*** 0.121*** 1
(13) BTM −0.536*** 0.262*** −0.008 −0.042*** 1
(14) SOE 0.130*** −0.130*** −0.157*** −0.101*** 0.068*** 1
(15) InvestorShare 0.176*** −0.171*** −0.210*** −0.084*** −0.020*** 0.290*** 1
(16) Big4 0.062*** −0.075*** −0.064*** 0.025*** 0.053*** 0.057*** 0.236*** 1
TABLE3 (Continued)
Panel C: Correlation analysis
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 09 frontiersin.org
Department of the CPC issued the “Opinions on Further
Regulating the Issue of Party and Government Leading Cadres’
Part-time Positions in Enterprises” (hereaer No. 18), which
prohibited ocials from serving as independent directors in
listed companies. No. 18 triggered many independent directors
with ocer status to resign in subsequent years. If well-connected
independent directors play an important role in restraining
controlling shareholders’ tunneling behavior, then the paper
expects that aer the issuance of No. 18, tunneling behavior in
companies with well-connected director networks will increase.
Post is set to 1 for the year immediately aer the No. 18 and 0
otherwise. In addition, following Fang etal. (2021), High_med
(calculated according to Networks_media) is equal to 1 for
director networks above the industry median and 0 otherwise
(the similar setting for High_mean and High_max). As shown in
Panel B of Table 6, the coecient on High_med*Post (3.40,
p< 0.01) is signicantly positive. Using High_mean and High_
max, the conclusion stands. e results indicate that controlling
shareholders’ tunneling behavior increases when companies lose
many broadly connected independent directors.
ird, as analyzed above, independent directors with rich
social networks may avoid companies with heavy tunneling
behavior due to reputational concerns. To address this issue, the
paper uses the changes in Networks from t-1 to t (ΔNetworks_
media, ΔNetworks_mean, and ΔNetworks_max) as the
dependent variable, Occupy with one lag as the independent
variable (Lag_Occupy), and controls for the lagged values of all
the above control variables. If this concern holds, then it should
be possible to observe a signicantly negative coecient on
Lag_Occupy in this regression. In Panel C of Table 6, the
coecient of Lag_Occupy is not signicant in any of the three
groups. Generally, the paper does not nd any evidence that
director networks are endogenously matched to controlling
shareholders’ tunneling behavior.
The mediating eect of internal
control
In the above analysis, the paper nds that director networks
signicantly inhibit controlling shareholders’ tunneling behavior.
One possible explanation is that well-connected independent
directors, with the information gained from the director networks
and extensive experience accumulated from serving in several
companies, help to improve the quality of internal controls,
which in turn inhibits controlling shareholders’ opportunistic
behavior. To test this conjecture, the research introduces the
mediating variable of internal control (IC). By constructing the
mediating eect model, the paper analyses whether director
networks inhibit controlling shareholders’ tunneling behavior by
improving the quality of internal control. Following Chan etal.
(2021), the paper uses the Internal Control Disclosure Index (in
natural logarithms) from the DIB Internal Control and Risk
Management Database to measure internal control quality (IC).
A higher value of IC indicates that the company has higher
internal control quality. In the baseline regressions, the results
have demonstrated the general eect of director networks on
controlling shareholders’ tunneling behavior. Next, the paper
further tests the mediating eect of internal control by
constructing the following two equations:
IC Networks Controls
Year Industry
it it it
it
,,
,
,
=+
++
å+å+
aa
e
01
(2)
TunnelNetworksIC
Controls Year Indu
it it it
it
,,,
,
=+ ++
+å +å
ll l
01 2
sstry it
+
e
, (3)
TABLE4 Multivariate results for director networks and tunneling.
(1) (2) (3)
Networks_media −0.001***
(−3.99)
Networks_mean −0.001***
(−5.63)
Networks_max −0.001***
(−6.52)
Size −0.003*** −0.002*** −0.002***
(−9.41) (−9.13) (−8.98)
ROA −0.083*** −0.083*** −0.083***
(−12.60) (−12.57) (−12.56)
Growth −0.001* −0.001* −0.001*
(−1.66) (−1.69) (−1.70)
Loss 0.003** 0.003** 0.003**
(2.44) (2.44) (2.44)
Lev 0.029*** 0.029*** 0.029***
(12.33) (12.27) (12.24)
CurrentRatio 0.001*** 0.001*** 0.001***
(10.44) (10.39) (10.36)
Dual 0.000 −0.000 −0.000
(0.02) (−0.02) (−0.05)
Ratio 0.004 0.004 0.005
(1.17) (1.09) (1.32)
BTM −0.016*** −0.016*** −0.016***
(−8.69) (−8.79) (−8.85)
SOE −0.005*** −0.005*** −0.005***
(−7.64) (−7.58) (−7.52)
InvestorShare −0.002*** −0.002*** −0.002***
(−2.69) (−2.66) (−2.65)
Big4 0.000 0.000 0.000
(0.29) (0.36) (0.47)
_cons 0.116*** 0.114*** 0.113***
(18.74) (18.54) (18.35)
Year Fixed Ye s Ye s Yes
Industry Fixed Yes Yes Yes
N 31,826 31,826 31,826
Adj. R-Square 0.214 0.214 0.215
***, **, and * denote signicance at the 1, 5, and 10% level, respectively.
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 10 frontiersin.org
TABLE5 Robustness test.
Panel A: Changing the measurement of the dependent variable
(1) (2) (3)
Networks_media −0.001**
(−2.49)
Networks_mean −0.001***
(−3.45)
Networks_max −0.001***
(−4.06)
Controls 0.107*** 0.106*** 0.105***
(11.44) (11.29) (11.16)
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
Firm Fixed Ye s Ye s Ye s
N 31,826 31,826 31,826
Adj. R-Square 0.217 0.218 0.218
Panel B: Cluster analysis at the rm level
(1) (2) (3)
Networks_media −0.001***
(−2.61)
Networks_mean −0.001***
(−3.58)
Networks_max −0.001***
(−4.19)
Controls Ye s Ye s Ye s
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
N 31,826 31,826 31,826
Adj. R-Square 0.214 0.214 0.215
Panel C: Avoiding the impact of the nancial crisis
(1) (2) (3)
Networks_media −0.002***
(−3.79)
Networks_mean −0.002***
(−3.18)
Networks_max −0.002***
(−2.71)
Controls Ye s Ye s Ye s
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
N 2,820 2,820 2,820
Adj. R-Square 0.403 0.403 0.402
Panel D: e impact of rm size
(1) Small (2) Large (3) Small (4) Large (5) Small (6) Large
Networks_media −0.001*** −0.000*
(−3.14) (−1.71)
Networks_mean −0.002*** −0.000***
(−3.71) (−2.94)
Networks_max −0.002*** −0.001***
(−3.79) (−4.08)
Controls Ye s Ye s Ye s Yes Ye s Ye s
Year Fixed Ye s Ye s Ye s Ye s Yes Ye s
Industry Fixed Yes Ye s Ye s Ye s Yes Ye s
(Continued)
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 11 frontiersin.org
Columns (1)–(3) in Table7 show the relationship between
director networks and internal control. e coecients of
Networks_media, Networks_mean, and Networks_max are all
signicantly positive at the 1% level, implying that enriched social
networks of independent directors contribute to companies’
internal control quality. is result is consistent with Sun etal.
(2012) that independent directors play an important role in
improving internal control quality. Columns (4)–(6) of Table7
show the results aer adding the mediating variable IC. e
coecients of Networks_media, Networks_mean, and Networks_
max remain signicant aer adding the mediating variable IC to
the baseline analysis, which indicates that the mediating eect of
internal control holds. Ge etal. (2021) provided evidence that
high-quality internal controls are eective in reducing controlling
shareholders’ resource extraction and protecting the interests of
minority shareholders. Based on this perspective, this paper
provides further evidence that well-connected independent
directors play a positive role in improving internal control quality
in companies, which in turn eectively restrains controlling
shareholders’ tunneling behavior.
Cross-sectional analyses
While the above analysis provides evidence that director
networks are signicantly and negatively related to controlling
shareholders’ tunneling behavior, this paper conjectures that the
eect of director networks on tunneling behavior will vary for
rms with dierent ownership types as well as governance
environments. Especially in rms with weak governance, the
restraining eect of director networks on tunneling behavior is
more remarkable. erefore, the following two sets of cross-
sectional analyses are carried out: (1) the degree of the audit rm’s
monitoring; (2) the degree of rm transparency.
Quality of audit monitoring
Previous studies show that audit plays an important role in
corporate governance (Ashbaugh and Wareld, 2003). Gao and
Kling (2008) and Jiang etal. (2010) also nd that auditors can
eectively monitor tunneling behavior and companies with a
high degree of tunneling are more likely to receive a qualied
opinion from the auditor. erefore, this paper argues that the
restraining eect of director networks on tunneling behavior is
not signicant in companies with higher quality audit
monitoring. Previous studies have shown that Big 4 audit rms
provide higher quality audits compared to non-Big 4 audit rms
N 7,957 23,869 7,957 23,869 7,957 23,869
Adj. R-Square 0.320 0.141 0.320 0.141 0.321 0.142
***, **, and * denote signicance at the 1, 5, and 10% level, respectively.
TABLE5 (Continued)
TABLE6 Endogeneity tests.
Panel A: Control for rm-level xed eects
(1) (2) (3)
Networks_media −0.000*
(−1.72)
Networks_mean −0.001***
(−2.90)
Networks_max −0.001***
(−3.73)
Controls Ye s Ye s Ye s
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
N 31,826 31,826 31,826
Adj. R-Square 0.039 0.039 0.039
Panel B: Impact of ocial independent directors’ departure
(1) (2) (3)
Post −0.045*** −0.045*** −0.045***
(−12.75) (−12.74) (−12.85)
High_med −0.002**
(−2.00)
High_med*Post 0.003***
(3.40)
High_ mean −0.002***
(−3.47)
High_mean*Post 0.002***
(3.14)
High_max −0.003***
(−5.10)
High_max*Post 0.003***
(3.59)
Controls Ye s Ye s Ye s
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
N 31,826 30,465 31,826
Adj. R-Square 0.214 0.216 0.214
Panel C: Changes in director networks
(1) (2) (3)
ΔNetworks_media ΔNetworks_mean ΔNetworks_max
Lag_Occupy −0.207 −0.238 −0.220
(−1.14) (−1.34) (−1.13)
Lag_Controls Ye s Ye s Ye s
Year Fixed Ye s Ye s Ye s
Industry Fixed Ye s Ye s Ye s
N 31,826 30,465 31,826
Adj. R-Square 0.214 0.216 0.214
***, **, and * denote signicance at the 1, 5, and 10% level, respectively.
Panel D: e impact of rm size
(1) Small (2) Large (3) Small (4) Large (5) Small (6) Large
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 12 frontiersin.org
(Khurana and Raman, 2004; De Franco etal., 2011). Big 4
auditors play a more eective external monitoring role in
corporate governance (Fan and Wong, 2005; Gul etal., 2010).
Consequently, following Gul etal. (2010), the sample is divided
into two groups based on whether the company is audited by
international Big 4 audit rms. at is, Big4 equals 1 if the
company is audited by Big 4 audit rms and 0 otherwise. It is
expected that the eect of director networks on tunneling
behavior is more signicant in companies audited by non-Big 4.
e results are shown in columns (1) and (2) of Table8. In the
non-Big4 audit sample, the coecient on Networks (−3.60,
p < 0.01) is signicantly negative. While in the Big4 group, the
coecient of Networks is not signicant. As expected, director
networks have more signicant inhibitory eects on tunneling
behavior in companies with poorer audit governance.
Corporate transparency
Bhat etal. (2006) provide evidence that companies with a
high level of transparency exhibit stronger corporate governance,
which in turn helps analysts issue more accurate earnings
forecasts. Improving corporate transparency also helps promote
the ecient allocating of resources (Francis etal., 2009), mitigate
the impact of investor sentiment on share prices (Firth etal.,
2015), as well as reduce IPO costs (Ang and Brau, 2002). For this
reason, this paper argues that companies with high transparency
have low information asymmetry with external stakeholders,
where the eect of director networks on tunneling behavior is not
signicant. In contrast, controlling shareholders’ tunneling
behavior is more likely to occur in companies with low
transparency, and the governance eect of director networks is
more pronounced. e Shanghai Stock Exchange and Shenzhen
Stock Exchange annually evaluate the information disclosure
work of listed companies, which is divided into four grades from
high to low: A, B, C, and D. Companies with A or B results are
classied in the high disclosure group, and Opacity equals 1.
2
e
remaining sample is in the low information disclosure group,
Opacity equals to 0. e regression results are shown in columns
(3) and (4) of Table 8. Consistent with the prediction, the
restraining eect of director networks on controlling
shareholders’ tunneling behavior is only signicant in the group
with lower transparency.
Conclusion
is paper examines how well-connected independent
directors inuence controlling shareholders’ tunneling behavior.
Studies based on social network theory argue that social networks
built by independent directors through serving on multiple boards
help them to access and pass on information, which is essential for
2 In the untabulated analysis, this paper defines the group with result A
as the high disclosure group, while B,C and D are classified as the low
disclosure group, the results are consistent.
TABLE7 Mediating eect.
(1) (2) (3) (4) (5) (6)
IC IC IC Occupy Occupy Occupy
Networks_media 0.004*** −0.000**
(6.22) (−2.28)
Networks_mean 0.005*** −0.001***
(7.67) (−3.55)
Networks_max 0.005*** −0.001***
(7.87) (−4.37)
IC −0.024*** −0.024*** −0.024***
(−11.69) (−11.66) (−11.64)
Controls Ye s Ye s Yes Ye s Ye s Yes
Year Fixed Yes Ye s Yes Ye s Ye s Yes
Industry Fixed Ye s Yes Yes Ye s Ye s Ye s
N 30,402 30,402 30,402 30,402 30,402 30,402
Adj. R-Square 0.297 0.298 0.298 0.226 0.226 0.227
***, **, and * denote signicance at the 1, 5, and 10% level, respectively.
TABLE8 Cross-sectional analyses.
(1) (2) (3) (4)
Big4= 1 Big4= 0 Opacity= 1 Opacity= 0
Networks 0.000 −0.001*** 0.000 −0.001***
(0.42) (−3.60) (1.10) (−4.21)
Controls Ye s Ye s Ye s Ye s
Year Fixed Ye s Ye s Yes Ye s
Industry
Fixed
Ye s Ye s Ye s Ye s
Prob > chi2 = 0.052 Prob > chi2 = 0.000
N 1892 29,934 17,612 14,214
Adj. R-Square 0.202 0.219 0.159 0.224
***, **, and * denote signicance at the 1, 5, and 10% level, respectively.
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 13 frontiersin.org
improving independent directors’ corporate governance
capabilities. In accordance with this perspective, the paper
proposes the hypothesis that director networks have a negative
relationship with controlling shareholders’ tunneling behavior.
Using a sample of Chinese listed companies from 2002 to 2020,
this paper provides evidence of how director networks inuence
controlling shareholders’ tunneling behavior. e results show that
well-connected director networks contribute to independent
directors’ governance capacity, which helps to inhibit controlling
stakeholders’ expropriation of minority shareholders. is paper
reports the same results through robustness tests, such as replacing
the measurement of the dependent variable, cluster analysis at the
rm level, and narrowing the sample, as well as endogeneity tests,
including controlling for rm-level xed eects and using ocer
director departures as exogenous events. Overall, the results indicate
that external networking of independent directors can help curb
controlling shareholders’ tunneling behavior and protect minority
shareholders’ interests. e Hypothesis is veried. Further, the
analysis shows that internal control plays a mediating role. Well-
connected independent directors help to improve the rms’ internal
control quality, which in turn reduces the risk of controlling
shareholders engaging in tunneling behavior. In addition, cross-
sectional analyses show that the relationship between director
networks and controlling shareholder tunneling behavior is more
pronounced in companies with poor external audit oversight and
less transparency. Overall, the results suggest that extensive director
networks strengthen independent directors’ capacity in
corporate governance.
is paper has some implications. e results have shown that
companies benet from independent directors’ networking. Well-
connected independent directors not only protect the company’s
sustainability and minority shareholders’ interests but also
compensate for the inadequacy of other governance mechanisms.
erefore, when appointing independent directors, companies
can broadly consider the network location of the independent
director and the resources they can bring to the company.
ere are limitations to this study. Although this paper adopts
some research design to alleviate the endogeneity matching issue
between director networks and tunneling behavior, it is dicult to
completely eliminate the inuence of omitted variables. Secondly, the
sample in this paper includes only listed companies. Unlisted
companies account for a large proportion of the Chinese capital
market. Moreover, compared to listed companies, unlisted
companies face less strict regulation, and controlling shareholders
are more likely to engage in tunneling. However, owing to limitations
on data availability, this paper is restricted to listed companies.
Consequently, the results in this paper should beinterpreted with
caution. If the data allow, future research could discuss whether the
negative relationship between director networks and controlling
shareholder tunneling behavior also holds in unlisted companies.
Cross-country studies are also recommended. rough multi-
country studies, it is further investigated whether director networks
dier in governance eectiveness in dierent countries and dierent
institutional contexts.
Data availability statement
Publicly available datasets were analyzed in this study. is
data can befound at: https://www.gtarsc.com/.
Author contributions
HC and JW participated in the study design, developed the
theoretical framework, and wrote the rst dra of the
manuscript. MX was responsible for data collection and
analysis. HC and JW reviewed the manuscript. All authors
contributed to the article and approved the submitted version,
led by HC.
Funding
e authors gratefully thank the nancial support provided by
the Postgraduate Research & Practice Innovation Program of
Jiangsu Province (KYCX21_0392).
Conflict of interest
e authors declare that the research was conducted in the
absence of any commercial or nancial relationships that could
beconstrued as a potential conict of interest.
Publisher’s note
All claims expressed in this article are solely those of the
authors and do not necessarily represent those of their aliated
organizations, or those of the publisher, the editors and the
reviewers. Any product that may be evaluated in this article, or
claim that may be made by its manufacturer, is not guaranteed or
endorsed by the publisher.
References
Aharony, J., Wang, J., and Yuan, H. (2010). Tunneling as an incentive for earnings
management during the Ipo process in China. J. Account. Public Policy 29, 1–26. doi:
10.1016/j.jaccpubpol.2009.10.003
Ahern, K. R. (2017). Information networks: evidence from illegal insider trading
tips. J. Financ. 125, 26–47. doi: 10.1016/j.jneco.2017.03.009
Akbas, F., Meschke, F., and Wintoki, M. B. (2016). Director networks and
informed traders. J. Account. Econ. 62, 1–23. doi: 10.1016/j.jacceco.2016.
03.003
Ang, J. S., and Brau, J. C. (2002). Firm transparency and the costs of going public.
J. Financ. Res. 25, 1–17. doi: 10.1111/1475-6803.00001
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 14 frontiersin.org
Ashbaugh, H., and Wareld, T. D. (2003). Audits as a corporate governance
mechanism: evidence from the German market. J. Int. Account. Res. 2, 1–21. doi:
10.2308/jiar.2003.2.1.1
Atanassov, J., and Mandell, A. J. (2018). Corporate governance and dividend
policy: evidence of tunneling from master limited partnerships. J. Corp. Finan. 53,
106–132. doi: 10.1016/j.jcorpn.2018.10.004
Berkman, H., Koch, P., and Westerholm, P. J. (2020). Inside the director network:
when directors trade or hold inside, interlock, and unconnected stocks. J. Bank.
Financ. 118:105892. doi: 10.1016/j.jbankn.2020.105892
Bhat, G., Hope, O. K., and Kang, T. (2006). Does corporate governance
transparency aect the accuracy of analyst forecasts? Accounting Finance 46,
715–732. doi: 10.1111/j.1467-629X.2006.00191.x
Bonacich, P. (1987). Power and centrality: a family of measures. Am. J. Sociol. 92,
1170–1182. doi: 10.1086/228631
Carpenter, M. A., and Westphal, J. D. (2001). e strategic context of external
network ties: examining the impact of director appointments on board
involvement in strategic decision making. Acad. Manag. J. 44, 639–660. doi:
10.5465/3069408
Chan, K. C., Chen, Y., and Liu, B. (2021). e linear and non-linear eects of
internal control and its ve components on corporate innovation: evidence from
Chinese rms using the Coso framework. European Accounting Revie w 30, 733–765.
doi: 10.1080/09638180.2020.1776626
Chan, K. H., Mo, P. L. L., and Tang, T. (2016). Tax avoidance and tunneling:
empirical analysis from an agency perspective. J. Int. Account. Res. 15, 49–66. doi:
10.2308/jiar-51345
Chen, J. J., Cheng, P., and Xiao, X. (2011). Related party transactions as a
source of earnings management. Appl. Financ. Econ. 21, 165–181. doi:
10.1080/09603107.2010.528361
Chen, D., Jian, M., and Xu, M. (2009). Dividends for tunneling in a regulated
economy: the case of China. Pac. Basin Financ. J. 17, 209–223. doi: 10.1016/j.
pacn.2008.05.002
Chen, Y., Wang, Y., and Lin, L. (2014). Independent Directors' Board networks
and controlling Shareholders' tunneling behavior. China J. Account. Res. 7, 101–118.
doi: 10.1016/j.cjar.2013.09.002
Cheung, Y.-L., Jing, L., Lu, T., Rau, P. R., and Stouraitis, A. (2009). Tunneling and
propping up:an analysis of related party transactions by Chinese listed companies.
Pac. Basin Financ. J. 17, 372–393. doi: 10.1016/j.pacn.2008.10.001
Cheung, Y., Rau, P., and Stouraitis, A. (2006). Tunneling, propping and
expropriation: evidence from connected party transactions in Hong Kong. J. Financ.
Econ. 82, 343–386. doi: 10.1016/j.jneco.2004.08.012
Core, J. E., Holthausen, R. W., and Larcker, D. F. (1999). Corporate governance,
chief executive ocer compensation, and rm performance. J. Financ. Econ. 51,
371–406. doi: 10.1016/S0304-405X(98)00058-0
De Franco, G., Gavious, I., Jin, J. Y., and Richardson, G. D. (2011). Do private
company targets that hire big 4 auditors receive higher proceeds? Contemp. Account.
Res. 28, 215–262. doi: 10.1111/j.1911-3846.2010.01047.x
Falato, A., Kadyrzhanova, D., and Lel, U. (2014). Distracted directors: does board
busyness hurt shareholder value? J. Financ. Econ. 113, 404–426. doi: 10.1016/j.
jneco.2014.05.005
Fan, J. P. H., and Wong, T. J. (2005). Do external auditors perform a corporate
governance role in emerging markets? Evidence from East Asia. J. Account. Res. 43,
35–72. doi: 10.1111/j.1475-679x.2004.00162.x
Fang, X., Pittman, J., and Zhao, Y. (2021). e importance of director external
social networks to stock Price crash risk. Contemp. Account. Res. 38, 903–941. doi:
10.1111/1911-3846.12647
Feng, Y., Song, K., and Tian, Y. S. (2019). Director networks and initial public
oerings. J. Bank. Financ. 106, 246–264. doi: 10.1016/j.jbankn.2019.07.001
Ferris, S. P., Jagannathan, M., and Pritchard, A. C. (2003). Too busy to mind the
business? Monitoring by directors with multiple board appointments. J. Financ. 58,
1087–1111. doi: 10.1111/1540-6261.00559
Fich, E. M., and Shivdasani, A. (2006). Are busy boards eective monitors? J.
Financ. 61, 689–724. doi: 10.1111/j.1540-6261.2006.00852.x
Fich, E., and White, L. J. (2003). Ceo compensation and turnover: the eects of
mutually interlocked boards. Wake Forest L. Rev. 38:935.
Field, L., Lowry, M., and Mkrtchyan, A. (2013). Are busy boards detrimental? J.
Financ. Econ. 109, 63–82. doi: 10.1016/j.jneco.2013.02.004
Firth, M., Wang, K., and Wong, S. M. (2015). Corporate transparency and the
impact of investor sentiment on stock prices. Manag. Sci. 61, 1630–1647. doi:
10.1287/mnsc.2014.1911
Francis, J. R., Huang, S., Khurana, I. K., and Pereira, R. (2009). Does corporate
transparency contribute to ecient resource allocation? J. Account. Res. 47, 943–989.
doi: 10.1111/j.1475-679X.2009.00340.x
Freeman, L. C. (1978). Centrality in Social Networks Conceptual Clarication.
Social Networks 1, 215–239. doi: 10.1016/0378-8733(78)90021-7
Friedman, E., Johnson, S., and Mitton, T. (2003). Propping and tunneling. J. Comp.
Econ. 31, 732–750. doi: 10.1016/j.jce.2003.08.004
Gao, L., and Kling, G. (2008). Corporate governance and tunneling: empirical evidence
from China. Pac. Basin Financ. J. 16, 591–605. doi: 10.1016/j.pacn.2007.09.001
Ge, W., Li, Z., Liu, Q., and McVay, S. (2021). Internal control over Financial
reporting and resource extraction: evidence from China. Contemp. Account. Res. 38,
1274–1309. doi: 10.1111/1911-3846.12653
Gong, M., Wang, Y., and Yang, X. (2021). Do independent directors restrain
controlling shareholders’ tunneling? Evidence from a natural experiment in China.
Econ. Model. 94, 548–559. doi: 10.1016/j.econmod.2020.01.023
Gul, F. A., Kim, J.-B., and Qiu, A. A. (2010). Ownership concentration, foreign
shareholding, audit quality, and stock Price synchronicity: evidence from China. J.
Financ. Econ. 95, 425–442. doi: 10.1016/j.jneco.2009.11.005
Hillman, A. J., Cannella, A. A., and Paetzold, R. L. (2000). e resource dependence
role of corporate directors: strategic adaptation of board composition in response to
environmental change. J. Manag. Stud. 37, 235–256. doi: 10.1111/1467-6486.00179
How, J. C. Y., Verhoeven, P., and Wu, C. L. (2008). Dividends and expropriation in
Hong Kong. Asian Academy of Management Journal of Accounting & Finance 4, 71–85.
Jian, M., and Wong, T. J. (2010). Propping through related party transactions. Rev.
Acc. Stud. 15, 70–105. doi: 10.1007/s11142-008-9081-4
Jiang, G., Lee, C. M. C., and Yue, H. (2010). Tunneling through Intercorporate loans:
the China experience. J. Financ. Econ. 98, 1–20. doi: 10.1016/j.jneco.2010.05.002
Johnson, S., La Porta, R., Lopez-de-Silanes, F., and Shleifer, A. (2000). Tunneling.
Am. Econ. Rev. 90, 22–27. doi: 10.1257/aer.90.2.22
Khurana, I. K., and Raman, K. K. (2004). Litigation risk and the nancial
reporting credibility of big 4 versus non-big 4 audits: evidence from Anglo-
American countries. Account. Rev. 79, 473–495. doi: 10.2308/accr.2004.79.2.473
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and Vishny, R. (2000). Investor
protection and corporate Govemance. J. Financ. Econ. 58, 3–27. doi: 10.1016/
S0304-405X(00)00065-9
Larcker, D. F., So, E. C., and Wang, C. C. Y. (2013). Boardroom centrality and rm
performance. J. Account. Econ. 55, 225–250. doi: 10.1016/j.jacceco.2013.01.006
Li, D., Jiang, Q., and Mai, Y. (2019). Board interlocks and capital structure dynamics:
evidence from China. Accounting & Finance 59, 1893–1922. doi: 10.1111/ac.12531
Liu, Y., Ma, S., and Tang, X. (2022). Independent director networks and executive
perquisite consumption—“collusion” or “coordination” in governance? Emerg. Mark.
Financ. Trade 1-16, 1–16. doi: 10.1080/1540496X.2022.2072202
Liu, H., Wang, H., and Wu, L. (2016). Removing vacant chairs: does independent
directors’ attendance at board meetings matter? J. Bus. Ethics 133, 375–393. doi:
10.1007/s10551-014-2402-6
Morck, R. (2004). "Behavioral nance in corporate governance-independent
directors, non-executive chairs, and the importance of the Devil's advocate". National
Bureau of Economic Research Cambridge, Mass., USA).
Omer, T. C., Shelley, M. K., and Tice, F. M. (2014). Do well-connected directors
aect rm value? J. Appl. Financ. 24, 17–32. doi: 10.2139/ssrn.2167354
Omer, T. C., Shelley, M. K., and Tice, F. M. (2020). Do director networks matter
for nancial reporting quality? Evidence from audit committee connectedness and
restatements. Manag. Sci. 66, 3361–3388. doi: 10.1287/mnsc.2019.3331
Renneboog, L., and Zhao, Y. (2011). Us knows us in the Uk: on director networks and
Ceo compensation. J. Corp. Finan. 17, 1132–1157. doi: 10.1016/j.jcorpn.2011.04.011
Schabus, M. (2022). Do director networks help managers forecast better? Account.
Re v. 97, 397–426. doi: 10.2308/TAR-2018-0074
Singh, D., and Delios, A. (2017). Corporate governance, board networks and
growth in domestic and international markets: evidence from India. J. World Bus.
52, 615–627. doi: 10.1016/j.jwb.2017.02.002
Sun, Y., Yi, Y., and Lin, B. (2012). Board Independence, internal information
environment and voluntary disclosure of auditors’ reports on internal controls.
China J. Account. Res. 5, 145–161. doi: 10.1016/j.cjar.2012.05.003
Wang, L. (2015). Protection or expropriation: politically connected independent
directors in China. J. Bank. Financ. 55, 92–106. doi: 10.1016/j.jbankn.2015.02.015
Ye, K. (2014). Independent director cash compensation and earnings
management. Journal of Accounting Public Policy 33, 391–400. doi: 10.1016/j.
jaccpubpol.2014.04.001
Cheng et al. 10.3389/fpsyg.2022.1011761
Frontiers in Psychology 15 frontiersin.org
Appendix A. Variable definitions.
Variab l e Denition
Tunne l Other receivables divided by total assets
Networks Comprehensive scores for Degree, Closeness, Betweenness and Eigenvector calculated by principal component analysis. In particular, Networks_media,
Networks_mean and Networks_max are the company-level network connectedness calculated based on the median, mean and maximum values of the
independent directors’ network indicators
Size e natural logarithm of total assets
ROA Net income divided by total assets
Growth Growth in sales for the period compared to sales in the last period
Loss 1 if net prot is less than zero, otherwise 0
Lev Total liabilities divided by total assets
CurrentRatio Current assets divided by current liabilities
Dual 1 if the chairman and CEO of the company are the same person, otherwise 0
Ratio Percentage of independent directors on the board
BTM Book-to-market ratio
SOE 1 if the rm’s ultimate controller is a government-owned entity, otherwise 0
IC e natural logarithm of the Internal Control Index score
InvestorShare Shares held by institutional investors
Big4 1 if the rm is audited by a Big 4 international audit rm, otherwise 0
Opacity 1 if annual corporate transparency rating is A or B, otherwise 0
Available via license: CC BY
Content may be subject to copyright.
Content uploaded by Hanxiu Cheng
Author content
All content in this area was uploaded by Hanxiu Cheng on Nov 17, 2023
Content may be subject to copyright.