International Review of Financial Analysis 36 (2014) 120130

Contents lists available at ScienceDirect

International Review of Financial Analysis

Ownership structure and risk-taking: Comparative evidence from private

and state-controlled banks in China
Yizhe Dong a,, Chao Meng b, Michael Firth c, Wenxuan Hou d
a
School of Management and Business, Aberystwyth University, Aberystwyth SY23 3AL, UK
b
Hongyuan Securities, Beijing, China
c
Department of Finance and Insurance, Lingnan University, Tuen Mun, New Territories, Hong Kong
d
Business School, University of Edinburgh, Edinburgh, United Kingdom

a r t i c l e i n f o a b s t r a c t

Available online 2 April 2014 This study examines the impact of ownership structure on Chinese banks' risk-taking behaviours. We classify the
Chinese commercial banks into three categories based on the types of controlling shareholder, and nd that
JEL classication: banks controlled by the government (GCBs) tend to take more risks than those controlled by state-owned
G21 enterprises (SOECBs) or private investors (PCBs). This is attributed to the severe political intervention and
G28 weak incentives to follow prudent bank management practices for GCBs. We also nd that the results are
G32
more pronounced among banks with concentrated ownership presumably because the large controlling power
G34
helps to enhance the monitoring of the management and promotes prudent operating procedures. Our ndings
Keywords: have important implications for the ongoing reform in the Chinese banking sector.
Bank 2014 Elsevier Inc. All rights reserved.
Risk-taking
State ownership
Ownership concentration
China

1. Introduction 2011), and banks have the ability to take on risk very quickly and in
ways that are not readily visible to directors or investors, thus posing
The Basel Committee on Banking Supervision (BCBS) recently issued a broader risk to the economy than non-nancial rms. To date, howev-
a set of Principles for enhancing sound corporate governance (BCBS, er, corporate governance studies in the literature have largely focused
2010) in the banking sector to discuss the link between governance on non-nancial rms. Therefore, the issue of corporate governance
quality and bank failure as well as economic development. Poor corpo- and risk-taking in the banking sector is of particular interest. To shed
rate governance has been found to motivate excessive risk-taking and light on this issue in the under-researched emerging markets, we
therefore been blamed as a contributory factor of the recent nancial study the role played by the controlling shareholders of Chinese banks
crisis (Laeven & Levine, 2009). The report highlighted some corporate by exploring the impact of their nature and the ownership concentra-
governance challenges including bank ownership structures that are tion on banks' risk-taking behaviours.
unduly complex, lack transparency, or impede appropriate checks and Since 1979, the Chinese authorities have undertaken gradual bank-
balances, and pointed out that Challenges can also arise when insiders ing reforms to address the institutional, political and organizational
or controlling shareholders exercise inappropriate inuences on the bank's problems faced by its banking industry. The speed of the reforms has
activities (2010, p.6). Corporate governance in the banking sector dif- accelerated since 2003, and the Chinese banking sector has been
fers from that in the non-nancial sectors in terms of transparency, dramatically reshaped. The latest round of banking reform measures
business complexity and regulation (Mehran, Morrison, & Shapiro, include nancial capital injections, shareholding restructures, the intro-
duction of foreign strategic investors, the listing of banks' share capital
on foreign and Chinese exchanges, and the establishment of a system
for the boards of directors. These reforms have changed the ownership
Corresponding author. Tel.: +44 1970622508; fax: +44 1970 62240.
E-mail addresses: yid1@aber.ac.uk (Y. Dong), Chaomeng@vip.126.com (C. Meng), structure of Chinese banks, and are expected to improve the governance
marth@ln.edu.hk (M. Firth), wenxuan.hou@ed.ac.uk (W. Hou). quality and have important implications on their behaviours.

http://dx.doi.org/10.1016/j.irfa.2014.03.009
1057-5219/ 2014 Elsevier Inc. All rights reserved.
 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130 121

In China and some other emerging market countries the banking better capital adequacy ratio (Shehzad, de Haan, & Scholtens, 2010).
sector operates under a two-tier ownership structure including state- These differences may partially be attributed to the different settings
owned banks and privately owned (domestic or foreign) banks. Both which embed different institutional features from the various countries
theoretical and empirical studies in the literature suggest that the and regulatory regimes.
performance and risk-taking behaviour of organizations depend on To perform our analysis, we hand collect the ownership information
the identity of the controlling shareholders (i.e., the ultimate owners) of 108 Chinese commercial banks over the period from 2003 to 2011.
(e.g., Barry, Lepetit, & Tarazi, 2011; John, Litov, & Yeung, 2008). In We regress the ownership structure characteristics, including the
terms of state ownership, political interference usually comes at the identity of the controlling shareholder and the ownership concentration
expense of corporate protability because of politicians' deliberate pol- and their interaction terms, on the bank's risk-taking proxies. We also
icy of transferring resources to their supporters (Shleifer, 1998; Shleifer incorporate other corporate governance characteristics as control
& Vishny, 1986). This suggests that state-owned banks might be seen as variables including the independence of the risk committee chair and
vehicles for raising capital to nance projects with high social returns, the proportion of female directors on the boards. We use three catego-
but possibly high-risk and low-prot returns, or to provide nance to ries of ownership identity to reect the nature of their largest share-
favoured groups such as state-owned enterprises (SOEs) (Clarke, Cull, holder: government-controlled banks (GCBs), SOE-controlled banks
& Shirley, 2005).1 State-owned banks nd it difcult to resist such (SOECBs), and privately controlled banks (PCBs). For the ownership
harmful government interference, whereas private banks are more concentration, we use the Herndahl index based on the ownership
able to oppose it, and typically employ more sensible prudential lending shares of the top ten shareholders and the percentage of shares held
policies and/or prot-maximizing strategies as a consequence (Shirley by the three largest shareholders. Our ndings show that SOECBs tend
& Nellis, 1991; Shleifer & Vishny, 1994). Moreover, lower performance to take less risk than GCBs. Unlike GCBs, SOECBs have greater incentives
incentives (Shleifer & Vishny, 1997) and soft budget constraints to pursue prot-maximizing strategies and exercise prudential lending
(Sheshinski & Lopez-Calva, 2003) in state-owned banks also result in practices. We also nd that the effect of controlling shareholders on
excessive risk-taking and the misallocation of resources. bank risk-taking depends on the ownership concentration. More specif-
These theoretical inferences have been supported by some empirical ically, concentrated ownership can reduce risk-taking in SOECBs and
evidence. For example, government-owned banks and large state own- PCBs, but increase risk-taking in GCBs presumably because of their
erships are associated with lower efciency (Bonin, Hasan, & Wachtel, different objectives. Finally, consistent with Aebi, Sabato, and Markus
2005; Fries & Taci, 2005), inferior long-term performance (Berger, (2012), we also nd that the presence of the Chief Risk Ofcer (CRO)
Clarke, Cull, Klapper, & Udell, 2005), greater risk-taking (Angkinand & on the executive team and a greater number of female directors signif-
Wihlborg, 2010; Iannotta, Nocera, & Sironi, 2007, 2013), and less icantly reduce risk-taking.
prudent lending behaviours (Jia, 2009). However, there are also some We believe that our study makes an important contribution to the
contradictory results. State-controlled banks have also been found literature in several ways. First, it adds to the literature of banking
to be associated with less risk in Russia (Fungov & Solanko, 2009) governance by providing original evidence on the impact of two dimen-
and higher efciency in India (Bhattacharyya, Lovell, & Sahay, 1997) sions of ownership structure (i.e., controlling shareholder type and
and Turkey (Isik & Hassan, 2002). Altunbas, Gardener, Molyneux, and ownership concentration) on banks' risk-taking. Some related studies
Moore (2001) nd little evidence that private banks are more efcient either focus on the nature of the bank (Barry et al., 2011; Forssbck,
than state-owned ones in Germany. Overall, the results are not conclu- 2011; Nichols, Wahlen, & Wieland, 2009) or on the degree of ownership
sive and little is known about the role of state controlling shareholders concentration (Iannotta et al., 2013; Laeven & Levine, 2009; Sullivan &
in Chinese banks' risk-taking behaviours.2 Spong, 2007). To the best of our knowledge, this is the rst study to
Besides the nature of the controlling shareholder, another important address how the ownership concentration affects the role of controlling
dimension of banks' ownership structure is ownership concentration shareholders. Second, this study contributes to the growing literature
(Iannotta et al., 2007). Opposite effects of ownership concentration on on emerging markets by exploring the rapidly developing Chinese
rm performance are predicted from theories from the literature. On banking sector from the largest emerging market in the world. The
the one hand, Shleifer and Vishny (1986) and Admati, Peiderer, and existing Chinese banking literature mainly examines the determinants
Zechner (1994) argue that concentrated ownership can overcome the of banks' (accounting) performance or efciency (Berger, Hasan &
free-rider problem and enhance rm performance by improving the Zhou, 2009; Fu & Heffernan, 2007, 2010; Kumbhakar & Wang, 2007;
monitoring of management. An agency problem is created when own- Lin & Zhang, 2009; Zhang, Jiang, Qu, & Wang, 2013), while our study
ership is dispersed because atomistic shareholders bear the full cost of focuses on the risk-taking behaviour of Chinese banks using three risk
monitoring while reaping only a fraction of the benets and therefore measure proxies, i.e., Z-score, non-performing loans, and the capital
have less incentive to monitor the rms. On the other hand, other theo- adequacy ratio. Finally, our ndings have important implications for
retical studies argue that large shareholders may exercise control rights regulators and investors. Our ndings suggest that the transfer of bank
to pursue private benets at the cost to the minority shareholders (La ownership from the government to marketized SOEs helps to improve
Porta, Lopez de Silanes, & Shleifer, 1999; Shleifer & Vishny, 1997). the stability of the banking system.
Mixed empirical evidence is also documented in this literature. Concen- The remainder of the paper is structured as follows. Section 2 intro-
trated ownership has been found to be associated with higher risks duces the institutional background of the Chinese banking sector.
(Laeven & Levine, 2009), higher insolvency risk and greater return vol- Section 3 develops our predictions on the impact of the controlling
atility (Haw, Ho, Hu, & Wu, 2010). In contrast, ownership concentration shareholders. Section 4 presents the research design. Section 5 provides
has been found to be associated with a lower level of risk-taking in the empirical results and Section 6 concludes.
Spanish commercial banks (Garcia-Marco & Robles-Fernndez, 2008),
better loan quality, lower asset risk and a lower insolvency risk 2. Institutional background
(Iannotta et al., 2007) and a lower non-performing loans ratio and
Over the last thirty years, the Chinese authorities have implemented
a series of signicant reforms aimed at transforming the country's
1
Firth, Lin, Liu, and Wong (2009) nd evidence that political connections play a role in banking sector from policy-driven, wholly state-owned and monopolis-
gaining access to bank nance in China. tic to market-oriented and competitive. One important aspect of the
2
There are a number of studies assessing the efciency or other performance of the Chi-
nese banking sector (e.g., Fu & Heffernan, 2007; Shin, Zhang, & Liu, 2007; Berger et al.,
reform is the ownership restructuring of the Chinese banks through
2009), but they do not explore either banks' risk-taking behaviour or the role of control- the introduction of foreign strategic investors, getting listed on stock
ling shareholders in banks. exchanges, and sales of shares to domestic rms. These gradual reforms
122 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130

have created a banking sector with multiple categories of banking Starting in 2003, the authorities restructured rural credit cooperatives
institutions operating in separate market segments with (generally) (RCCs) into rural commercial banks (RCBs). Over the past few years of de-
clearly delineated functions. According to the China Banking Regulatory velopment, RCBs have gradually been transformed from policy-driven,
Commission (CBRC), Chinese banks are classied into wholly state- rural businesses into market-oriented, urban ones. Following rapid ex-
owned policy banks, large-scale (state-owned) commercial banks, pansion in recent years, there were 212 RCBs operating in China at the
joint stock commercial banks (JSCBs), city commercial banks (CCBs), end of 2011, valued at RMB 4.25 trillion, and accounting for 3.75% of
rural commercial banks (RCBs), locally incorporated foreign banks, the total banking institution assets in the country (CBRC, 2011).
and other nancial institutions. Prior studies on Chinese bank ownership structure have generally
One of the main features of the Chinese banking sector is the focused on three types of banks in China, namely state-owned commer-
dominance of the ve largest state-owned commercial banks (SOCBs) cial banks (SOCBs), joint stock commercial banks (JSCBs) and city com-
(known as the Big Five): the Agricultural Bank of China (ABC), the mercial banks (CCBs) (e.g., Garcia-Herrero, Gavila, & Santabarbara,
Bank of China (BOC), the China Construction Bank (CCB), the Industrial 2009; Jia, 2009; Lin & Zhang, 2009). However, there are some problems
and Commercial Bank of China (ICBC) and the Bank of Communications with this classication. For example, China Everbright Bank and China
(BOCOM).3 One of the most important targets of the Chinese authorities Mingsheng Bank are both categorized as JSCBs, but their ownership
is to establish sound corporate governance through the restructuring of structures are very different. The majority of the shares in the former
the SOCBs. Starting in 2003, the Chinese authorities introduced a partial are held by the central government and SOEs, while the latter is fully
privatization strategy to encourage the Big Four (later the Big Five) to owned by private investors. In other words, the issued shares could be
adopt the shareholding ownership structure. Under this strategy, two ap- held by state and private shareholders, or among private ones only,
proaches were initially undertaken. The rst was to sell strategic stakes to and therefore the category of JSCBs fails to reect the difference in the
foreign investors and the second was to list the banks on foreign and Chi- presence or the inuence of state shareholders. To address this issue,
nese exchanges. Although some portion of the shares of the Big Five have we classify the banks based on the nature of the largest shareholders,
been transferred to domestic institutions, foreign investors and the public which is dened as the controlling shareholders who tend to dominate
through these measures, the ultimate voting control of the banks remains the bank's decision-making and control the bank's property by virtue of
with the state. Therefore, the government continues to exert signicant their superior control rights (See Chen, Firth, & Xu, 2009). We classify
inuence over the lending practices and administration of these banks. the Chinese commercial banks into 3 types, namely GCBs, SOECBs, and
There are currently twelve JSCBs with national operating licences, PCBs. The GCBs are banks whose largest shareholders are government
representing the second tier of Chinese banks. Because the JSCBs were agencies, such as central or local government, government bureaus, or
established more recently than the SOCBs, they are not burdened with state asset operating companies. The SOECBs have SOEs as their largest
any historical policy lending (in particular to those relating to non- shareholders. The PCBs' largest shareholders are private rms, foreign
performing loans) and, therefore, tend to be more agile and responsive nancial institutions or individuals. Unlike the government agencies,
to market requirements. JSCBs' shares are distributed among the central SOEs and private rms are prot-making entities. Our ownership classi-
government, the local government, SOEs, and private and foreign cation better reects the institutional features of the Chinese banking
investors.4 The ownership structure varies widely across JSCBs. In sector, and in particular captures the inuence of the prevailing state-
some, such as China Mingsheng Bank and China Zheshang Bank, the related shareholders in China.
majority of the shares are owned by private investors, while in others
the majority are held by the state or SOEs. JSCBs are allowed to offer a
wide variety of banking services, including accepting deposits, extend- 3. Hypothesis development
ing loans, and providing foreign exchange and international transaction
services. They also regularly nance small and medium enterprises 3.1. Government-controlled banks
(SMEs), which tend to be ignored by the Big Five.
Since 1995, city commercial banks (CCBs) have been created For historical reasons, a large proportion of Chinese commercial
through the restructuring and consolidation of urban credit coopera- banks are owned or controlled by the state, either directly through
tives (UCCs). They represent the third category of Chinese banks. Most central or local government agencies or indirectly through marketized
were originally wholly owned or controlled by local government, the SOEs. These two types of state controls are likely to have different
aim being to provide nancial support for local economic development. impacts on Chinese banks, in particular on their risk-taking behaviour.
However, since the new millennium, CCBs have gradually been trans- When government agencies are the controlling shareholders of a
formed into (private) joint-equity corporations with a more diversied bank, its board of directors and senior ofcers will generally be
set of shareholders, including the treasuries of local governments, SOEs, appointed or approved by the government.5 Moreover, the promotions
private enterprises, foreign investors and individuals. However, local and rewards granted to this type of bank's senior managers largely
government is still the largest shareholder in many CCBs. As of 2011, depend on how well they carry out the instructions of the central
we nd that an average of 17.1% of the shares in CCBs was owned by or local government, and less on the creation of bank value (Cao,
local governments. Because of their smaller size, the CCBs have strug- Lemmon, Tian, & Pan, 2011; Chen et al., 2009). The ofcials do not
gled to compete with the Big Five and the JSCBs. However, in recent bear the consequences of any inappropriate decisions they make. There-
years, they have made signicant progress in upgrading their opera- fore, senior ofcers have less incentive to monitor the banks and may
tional and managerial capabilities, as well as promoting innovative not fully comply with prudent bank management practices. The strate-
products and technologies. They have gradually nurtured their own gies and operations of government-controlled banks are more likely to
brands and corporate cultures, and have begun to play a pivotal role be subject to political intervention as they generally serve as policy-
in underpinning the development of small and micro-enterprises and lending conduits for the government to provide loans to SOEs.6 Thus,
the consumer nance business. the incentives to follow prudential risk management rules and to
adhere to commercial objectives are weak for GCBs.
3
BOCOM used to be classed as a JSCB. However, it is much larger than the other JSCBs,
5
and its share ownership is spread among different state-owned entities. Therefore, in The senior bank ofcers of government-controlled banks are generally members of
2006, the CBRC redened it as a SOCB. Thus, it joined the other four big state-owned banks the Chinese Communist Party. They are also likely to rank highly in the Chinese govern-
(previously known as the Big Four) to form the Big Five. For consistency, we treat ment's hierarchy (Martin, 2012).
6
BOCOM as a SOCB rather than a JSCB throughout the entire period of our study. According to Article 34 of the Commercial Banking Law, a commercial bank shall con-
4
SOEs generally invest in JSCBs purely for the expected returns (i.e., dividends and cap- duct its loan business in accordance with the need for the development of the national
ital gains) and do not engage in management activities. economy and social progress under the guidance of the state industrial policy.
 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130 123

We nd that the total loans issued by Chinese banks grew by 95.3% 3.4. Ownership concentration
hitting a record high of RMB 9590 billion in 2009 as part of the Chinese
economic stimulus programme and the majority of these bank loans We argue that the degree of ownership concentration also has a
were lent by the state-controlled commercial banks and driven mainly signicant impact on banks' risk-taking behaviour. Ownership con-
by the policy directives of the central and local governments (CBRC, centration refers to the distribution of the ownership among differ-
2010). As Greenwald and Stiglitz (1993) point out, the granting of ent institutions and individuals and is related to shareholders'
enormous policy-directed loans during an economic downturn is likely controlling power. Previous literature (e.g., Azofra & Santamara,
to increase the riskiness of state-controlled banks. Furthermore, 2011; Iannotta et al., 2007; Shehzad et al., 2010) suggests that the
government-controlled commercial banks generally enjoy the advan- ownership concentration could signicantly affect a bank's perfor-
tage of either implicit or explicit nancial and regulatory support from mance and riskiness. According to Shleifer and Vishny (1986) and
the government (Faccio, Masulis, & McConnell, 2006). For example, Edwards and Nibler (2000), concentrated ownership enhances corpo-
we nd that as part of its efforts to rescue the major state-owned rate control by improving the monitoring of management. Moreover,
banks, the State Council transferred around 1245 billion Yuan in non- dispersed ownership may prevent efcient decision-making (Shehzad
performing loans from the Big Five banks to asset management compa- et al., 2010) and create a free-rider problem in corporate control
nies set up by the government during 20032005 (Okazaki, 2007). This (Gorton & Schmid, 1999). However, greater ownership concentration
governmental protection encourages bankers to take excessive risks as may not reduce a bank's riskiness because the interests of the large
the losses and excess costs are invariably covered by the government shareholders may conict with those of the minority ones (Gomes &
(Demirg-Kunt & Detragiache, 2002). Novaes, 2005).

3.2. SOE-controlled banks

4. Research design
SOE controlling shareholders are different from government
controlling shareholders in many respects. First, SOECBs are more 4.1. Data and sample selection
empowered and have greater autonomy as there is less interference
from the government (Martin, 2012). Unlike the case of government To investigate the impact of ownership structure on banks' risk-
controlled banks, the board of directors and senior ofcers of SOECBs taking, we construct our sample from two sources. We hand-collect
are appointed by SOE controlling shareholders rather than being directly the detailed information about the banks' ownership structures and
appointed by the organization department of the Communist Party. corporate governance, such as the percentage of ownership held by
Chinese SOEs have become market-oriented and are now responsible the top ten shareholders or the independence of the risk committee,
for their own gains and losses after the third round of reforms (Delios, from the banks' annual reports.8 The bank-specic accounting data are
Wu, & Zhou, 2006). Thus, SOE controlling shareholders are motivated extracted from the BankScope database and from the banks' annual
to appoint good managers and to monitor them to ensure that banks reports. After eliminating observations with missing values, our nal
they control operate in a safe and sound manner. Second, SOECBs sample comprises 667 yearly observations covering 108 Chinese com-
have strong nancial constraints and do not receive as much nancial mercial banks over the period from 2003 to 2011. The sample includes
support from the state as government-controlled banks. Therefore, the ve large SOCBs, 12 JSCBs, 84 CCBs, and 7 RCBs, and represents
they have greater incentives to pursue prot-maximizing strategies about 74% of the total assets of Chinese banking institutions at the end
and to exercise prudential lending standards than do GCBs. Finally, of 2011. We explore two major dimensions of ownership structure,
SOECBs generally adopt a performance-related compensation system, namely owner identity and ownership concentration. These two cate-
with the top executives given sufcient incentives to do a good job gories of ownership measures incorporate both the owners' incentives
through monetary rewards. In 2011, for example, the annual report of and the controlling power of shareholders. Table 1 shows the owner-
China CITIC bank, a SOECG, shows an annual CEO remuneration of ship structures of the banks in our sample over the period 20032011.
49.618 million Yuan, 4.5 times higher than the CEO annual remunera- Based on the detailed ownership data, we categorize the banks based
tion in ICBC, the biggest government controlled bank. on the identity of the controlling (largest) shareholder into three
types, namely 349 banks controlled by the state (GCBs), 170 banks
controlled by the SOEs (SOECBs), and 148 controlled by private share-
3.3. Private-controlled banks
holders (PCBs).
The third type of controlling shareholder is a private shareholder.
Generally, banks controlled by private shareholders (PCBs) are more 4.2. Empirical models
prot-motivated than either of the previous two types (GCB and
SOECB). Private controlling shareholders are likely to select senior We employ three proxies to measure the risk-taking of Chinese
managers with a detailed knowledge of the banking industry and the banks. The Z-score, proposed by Boyd and Graham (1986), equals the
capability to maximize the wealth of the shareholders. Moreover, unlike return on assets (ROA) plus the capital to assets ratio (E/A) divided by
SOECBs or GCBs, PCBs can face the threat of a hostile takeover or the the standard deviation of asset returns ((ROA)). It measures the
possibility of bankruptcy because of the lack of implicit government distance to default since it is the inverse of the probability that losses
guarantees. These traits provide a natural incentive for the managers exceed bank capital (i.e., prob ( ROA b E/A)). A higher Z-score indi-
of private banks to install more efcient and prudent operating proce- cates that a bank is more stable and less risky. Since the Z-score is highly
dures than may be seen in the state-owned counterparts (Berglf & skewed, we use its natural logarithm to smooth out the skewness
Roland, 1998; Sheshinski & Lopez-Calva, 2003). However, PCBs may (Laeven & Levine, 2009). The Z-score is commonly used in literature
still allocate loans and resources in ways that deviate from optimal to measure bank risk (e.g., Angkinand & Wihlborg, 2010; Barry et al.,
business practices. They may also face pressure from their private share- 2011; Demirg-Kunt & Huizinga, 2013; Laeven & Levine, 2009).
holders to provide preferential treatment to their companies, families
and/or related companies.7 8
In recent years, most Chinese commercial banks have complied with the People's Bank
of China's (PBOC) 2002 Provisional Rules on Information Disclosure of Commercial Banks,
7
Some of the private shareholders or related companies are SMEs, which generally are which require banks to disclose nancial statements and provide information on their cor-
hard to appraise, light on assets and quick to capsize when economic winds change. There- porate governance and risk management activities in the annual reports. The annual re-
fore, they are perceived as higher risk than larger companies such as SOEs (Martin, 2012). ports are either published on the banks' websites or are available upon request.
124 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130

Table 1
The ownership structure of Chinese banks, 20032011.

State control (Obs. 349) SOE control (Obs. 170) Private control (Obs. 148) ALL (Obs. 667)

Mean SD Min Max Mean SD Min Max Mean SD Min Max Mean SD Min Max

Largest shareholder 0.2730 0.2230 0.042 1 0.2279 0.1549 0.0644 0.87 0.1802 0.1687 0.0481 0.9075 0.2409 0.1993 0.042 1
Second shareholder 0.1085 0.0761 0 0.5 0.1287 0.0563 0.0459 0.2667 0.0959 0.0454 0.0054 0.2267 0.1109 0.0665 0 0.5
Third shareholder 0.0679 0.0338 0 0.1935 0.0795 0.0382 0.0125 0.2 0.0728 0.035 0.0049 0.1778 0.0720 0.0355 0 0.2
Ownership of top three shareholders 0.4495 0.2255 0.0707 1 0.4362 0.1789 0.1616 0.9785 0.3490 0.1653 0.1269 0.9194 0.4238 0.2058 0.0707 1
HHI 0.1593 0.2138 0.0033 1 0.1166 0.1251 0.0141 0.7598 0.0936 0.1549 0.0097 0.8237 0.1338 0.1842 0.0033 1
Total ownership by all block holders 0.5401 0.2379 0 1 0.5253 0.2258 0.0644 1 0.4623 0.2315 0 0.9075 0.5191 0.2352 0 1
No. of blockholders 4.0516 2.3825 0 10 4.0647 1.8024 1 9 4.3378 2.5270 0 10 4.1184 2.2867 0 10

Notes: This table shows the percentage of a bank's share capital owned by the largest, second largest, third largest, and largest three shareholders. HHI is the Herndahl index, which is
based on the ownership held by the ten largest shareholders of the bank. Blockholders are dened as shareholders that hold 5% or more of the shares of a bank.

We also use the non-performing loan ratio and the capital adequacy value of the Herndahl index, the more concentrated is the ownership
ratio (CAR) by following Demirg-Kunt, Detragiache, and Tressel of the bank. The CR3 variable is dened as the sum of the percentage
(2006), Shehzad et al. (2010), and Delis and Kouretas (2011). The of shares owned by the largest three shareholders and is used as an al-
non-performing loan ratio, calculated as the ratio of non-performing ternative ownership concentration measure for robustness checks
loans to total loans, reects the quality of a bank's assets.9 Because (Demsetz & Villalonga, 2001; Omran, 2009).
non-performing loans cause losses for banks, a higher non-performing CONTROL denotes a set of control variables including governance,
loan ratio is associated with higher credit risk (Delis & Kouretas, bank-specic, and macroeconomic variables that may affect a bank's
2011). The capital adequacy ratio (CAR) is the ratio of a bank's capital risk-taking behaviour. We incorporate a dummy variable to capture
to its risk-weighted assets. The level of bank equity provides a cushion whether the bank's chief risk ofcer (CRO), who oversees all relevant
against its portfolios' losses and nancial distress, therefore, the CAR is bank risk, is a member of the executive team and, if so, we expect
closely related to a bank's insolvency risk (see Berger & Mester, 1997; them to have greater authority and inuence to reduce the amount of
Hughes & Mester, 2012; Mester, 1997). Moreover, lower capitalization risk a bank takes (Aebi et al., 2012; Mongiardino & Plath, 2010). We
also reects more severe agency problems between shareholders and dene the risk management committee independence based on
managers, and implies higher-risk bank strategies (Shehzad et al., whether the committee chair is an independent board member (Aebi
2010). et al., 2012; Ellul & Yerramilli, 2013), and expect the independent status
To examine the impact of ownership structure and risk management- could enhance the transparency of the bank's risk management infor-
related governance factors on the risk-taking of banks in China, we use mation. We also use the proportion of female directors on a bank's
the following model10: board to explore whether a higher representation of female directors
could reduce the amount of risk a bank takes (e.g., Almazan & Suarez,
k 2003; Berger, Kick & Schaeck, 2014; Fields, Fraser, & Subrahmanyam,
BRTit 1 SOECBit 2 PCBit CONCit k k CONTROLit it
2012; Pathan, 2009).
1 With regard to other control variables, bank size is measured by the
natural logarithm of the bank's total assets (Delis & Kouretas, 2011;
where the dependent variable BRT is one of the three bank risk-taking Laeven & Levine, 2009). Large banks could be less risky due to their
measures: the natural logarithm of the Z-score (LnZ-score), the non- greater ability to diversify risk across product lines or could be more
performing loan ratio (NPL), or the risk-weighted CAR. The dummy risky due to the implicit assumption that they are too-big-to-fail
variable SOECB is set equal to 1 for SOE-controlled banks where the (Brown & Din, 2011; Demirg-Kunt & Huizinga, 2013). The cost-to-
controlling shareholders are SOEs and 0 otherwise. PCB is set equal to income ratio, dened as the ratio of total expenses to operating income
1 for private-controlled banks where the controlling shareholders are (interest and non-interest income), reects operations both on and off
private entities or individuals, and 0 otherwise. The government the balance sheet. It is expected to be negatively related to a bank's
controlled banks (GCBs) serve as the benchmark and omitted category. risk because less efcient banks are likely to take on greater risk to
Because they can obtain political and nancial support in the event of a generate prots (Agoraki, Delis, & Pasiouras, 2011; Boyd, Gianni, &
nancial crisis, we hypothesize that GCBs would take more risks than Jalal, 2006). The ratio of total loans to total deposits assesses the extent
other types of banks. We therefore expect the coefcients on SOECG to which customer loans are nanced by customer deposits, and is
and PCB to be signicantly positive in the regression models of related to the bank's liquidity. As diversication may be related to a
Z-score and CAR regressions, and signicantly negative in the regres- bank's risk level, we control for the banks' diversication activities
sion model of non-performing loan ratios. CONC denotes one of the using Laeven and Levine's (2007) asset diversity measure (as dened
two measures of ownership concentration, namely the ownership in the Appendix). Berger et al. (2009) and Hasan and Xie (2012) suggest
HerndahlHirschman index (HHI) or the top three shareholders' that foreign strategic investment improves the corporate governance of
concentration ratio (CR3). The HHI equals the sum of the squared own- Chinese banks and reduces their risk-taking. We use a dummy variable
ership shares of the ten largest shareholders of the bank (Demsetz & to capture whether a bank has foreign strategic investment and expect
Lehn, 1985; Hou, Lee, Stathopoulos, & Tong, 2013). The higher is the that it will help the bank to control its risk. At the macroeconomic level,
we include the real GDP growth rate to control for the general economic
environment in China over the sample period, and a dummy variable of
9
According to the ve-category loan classication system that was adopted by Chinese nancial crisis (years 20082011) to capture the impact of the recent
banks in 2002, performing loans include normal and special mention loans and non- nancial crisis (Aebi et al., 2012; DeYoung, Peng, & Yan, 2013). The Ap-
performing loans consist of sub-standard, doubtful and loss loans. pendix presents a summary of the variable denitions and data sources.
10
The models are estimated by using ordinary least squares (OLS). However, we also
In addition to the OLS regression estimation, we also use the two-step
employ the system Generalised Method of Moments (GMM) to address potential
endogeneity problems in robustness checks. Details of the GMM estimation results are system generalized method of moment (GMM) estimation to address
given in Section 5.2. the possible endogeneity problems as a robustness check.
 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130 125

We also argue that the role played by the various types of controlling the coefcients of the square terms of each the ownership variables,
shareholders in banks' risk-taking behaviours is conditional on their 4, 5 and 6, are statistically signicant, this would suggest a non-
power and their incentives as reected by the ownership concentration, linear effect of ownership of controlling shareholders.
i.e., the types of controlling shareholder affect the relation between risk-
taking and the degree of ownership concentration. To explore this issue,
4.3. Descriptive statistics
we use the following regression model:
Table 2 shows summary statistics for the variables that are used in
BRTit 1 SOECBit 2 PCBit 1 CONCit 2 CONCit  SOECBit
k the study. The mean (median) of Z-score is 14.73 (12.24) which is
3 CONCit  PCBit k k CONTROLit it
comparable with the gure given by Zhang, Wang, and Qu (2012).
2 The mean (median) NPL ratio is 2.62% (1.46%), with a large degree of
variation across banks. The capital adequacy ratio ranges from 52.15%
where two interaction terms are incorporated in the model as the to 21.70%, with an average of 11.52%. Regarding the ownership
products of the ownership concentration (CONC) with the ownership variables, the average value of the state-controlled banks is 0.5232, indi-
type dummy variables SOECB and PCB, respectively. Other variables cating that the state is the largest single shareholder in about 52.32% of
are as dened previously. If the coefcients of the interaction terms, cases. In our sample, 25.49% and 22.18% of banks are controlled by SOEs
2 and 3, are statistically signicant, this will suggest that the impacts and private companies, respectively. 29.53% of our sample banks have a
of ownership concentration on risk-taking varies across ownership strategic foreign investment. The mean (median) of the Herndahl
type. index is 0.1393 (0.0700). The mean (median) of the percentage of shares
Finally, we explore whether the ownership of the largest shareholder owned by the top three shareholders is 42.4% (37.46%), indicating a
has a non-monotonic impact on the bank's risk-taking activity. On the strong ownership concentration in the Chinese banking sector. With
one hand, a controlling shareholder with a greater proportion of shares regard to the control variables, we nd that 12.14% of the banks in our
in a bank would have a stronger incentive and more power to monitor sample have a CRO in their executive team. This gure is comparable
the management and thus reduce the amount of risk the bank takes. On to the gure reported in Aebi et al. (2012) using U.S. data. About
the other hand, the dominant shareholder makes it possible to expro- 17.69% of the banks in our sample have an independent risk committee.
priate funds from its controlled bank and increase the risks of the The proportion of the female directors are about 10%, which are slightly
bank. In addition, some studies in the literature argue that ownership higher than those of U.S. bank holding companies as reported in Pathan
concentration above a certain level would allow larger shareholders to and Faff (2013). Finally, only 4.4% of banks in our sample are listed on
become entrenched and expropriate the wealth of minority share- stock exchanges (Table 2).
holders (e.g., Chen, Firth, & Rui, 2006; Gul, Kim, & Qiu, 2010; Loderer
& Martin, 1997; Wei, Xie, & Zhang, 2005). Therefore, there could be a
non-linear relation between ownership concentration (ownership 5. Empirical analysis
held by the largest shareholder) and bank risk-taking. To explore such
possibilities, we use the following regression model: 5.1. OLS estimation results

2
BRTit 1 GCOit 2 SOECOit 3 PCOit 4 GCOit 5 SOECOit
2 Table 3 presents the regression results from using the natural
logarithm of the Z-score as the dependent variable (a higher value
2
X k 3
6 PCOit k CONTROLit it of the Z-score indicates less risk-taking). All the models are estimated
k by ordinary least squares (OLS) techniques with clustering at the
bank level. Robust standard errors are used to correct potential
where GCO, SOECO and PCO represent the percentage of shares held by heteroskedasticity and potential time series autocorrelation within
the controlling shareholders in GCBs, SOECBs, and PCBs respectively. If each bank. Column 1 reports the results of Eq. (1). The coefcients on

Table 2
Summary sample statistics.

Mean Median STDEV. Min. Max.

Z-score 14.7296 14.2411 5.4247 0.0400 36.6580

Non-performing loan ratio (NPL ratio) 0.0262 0.0146 0.0359 0.0000 0.3066
Capital adequacy ratio (CAR) 0.1152 0.1162 0.0508 0.2170 0.5215
Bank size (LnTA) 4.8673 4.6654 0.7927 3.3287 7.1760
Cost/income (EFFICIENCY) 0.6048 0.5986 0.0935 0.32038 1.0738
Asset diversity (ADIV) 0.6468 0.6359 0.1864 0.0553 1.0000
Loans/deposits (TL/TD) 0.6478 0.6667 0.1178 0.2062 0.9259
Listed bank (LIST) 0.0439 0.0000 0.0778 0.0000 1.0000
Economic growth (GDP) 0.1070 0.1040 0.0170 0.0920 0.1420
Post-global nancial crisis (Post-GFS) 0.6192 1.0000 0.4859 0.0000 1.0000
Government-controlled banks (GCBs) 0.5232 1.0000 0.4998 0.0000 1.0000
SOE-controlled banks (SOECBs) 0.2549 0.0000 0.4369 0.0000 1.0000
Private controlled banks (PCBs) 0.2218 0.0000 0.4158 0.0000 1.0000
Foreign strategic investment (FSI) 0.2953 0.0000 0.4565 0.0000 1.0000
Ownership ratio of the government-controlled ownership (GCO) 0.2730 0.0700 0.2230 0.042 1
Ownership ratio of the SOE-controlled shareholders (SOECO) 0.2279 0.0000 0.1549 0.0644 0.87
Ownership ratio of the Privately controlled shareholders (PCO) 0.1802 0.0000 0.1802 0.1687 0.0481
Ownership Herndahl index (HHI) 0.1339 0.0731 0.1842 0.0033 1.0000
Ownership concentration ratio (CR3) 0.4240 0.3746 0.2061 0.0070 1.0000
CRO on executive team (CRO) 0.1214 0.0000 0.3268 0.0000 1.0000
Independent risk committee (IRMC) 0.1769 0.0000 0.3819 0.0000 1.0000
% female directors (FEMALE) 0.0977 0.0800 0.0927 0.0000 0.4286

Note: all variables are dened in the Appendix.

126 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130

Table 3 after obtaining a listing (presumably because of the market discipline

The impact of ownership structure on the Z-score. and scrutiny).
Dependent variable (1) (2) (3) (4) As reported in regression 2, the coefcients on HHI*SOECB and
(LnZ-score) HHI*PCB are both signicantly positive, showing that concentrated
Constant 5.0929 4.9384 5.1047 4.7383 ownership in SOECBs and PCBs enhances risk control. This is in line
(16.19) (16.85) (17.39) (18.83) with the literature that blockholders help to enhance governance
Ln(TA) 0.2851 0.2555 0.2853 0.2780 quality (e.g., Shehzad et al., 2010; Shleifer & Vishny, 1986). Since the
(3.85) (3.55) (3.96) (3.76)
coefcient on PCB becomes insignicant in regression 2, it is worth
EFFICIENCY 1.4722 1.4991 1.470 1.4745
(6.57) (6.79) (6.78) (5.63) noting that the impact of private controlling shareholders to reduce
ADIV 0.2113 0.2103 0.2108 0.1462 risks is only documented among banks with concentrated ownership.
(1.04) (0.99) (1.06) (0.90) In regressions 3 and 4, we use the ownership ratio of the top three
TL/TD 0.1702 0.2595 0.1663 0.0995 shareholders (CR3) to replace HHI as robustness checks and nd that
(0.86) (1.29) (0.87) (0.48)
the results remain consistent.
GDP growth 5.6249 5.504 5.6229 5.7084
(4.95) (4.85) (5.36) (5.63) Table 4 presents the results of the regressions with the non-
FSI 0.0105 0.0165 0.0089 0.0314 performing loan ratio (the NPL ratio) as the dependent variable (a
(0.09) (0.14) (0.07) (0.25) lower value of the NPL ratio indicates less risk-taking). The coefcients
LIST 0.2855 0.2630 0.2845 0.4106
on SOECB (0.0105) and PCB (0.0104) are negative and statistically
(1.70) (1.51) (1.66) (2.11)
Post-GFS 0.2048 0.2402 0.2084 0.1255 and economically signicant as shown in regression 1. For a typical bank
(3.36) (3.34) (3.23) (2.59) with the median level of non-performing loan (1.46%), the NPL ratio
SOECB 0.0368 0.0432 0.0587 0.0603
(2.53) (1.69) (2.97) (2.10)
PCB 0.0303 0.0218 0.0197 0.0113
Table 4
(1.92) (1.42) (1.48) (1.36)
The impact of ownership structure on the non-performing loan ratio.
HHI 0.2764 0.2357
(2.05) (1.86) Dependent variable (1) (2) (3) (4)
CR3 0.3182 0.2451 (NPL ratio)
(1.80) (1.69)
IRMC 0.0649 0.0532 0.0665 0.0929 Constant 0.1185 0.1076 0.1511 0.1505
(0.99) (0.80) (0.95) (1.51) (3.52) (3.86) (3.09) (3.23)
CRO 0.2243 0.2246 0.2258 0.1754 Ln(TA) 0.0189 0.0077 0.0172 0.0150
(2.49) (2.52) (2.50) (2.62) (2.03) (1.79) (2.14) (2.14)
FEMALE 0.5199 0.5129 0.5236 0.4835 Efciency 0.1193 0.1213 0.1268 0.1304
(2.01) (1.95) (1.99) (1.79) (5.19) (5.39) (4.77) (4.80)
HHI*SOECB 0.8675 ADIV 0.0066 0.0037 0.0109 0.0083
(1.74) (0.49) (0.31) (0.79) (0.132)
HHI*PCB 0.5877 TL/TD 0.0216 0.0167 0.0168 0.0173
(1.99) (1.25) (0.97) (1.02) (1.01)
CR3*SOECB 0.6879 GDP growth 0.1869 0.1773 0.2344 0.2171
(2.32) (2.29) (2.13) (2.13) (2.47)
CR3*PCB 0.4168 FSI 0.0011 0.0021 0.0056 0.0048
(2.02) (0.20) (0.43) (0.87) (0.79)
No. of observations 667 667 667 667 LIST 0.0174 0.0170 0.0223 0.0231
R2 0.2070 0.2141 0.2081 0.2336 (2.24) (2.26) (2.33) (2.40)
F-statistics 10.67 9.46 10.63 9.52 Post-GFS 0.0045 0.0049 0.0038 0.0043
(1.68) (1.69) (1.93) (1.70)
Notes: This table reports the results from ordinary least squares (OLS) with clustering at SOECB 0.0105 0.0051 0.0123 0.0054
the bank level. All variables are dened in the Appendix. Numbers in parentheses are (2.69) (1.24) (2.98) (0.60)
t-statistics, computed using heteroskedasticity-robust standard errors. PCB 0.0104 0.0039 0.0107 0.0076
Indicates estimations that are signicant at 10% level. (1.92) (0.89) (2.20) (0.63)
Indicates estimations that are signicant at 5% level. HHI 0.0495 0.0789
Indicates estimations that are signicant at 1% level. (1.78) (2.18)
CR3 0.0298 0.0347
(1.97) (1.80)
IRMC 0.0153 0.0142 0.0173 0.0172
(3.09) (3.14) (2.83) (2.83)
SOECB and PCB are signicantly positive, indicating that banks con- CRO 0.0092 0.015 0.0094 0.0089
trolled by SOEs and private investors tend to take less risk than banks (2.31) (2.55) (2.36) (2.45)
controlled by government agencies. The coefcient on HHI (Herndahl FEMALE 0.0381 0.0299 0.0423 0.0376
(1.76) (1.55) (1.82) (1.73)
index) is signicantly negative as 0.2764 suggesting that higher
HHI*SOECB 0.0887
ownership concentration could promote risk-taking. Furthermore, we (2.00)
nd that the coefcient on CRO to be signicantly positive indicating HHI*PCB 0.1086
that CROs at the executive level help banks to improve their risk control. (2.75)
The coefcient on board gender diversity (FEMALE) is signicantly pos- CR3*SOECB 0.0370
(1.83)
itive suggesting that including more female directors on boards could CR3*PCB 0.0486
help to restrain risk-taking behaviours. Our ndings support those of (2.28)
Almazan and Suarez (2003) and Farrell and Hersch (2005). The inde- No. of observations 667 667 667 667
pendence of the risk committee (IRMC) is found to have no impact on R2 0.3456 0.3874 0.3129 0.3219
F-statistics 5.38 4.81 4.96 4.32
the Z-score. Regarding other control variables, the coefcients are sig-
nicant and negative for bank size (LnTA) and the cost to income ratio Notes: This table reports the results from ordinary least squares (OLS) with clustering
(EFFICIENCY) showing that larger and more inefcient banks tend to at the bank level. All variables are dened in the Appendix. Numbers in parentheses are
t-statistics, computed using heteroskedasticity-robust standard errors.
take more risks. The coefcients are signicant and positive for Post- Indicates estimations that are signicant at 10% level.
global nancial crisis (Post-GFS) and listing status of banks indicating Indicates estimations that are signicant at 5% level.
that banks tend to take less risk after the global nancial crisis and Indicates estimations that are signicant at 1% level.
 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130 127

decreases by around 70% to 0.41% when the controlling shareholder in women on the board all help improve banks' risk management and
GCBs is replaced by either a SOE or private company. This is consistent reduce their non-performing loans.
with our prediction that the SOECBs and PCBs rather than GCBs tend Table 5 presents the results based on the third risk measure of risk-
to take less risk. The coefcients on ownership concentration measures, taking, namely the capital adequacy ratio (CAR). A higher value of CAR
HHI and CR3, are signicantly positive across all models. The results are indicates less risk-taking. The coefcients on SOECB are signicantly
also economically signicant suggesting that banks with concentrated positive in regressions 1, 3, and 4, while the coefcients on PCB are
ownership tend to pursue risk-taking activities. not signicant. This indicates lower risk-taking in SOECBs, but not in
When we incorporate the interaction terms of controlling share- PCBs, when compared with the GCBs. The coefcients on HHI and CR3
holders' identity and ownership concentration in regressions 2 and 4, are statistically signicant and negative showing that concentrated
the signicant and negative coefcients show that a higher degree of ownership helps to increase CAR. When the interaction terms of the
ownership concentration further reduces the non-performing loans of controlling shareholder identity and ownership concentration are
SOECBs and PCBs, leading to higher asset quality. Since the coefcients incorporated in regressions 2 and 4, the coefcients for HHI*SOECB
on SOECB and PCB become insignicant in regressions 2 and 4, the and CR3*SOECB are signicant and positive while these for SOECB itself
impact of private and SOE controlling shareholders to reduce risk- become insignicant. This shows that the positive impact of SOE
taking is only pronounced in banks with concentrated ownership. controlling shareholders on increasing CAR and reduced risks are only
Finally, the coefcients on IRMC, CRO and FEMALE are negative and pronounced among banks with concentrated ownership. Similarly, we
signicant showing that the independence of the risk committee, the nd that the coefcients for HHI*PCB are positive and statistically
presence of the CRO on the executive team, and a greater number of signicant at the 5% level, suggesting that the PCBs with concentrated
ownership tend to take less risk than those with dispersed ownership.
Table 6 reports the results of Eq. (3) that examines how the owner-
ship ratio of various types of controlling shareholder affects the banks'
risk taking behaviours. For each of the risk measures, we run two
Table 5 types of OLS regressions, one (regressions 1, 3 and 5) intended to cap-
The impact of ownership structure on the capital adequacy ratio.
ture the linear impact, and one (columns 2, 4 and 6) intended to capture
Dependent variable (CAR) (1) (2) (3) (4) the non-linear impact by including squared terms GCO2, SOECO2, and
Constant 0.3898 0.3653 0.4182 0.4068 PCO2. The coefcients on GCO are signicantly positive in the regression
(8.76) (10.01) (6.97) (7.20) of NPL ratio (regression 3) and negative in the regression of CAR
Ln(TA) 0.0249 0.0201 0.0308 0.0267 (regression 5), suggesting that the higher the ownership ratio of the
(3.77) (4.11) (3.20) (3.22) government controlling shareholders, the higher the risks taken by
Efciency 0.1382 0.1425 0.1455 0.1516
(4.51) (4.66) (4.29) (4.37)
the banks. In regressions 4 and 6, the coefcients on the quadratic
ADIV 0.0320 0.0559 0.0362 0.0359 terms (GCO2) are statistically signicant and positive, while the coef-
(2.04) (2.70) (2.21) (2.26) cients on the level terms (GCO) become insignicant. The results
TL/TD 0.0698 0.0687 0.0645 0.0571 suggest that when the government-controlled ownership is at a low
(3.43) (3.25) (3.39) (2.92)
level, the controlling shareholder does not signicantly affect the bank's
GDP growth 0.5659 0.541 0.6111 0.5843
(5.19) (5.29) (5.05) (4.99) risk taking behaviours. However, a further increase in government own-
FSI 0.0088 0.0079 0.0128 0.0113 ership concentration may create control ambitions and the capability
(1.39) (1.26) (1.71) (1.60) for the controlling shareholder to expropriate minority shareholders
LIST 0.0283 0.0246 0.337 0.0340 by taking extra risks. In addition, we nd that the coefcients on
(2.83) (2.73) (2.82) (2.90)
Post-GFS 0.0092 0.0105 0.0126 0.0908
SOECO are statistically signicant but not the coefcients of SOECO2,
(2.10) (2.41) (2.24) (1.85) suggesting a linear relationship between SOE ownership concentration
SOECB 0.0139 0.0028 0.0126 0.0197 and bank risk taking. Higher ownership concentration increases the
(2.03) (0.41) (1.97) (1.76) incentive and power for a SOE controlling shareholder to monitor
PCB 0.0032 0.0032 0.0038 0.0156
management and reduce bank risk-taking: the larger is the ownership
(0.65) (0.72) (0.86) (1.30)
HHI 0.0645 0.0749 of the SOE controlling shareholder, the less risk the banks take. Regard-
(1.98) (2.02) ing private controlling ownership, we nd that the coefcients on the
CR3 0.0366 0.0371 quadratic terms (PCO2) are statistically signicant and positive only in
(1.73) (1.80) the Z-score model, suggesting that high levels of private controlling
IRMC 0.0122 0.0102 0.0137 0.0120
(1.79) (1.63) (1.78) (1.55)
ownership help banks to control their risks in terms of the Z-score.
CRO 0.0172 0.0182 0.0171 0.0167 Finally, the results also show that a higher foreign ownership could
(2.94) (2.99) (3.04) (3.11) reduce banks' risk-taking.
FEMALE 0.0496 0.0422 0.0528 0.0486
(1.87) (1.70) (1.86) (1.82)
HHI*SOECB 0.1389
(2.30) 5.2. Robustness checks
HHI*PCB 0.0989
(2.13) This section intends to address the concern of endogeneity. We rst
CR3*SOECB 0.0729 argue that our setting is unlikely to be subject to the endogeneity
(2.79)
CR3*PCB 0.0492
problem because the type of controlling shareholders of Chinese
(1.53) banks is exogenously determined by the Chinese regulatory authorities
No. of observations 667 667 667 667 according to the agenda of the reform and relevant policies. However, as
R2 0.3416 0.3643 0.3290 0.3406 a robustness check we apply the system generalized method of
F-statistics 6.01 7.29 5.64 5.33
moments (GMM) to further mitigate the concern about endogeneity.
Notes: This table reports the results from ordinary least squares (OLS) with clustering The system GMM estimation results for Eq. (1) are presented in
at the bank level. All variables are dened in the Appendix. Numbers in parentheses are Table 7. The results show that SOE controlled banks tend to take less
t-statistics, computed using heteroskedasticity-robust standard errors.
Indicates estimations that are signicant at 10% level. risk than government controlled banks and concentrated ownership
Indicates estimations that are signicant at 5% level. leads to more risk-taking behaviours. Private controlled banks also
Indicates estimations that are signicant at 1% level. tend to control the risk by reducing the non-performing loans.
128 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130

Table 6
The impact of ownership ratios of controlling shareholders on banks' risk-taking behaviour.

Models (1) (2) (3) (4) (5) (6)

Dependent variables Ln(Z-scores) NPL ratio CAR

Constant 4.8039 4.7862 0.0975 0.0709 0.3379 0.3152

(16.15) (15.08) (3.52) (3.22) (9.14) (12.06)
Ln(TA) 0.2019 0.1947 0.0048 0.0013 0.0140 0.0103
(3.68) (3.70) (1.23) (0.38) (3.16) (3.41)
Efciency 1.5083 1.4994 0.1283 0.1241 0.1470 0.1433
(7.12) (7.07) (4.93) (5.46) (4.20) (4.53)
ADIV 0.0916 0.0956 0.0019 0.0046 0.0288 0.0214
(0.62) (0.0.57) (0.16) (0.42) (1.89) (1.54)
TL/TD 0.1425 0.2362 0.0135 0.0102 0.0548 0.0511
(0.74) (0.109) (0.80) (0.59) (2.86) (2.40)
GDP growth 5.0587 5.1367 0.1495 0.0054 0.4963 0.5045
(4.27) (4.37) (1.86) (3.71) (4.44) (4.73)
Post-GFS 0.1128 0.0943 0.0056 0.0045 0.0067 0.0058
(2.75) (1.90) (2.12) (1.66) (1.74) (1.52)
GCO 0.2163 0.5366 0.0614 0.0704 0.0601 0.0424
(0.95) (1.03) (2.42) (1.57) (2.05) (0.79)
SOECO 0.3653 1.1357 0.0508 0.0743 0.0385 0.0351
(2.03) (1.86) (2.05) (2.04) (1.92) (0.56)
PCO 0.0688 0.8191 0.084 0.0485 0.0292 0.0353
(0.37) (1.40) (0.38) (0.84) (1.11) (0.67)
FSO 0.6060 0.7266 0.0351 0.0969 0.0799 0.0727
(1.76) (1.92) (1.32) (1.08) (2.32) (2.18)
GCO2 0.1896 0.1621 0.1359
(0.26) (2.22) (1.80)
SOECO2 1.8186 0.0785 0.0241
(1.34) (1.50) (0.27)
PCO2 1.2565 0.0263 0.0471
(2.07) (0.43) (0.89)
IRMC 0.0678 0.0669 0.0157 0.0108 0.0117 0.0077
(0.98) (1.09) (2.83) (2.98) (1.58) (1.32)
CRO 0.1871 0.1978 0.0079 0.0074 0.0143 0.0143
(2.45) (2.62) (1.85) (1.93) (2.17) (2.27)
FEMALE 0.5289 0.5489 0.0367 0.0239 0.0500 0.0393
(2.01) (2.09) (1.72) (1.25) (1.87) (1.58)
No. of observations 667 667 667 667 667 667
R2 0.1989 0.2095 0.3306 0.3908 0.3292 0.3567
F-statistics 12.16 10.23 4.78 5.3 6.46 9.39

Notes: This table reports the results from ordinary least squares (OLS) with clustering at the bank level. All variables are dened in the Appendix. Numbers in parentheses are t-statistics,
computed using heteroskedasticity-robust standard errors.
Indicates estimations that are signicant at 10% level.
Indicates estimations that are signicant at 5% level.
Indicates estimations that are signicant at 1% level.

Overall, the GMM results are generally consistent with the main compared with GCBs. The results support our argument that the incen-
ndings.11 tives to engage in prudent lending practices and to adhere to commer-
cial objectives are weak for banks controlled by the government. We
6. Conclusion also nd that the effect to reduce risks by the SOE and private control-
ling shareholders is more pronounced among banks with concentrated
The Principles for enhancing sound corporate governance (BCBS, ownership.
2010) issued by the Basel Committee on Banking Supervision (BCBS) Our ndings should have relevance for the work of policy
highlighted the possible inappropriate inuence from the controlling makers. Market-oriented SOEs seem to be more efcient control-
shareholders of banks and the serious consequence of excessive risk- ling shareholders for rms in countries with weak institutional
taking of banks on the economy. The ongoing reform in the banking environments (Stiglitz, 1999). Therefore, an important policy im-
sector in China also largely focuses on the ownership structure in plication of this study is that the Chinese government should con-
order to improve the efciency and prudence of Chinese banks. To tinue to transfer its bank ownership to marketized SOEs as this
shed light on this interesting yet under-researched issue, our study helps to improve the stability of the Chinese banking system. Sec-
examines how ownership structure and risk management-related cor- ondly, because concentrated ownership has been found to further
porate governance inuence the risk-taking behaviour of Chinese promote risk control in SOECBs and PCBs, Chinese banking regula-
banks. Based on the type of controlling shareholder, we classify Chinese tors should be cautious about the dispersed ownership of banks. In
commercial banks into government-controlled banks (GCBs), SOE- the weak governance environment of China, investors with dis-
controlled banks (SOECBs), and privately controlled banks (PCBs). Our persed ownerships could nd it difcult to exert an impact on
empirical results show that SOECBs and PCBs take less risk when bank management. Finally, this study nds evidence that listing
banks on the stock market could enhance their governance proce-
dures and reduce the level of risk-taking. Chinese authorities
should encourage banks to list on the stock exchanges, which
11
The untabulated GMM estimation results for Eqs. (2) and (3) remain consistent with exert effective discipline over bank management and in turn re-
our main ndings and are available upon request. strains bankers from taking excessive risks.
 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130 129

Table 7 References
Robustness checks with generalized method of moment (GMM) estimates.

Dependent variables Ln(Z-scores) NPL ratio CAR Admati, A.R., Peiderer, P., & Zechner, J. (1994). Large shareholder activism, risk sharing,
and nancial market equilibrium. Journal of Political Economy, 102(6), 10971130.
Constant 1.8184 0.2456 0.0158 Aebi, V., Sabato, G., & Markus, S. (2012). Risk management, corporate governance, and
(0.86) (2.04) (0.11) bank performance in the nancial crisis. Journal of Banking & Finance, 36(12),
Ln(TA) 0.0567 0.3174 0.0671 32133226.
(0.16) (1.78) (2.49) Agoraki, M. E. K., Delis, M.D., & Pasiouras, F. (2011). Regulations, competition and bank
Efciency 0.9491 0.1718 0.0729 risk-taking in transition countries. Journal of Financial Stability, 7(1), 3848.
(3.92) (1.20) (1.26) Almazan, A., & Suarez, J. (2003). Entrenchment and severance pay in optimal governance
ADIV 0.7500 0.0645 0.0015 structures. Journal of Finance, 58, 519547.
(0.99) (1.30) (0.04) Altunbas, Y., Gardener, E. P.M., Molyneux, P., & Moore, B. (2001). Efciency in European
TL/TD 2.5832 0.2354 0.1531 banking. European Economic Review, 45(10), 19311955.
(1.67) (1.10) (1.97) Angkinand, A., & Wihlborg, C. (2010). Deposit insurance coverage, ownership, and banks'
risk-taking in emerging markets. Journal of International Money and Finance, 29(2),
GDP growth 7.0498 0.2182 0.0901
252274.
(1.98) (1.09) (0.34)
Azofra, V., & Santamara, M. (2011). Ownership, control, and pyramids in Spanish
FSI 0.0685 0.0039 0.0654
commercial banks. Journal of Banking & Finance, 35(6), 14641476.
(0.16) (0.22) (1.36) Barry, T. A., Lepetit, L., & Tarazi, A. (2011). Ownership structure and risk in publicly held
LIST 0.8512 0.0443 0.3619 and privately owned banks. Journal of Banking & Finance, 35, 13271340.
(2.29) (2.13) (1.08) Basel Committee on Banking Supervision (2010). Principles for enhancing sound corporate
Post-GFS 1.518 0.0050 0.0063 governance. Basel: BIS.
(1.39) (0.76) (0.73) Berger, A. N., Clarke, G. R. G., Cull, R., Klapper, L., & Udell, G. F. (2005). Corporate
SOECB 0.4813 0.0404 0.1375 governance and bank performance: A joint analysis of the static, selection, and
(2.27) (2.20) (3.07) dynamic effects of domestic, foreign, and state ownership. Journal of Banking &
PCB 1.2715 0.0750 0.1202 Finance, 29, 21792221.
(1.35) (2.08) (1.94) Berger, A. N., Hasan, I., & Zhou, M. (2009). Bank ownership and efciency in China: What
HHI 0.3268 0.1462 0.2382 will happen to the world's largest nation? Journal of Banking & Finance, 33, 113130.
Berger, A. N., Kick, T., & Schaeck, K. (2014). Executive board composition and bank risk
(0.89) (2.22) (3.67)
taking. Journal of Corporate Finance (in press).
IRMC 0.7489 0.1014 0.0078
Berger, A. N., & Mester, L. J. (1997). Inside the black box: What explains differences in the
(1.39) (0.49) (0.35)
efciencies of nancial institutions? Journal of Banking & Finance, 21, 895947.
CRO 0.2986 0.0431 0.0536 Berglf, E., & Roland, G. (1998). Soft budget constraints and banking in transition. Journal
(1.80) (1.74) (1.99) of Comparative Economics, 26, 1840.
FEMALE 0.5614 0.2902 0.3015 Bhattacharyya, A., Lovell, C. A. K., & Sahay, P. (1997). The impact of liberalisation on the
(0.29) (4.73) (2.86) productive efciency of Indian commercial banks. European Journal of Operational
AR(1)/AR(2) 0.136/0.497 0.329/0.828 0.136/0.497 Research, 98(2), 332345.
Sargan/Hansen tests 0.673/0.969 0.982/0.948 0.908/0.548 Bonin, J. P., Hasan, I., & Wachtel, P. (2005). Bank performance, efciency and ownership in
transition countries. Journal of Banking & Finance, 29, 3153.
Notes: This table reports the results from two step system generalized method of Boyd, J. H., Gianni, D. N., & Jalal, A.M. (2006). Bank risk-taking and competition revisited: New
moments (GMM). All variables are dened in the Appendix. Numbers in parentheses theory and new evidence. IMF working papers 06/297. International Monetary Fund.
are z-statistics, Sargan and Hansen are the p value of the Sargan and Hansen test statistics Boyd, J. H., & Graham, S. L. (1986, Spring). Risk, regulation, and bank holding company
of over-identifying restrictions. AR(1)/AR(2) reports the p value of the rst- and second- expansion into nonbanking. Federal Reserve Bank of Minneapolis Quarterly Review,
order autocorrelation test statistics. 10(2), 217.
Indicates estimations that are signicant at 10% level. Brown, C., & Din, S. (2011). Too many to fail? Evidence of regulatory forbearance when
Indicates estimations that are signicant at 5% level. the banking sector is weak. Review of Financial Studies, 24, 13781405.
Indicates estimations that are signicant at 1% level. Cao, J., Lemmon, M., Tian, G., & Pan, X. (2011). Political promotion, CEO compensation,
and their effect on rm performance. Research Collection, Lee Kong Chian School of
Business (Open Access). Available at. http://works.bepress.com/xpan/1
Appendix A. Denition of variables and data source CBRC (2010). CBRC annual report. Beijing: CBRC.

Variables Denition
ROAE=TA
Z-score ROA ,where ROA is return on assets, E/TA is the ratio of equity to total assets and (ROA) is the standard deviation of return on assets.
Non-performing loan ratio (NPL ratio) Ratio of non-performing loans to total loans
Capital adequacy ratio (CAR) Risk-weighted capital adequacy ratio
Bank size (LnTA) Log of total assets of a bank
Cost/income (EFFICIENCY) Ratio of total costs to total income
 
Asset diversity (ADIV)  earning assets
1Net loansOther
Total earning assets 
Loans/deposits (TL/TD) Ratio of total loans to total funding
Listing status (LIST) A dummy variable equal to 1 if a bank has been listed at the end of the year and 0 otherwise
Economic growth (GDP %) Annual growth rate of GDP
Post-global nancial crisis (Post-GFS) A dummy variable equal to 1 for the period following the start of the global nancial crisis (20082011) and 0 otherwise
Government-controlled banks (GCBs) A dummy variable equal to 1 if the biggest shareholder is a state bureau or a state asset operating company and 0 otherwise.
(Omitted)
SOE-controlled banks (SOECBs) A dummy variable equal to 1 if the biggest shareholder is a state-owned enterprise (SOE) and 0 otherwise
Privately controlled banks (PCBs) A dummy variable equal to 1 if the biggest shareholder is a private rm or individual and 0 otherwise
Foreign strategic investment (FSI) A dummy variable equal to 1 if a bank has foreign strategic investment and 0 otherwise
Government-controlled ownership The percentage of shares owned by a state bureau or a state asset operating company if that owner is the largest shareholder
(GCO)
SOE-controlled ownership (SOECO) The percentage of shares owned by a SOE if that owner is the largest shareholder
Privately controlled ownership (PCO) The percentage of shares owned by a private company (or an individual) if that owner is the largest shareholder
Foreign strategic ownership (FSO) The percentage of shares owned by foreign strategic investors
Ownership Herndahl index (HHI) Herndahl index based on the ownership held by the ten largest shareholders of the bank.
Ownership concentration ratio (CR3) The percentage of shares owned by the top three shareholders
CRO on the executive team (CRO) A dummy variable equal to 1 if the Chief Risk Ofcer (CRO) is a member of the executive team and 0 otherwise.
Independent risk management A dummy variable equal to 1 if the chair of the committee is an independent director and 0 otherwise.
committee (IRMC)
% female directors (FEMALE) The proportion of female directors on the board
130 Y. Dong et al. / International Review of Financial Analysis 36 (2014) 120130

Chen, G., Firth, M., & Rui, O. (2006). Have China's enterprise reforms led to improved Iannotta, G., Nocera, G., & Sironi, A. (2007). Ownership structure, risk and performance in
efciency and protability? Emerging Markets Review, 7(1), 82109. the European banking industry. Journal of Banking & Finance, 31, 21272149.
Chen, G., Firth, M., & Xu, L. (2009). Does the type of ownership control matter? Evidence Iannotta, G., Nocera, G., & Sironi, A. (2013). The impact of government ownership on bank
from China's listed companies. Journal of Banking & Finance, 33(1), 171181. risk. Journal of Financial Intermediation, 22(2), 162176.
China Banking Regulatory Commission (CBRC) (2011). CBRC annual report. Beijing: CBRC. Isik, I., & Hassan, M. K. (2002). Cost and prot efciency of the Turkish banking industry:
Clarke, G. R. G., Cull, R., & Shirley, M. M. (2005). Bank privatization in developing countries: An empirical investigation. The Financial Review, 37(2), 257279.
A summary of lessons and ndings. Journal of Banking & Finance, 29, 19051930. Jia, C. (2009). The effect of ownership on the prudential behavior of banks The case of
Delios, A., Wu, Z. J., & Zhou, N. (2006). A new perspective on ownership identities in China. Journal of Banking & Finance, 33(1), 7787.
China's listed companies. Management and Organization Review, 2(3), 319343. John, K., Litov, L., & Yeung, B. (2008). Corporate governance and risk-taking. Journal of
Delis, M.D., & Kouretas, G. P. (2011). Interest rates and bank risk-taking. Journal of Banking Finance, 63, 16791728.
& Finance, 35(4), 840855. Kumbhakar, S., & Wang, D. (2007). Economic reforms, efciency and productivity in
Demirg-Kunt, A., & Detragiache, E. (2002). Does deposit insurance increase banking Chinese banking. Journal of Regulatory Economics, 32, 105129.
system stability? An empirical investigation. Journal of Monetary Economics, 49, La Porta, R., Lopez de Silanes, F., & Shleifer, A. (1999). Corporate ownership around the
13731406. world. Journal of Finance, 54, 471517.
Demirg-Kunt, A., Detragiache, E., & Tressel, T. (2006). Banking on the principles: Laeven, L., & Levine, R. (2007). Is there a diversication discount in nancial conglomer-
Compliance with Basel core principles and bank soundness. World Bank policy research ates? Journal of Financial Economics, 85(2), 331367.
working paper no. 3954. Laeven, L., & Levine, R. (2009). Bank governance, regulation, and risk taking. Journal of
Demirg-Kunt, A., & Huizinga, H. (2013). Are banks too big to fail or too big to save? Financial Economics, 93, 259275.
International evidence from equity prices and CDS spreads. Journal of Banking & Lin, X., & Zhang, Y. (2009). Bank ownership reform and bank performance in China.
Finance, 37(3), 875894. Journal of Banking & Finance, 33, 2029.
Demsetz, H., & Lehn, K. (1985). The structure of corporate ownership: Causes and conse- Loderer, C., & Martin, K. (1997). Executive stock ownership and performance: Tracking
quences. Journal of Political Economy, 93, 11551177. faint traces. Journal of Financial Economics, 45, 223255.
Demsetz, H., & Villalonga, B. (2001). Ownership structure and corporate performance. Martin, M. F. (2012). China's banking system: Issues for congress. Report for congress.
Journal of Corporate Finance, 7, 209233. Report no. R42380.
DeYoung, R., Peng, E., & Yan, M. (2013). Executive compensation and business policy Mehran, H., Morrison, A., & Shapiro, J. (2011). Corporate governance and banks: What have
choices at U.S. commercial banks. Journal of Financial and Quantitative Analysis, we learned from the nancial crisis? Federal Reserve Bank of New York. (Staff Report
48(01), 165196. No. 502).
Edwards, J., & Nibler, M. (2000). Corporate governance in Germany: The role of banks and Mester, L. J. (1997). Measuring efciency at U.S. banks: Accounting for heterogeneity is
ownership concentration. Economic Policy, 15(31), 237267. important. European Journal of Operational Research, 98, 230242.
Ellul, A., & Yerramilli, V. (2013). Stronger risk controls, lower risk: Evidence from U.S. Mongiardino, A., & Plath, C. (2010). Risk governance at large banks: Have any lessons
bank holding companies. Journal of Finance, 68(5), 17571803. been learned? Journal of Risk Management in Financial Institutions, 3, 116123.
Faccio, M., Masulis, R. M., & McConnell, J. J. (2006). Political connections and corporate Nichols, D., Wahlen, J., & Wieland, M. (2009). Publicly-traded versus privately-held:
bailouts. Journal of Finance, 61(6), 25972635. implications for conditional conservatism in bank accounting. Review of Accounting
Farrell, K. A., & Hersch, P. L. (2005). Additions to corporate boards: The effect of gender. Studies, 14, 88122.
Journal of Corporate Finance, 11, 85106. Okazaki, K. (2007). Banking system reform in China: The challenges of moving toward a
Fields, P. L., Fraser, D. R., & Subrahmanyam, A. (2012). Board quality and the cost of debt market-oriented economy. RAND Corporation.
capital: The case of bank loans. Journal of Banking & Finance, 36, 15361547. Omran, M. (2009). Post-privatization corporate governance and rm performance: The
Firth, M., Lin, C., Liu, P., & Wong, S. M. L. (2009). Inside the black box: Bank credit role of private ownership concentration, identity and board composition. Journal of
allocation in China's private sector. Journal of Banking & Finance, 33, 11441155. Comparative Economics, 37(4), 658673.
Forssbck, J. (2011). Owership structure, market discipline, and banks' risk-taking incen- Pathan, S. (2009). Strong boards, CEO power and bank risk-taking. Journal of Banking &
tives under deposit insurance. Journal of Banking & Finance, 35(10), 26662678. Finance, 33(7), 13401350.
Fries, S., & Taci, A. (2005). Cost efciency of banks in transition: Evidence from 289 banks Pathan, S., & Faff, R. (2013). Does board structure in banks really affect their performance?
in 15 post-communist countries. Journal of Banking & Finance, 29(1), 5581. Journal of Banking & Finance, 37(5), 15731589.
Fu, X., & Heffernan, S. (2007). Cost X-efciency in China's banking sector. China Economic Shehzad, C. T., de Haan, J., & Scholtens, B. (2010). The impact of bank ownership concen-
Review, 18(1), 3553. tration on impaired loans and capital adequacy. Journal of Banking & Finance, 34(2),
Fu, X., & Heffernan, S. (2010). Determinants of nancial performance in Chinese banking. 399408.
Applied Financial Economics, 20, 15851600. Sheshinski, E., & Lopez-Calva, L. F. (2003). Privatisation and its benets: Theory and
Fungov, Z., & Solanko, L. (2009). Risk-taking by Russian banks: Do location, ownership evidence. CESinfo Economic Studies, 49(3), 429459.
and size matter? In Current trends in the Russian nancial system. Vienna: SUERF. Shin, V., Zhang, Q., & Liu, M. (2007). Comparing the performance of Chinese banks: A
Garcia-Herrero, A., Gavila, S., & Santabarbara, D. (2009). What explains the low protability principal component approach. China Economic Review, 18(1), 1534.
of Chinese banks? Journal of Banking & Finance, 33, 20802092. Shirley, M., & Nellis, J. (1991). Public enterprise reform: The lessons of experience. World
Garcia-Marco, T., & Robles-Fernndez, M.D. (2008). Risk-taking behaviour and ownership Bank discussion papers, no. WDP 119.
in the banking industry: The Spanish evidence. Journal of Economics and Business, Shleifer, A. (1998). State versus private ownership. Journal of Economic Perspectives, 12(4),
60(4), 332354. 133150.
Gomes, A.R., & Novaes, W. (2005). Sharing of control as a corporate governance mechanism. Shleifer, A., & Vishny, R. W. (1986). Large shareholders and corporate control. Journal of
PIER working paper no. 01-029. Political Economy, 94, 461488.
Gorton, G., & Schmid, F. (1999). Corporate governance, ownership dispersion and ef- Shleifer, A., & Vishny, R. (1994). Politicians and rms. Quarterly Journal of Economics,
ciency: Empirical evidence from Austrian cooperative banking. Journal of Corporate 109(4), 9951025.
Finance, 5(2), 119140. Shleifer, A., & Vishny, R. W. (1997). A survey of corporate governance. Journal of Finance,
Greenwald, B., & Stiglitz, J. E. (1993, March). Monetary policy and the theory of the risk- 52, 737783.
averse bank. Proceedings. Federal Reserve Bank of San Francisco. Stiglitz, J. E. (1999). Wither reform? Ten years of transition. Keynote address, Annual Bank
Gul, F. A., Kim, J., & Qiu, A. (2010). Ownership concentration, foreign shareholding, audit Conference on Development Economics. The World Bank.
quality, and stock price synchronicity: Evidence from China. Journal of Financial Sullivan, R. J., & Spong (2007). Manager wealth concentration, ownership structure and
Economics, 95, 425442. risk in commercial banks. Journal of Financial Intermediation, 16(2), 229248.
Hasan, I., & Xie, R. (2012). A note on foreign bank entry and bank corporate governance in Wei, Z., Xie, F., & Zhang, S. (2005). Ownership structure and rm value in China's
China. BOFIT discussion papers (pp. 20122018). privatized rms: 19912001. Journal of Financial and Quantitative Analysis, 40(1),
Haw, I. M., Ho, S. M., Hu, B., & Wu, D. (2010). Concentrated control, institutions, and 87108.
banking sector: An international study. Journal of Banking & Finance, 34, 485497. Zhang, J., Jiang, C., Qu, B., & Wang, P. (2013). Market concentration, risk-taking, and bank
Hou, W., Lee, E., Stathopoulos, K., & Tong, Z. (2013). Executive compensation and the split performance: Evidence from emerging economies. International Review of Financial
share structure reform in China. European Journal of Finance (in press). Analysis, 30, 149157.
Hughes, J. P., & Mester, L. J. (2012). Efciency in banking: Theory, practice, and evidence. Zhang, J., Wang, P., & Qu, B. (2012). Bank risk taking, efciency, and law enforce-
In A. Berger, P. Molyneux, & J. Wilson (Eds.), The Oxford handbook of banking. Oxford: ment: Evidence from Chinese city commercial banks. China Economic Review,
Oxford University Press. 23, 284295.