Chinese commercial banks: Benefits from foreign strategic investors?

Pacific-Basin Finance Journal 40 (2016) 147–172

Contents lists available at ScienceDirect

Pacific-Basin Finance Journal journal homepage: www.elsevier.com/locate/pacfin

Chinese commercial banks: Benefits from foreign strategic investors? Maoyong Cheng a,⁎, Hongyan Geng b, Junrui Zhang b a b

School of Economics and Finance, Xi'an Jiaotong University, Xi'an, Shaanxi, China School of Management, Xi'an Jiaotong University, Xi'an, Shaanxi, China

a r t i c l e

i n f o

Article history: Received 16 February 2016 Received in revised form 30 September 2016 Accepted 26 October 2016 Available online 27 October 2016 JEL classification codes: G21 G32 G34 Keywords: Foreign strategic investors Bank risks State ownership China

a b s t r a c t Introducing foreign strategic investors (FSIs) has been vital to China's bank ownership reforms. Using relevant data between 1995 and 2014, we employ the propensity score matching and difference in differences approaches to investigate the effects of FSIs on bank risks, including insolvency risk, capital risk, liquidity risk, asset quality, and credit risk. We make several findings. First, FSIs may significantly reduce bank risks, with the exception of insolvency risk. Second, the effects of FSIs on capital risk, liquidity risk, asset quality, and credit risk are weaker in state-owned banks than in non-state-owned banks. In addition, FSIs-assigned directors and managers could further decrease bank capital risk, liquidity risk, and credit risk, and improve bank asset quality. And directors and managers have weaker effects on bank risks in stateowned banks. Finally, bank risks exhibit no significant changes after FSI exits, and the effects of FSIs on bank risks do not differ between banks without and with exited FSIs. This suggests that spillover effects work more than monitoring effects in the context of China's financial background. © 2016 Elsevier B.V. All rights reserved.

1. Introduction In this paper, we examine the effects of foreign strategic investors (FSIs) on bank risks, including insolvency risk, capital risk, liquidity risk, asset quality, and credit risk, in the Chinese banking sector for the period between 1995 and 2014. In contrast to the literature (Berger et al., 2009; Lin and Zhang, 2009; Xu, 2011; Jiang et al., 2013; Sun et al., 2013), we not only investigate the effects of FSIs on bank risks but also question whether FSI exits from China's banks affect their partners. Additionally, this paper explores whether the effects of FSIs on bank risks differ between state-owned banks and non-state-owned banks. In 1996, the Asian Development Bank (ADB) spent US$19 million to purchase 92.222 million shares, a 1.9% stake in China Everbright Bank, which is a national joint-stock bank. This is the first case for introducing FSIs of China's commercial banks. Since then, numerous Chinese banks have introduced FSIs. By the end of 2014, 43 of China's banks had introduced FSIs. However, contrary to the initial expectations of both FSIs and Chinese banks, some FSIs began to sell their shares of Chinese banks in 2007. This setback has also spread to three of the four largest state-owned banks and joint-stock banks. In fact, almost all of these

⁎ Corresponding author at: School of Economics and Finance, Xi'an Jiaotong University, Xi'an, Shaanxi, China. E-mail address: [email protected] (M. Cheng).

http://dx.doi.org/10.1016/j.pacfin.2016.10.011 0927-538X/© 2016 Elsevier B.V. All rights reserved.

148

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

foreign investments are profitable. Why do FSIs choose to exit Chinese banks? Policymakers and researchers have focused on the effects of FSIs in China with mixed results (Berger et al., 2009; Lin and Zhang, 2009; Xu, 2011; Jiang et al., 2013; Sun et al., 2013). Some aggressive commentators1 even argue that introducing foreign banks has been a total failure for Chinese banks. With this background, we attempt to answer four specific questions to provide information for policymakers regarding further reform of the Chinese banking system. How do FSIs affect bank risks? Do the effects of FSIs differ between state-owned and non-stateowned banks? How FSIs-assigned directors and managers influence bank risks? Do FSI exits impact the risks of Chinese banks? Policy considerations motivate our research. As emphasized by Bernanke (1983), Keeley (1990), Calomiris and Mason (1997, 2003a, 2003b), and Mohsni and Otchere (2014), the risk-taking behavior of banks affects financial and economic fragility. This relation is confirmed by the financial crisis of 2007. Under this background, the government proposes several privatization steps (including introducing FSIs) to shape bank risks in China. However, researchers have not reached a definite conclusion on how introducing FSIs affects their partners. FSI exits from China's banking sector provide further motivation. Why do FSIs choose to exit from Chinese commercial banks, especially state-owned banks, if all foreign investments are profitable? How do FSI exits affect their Chinese partners? The literature has also highlighted the value of examining the effects of FSIs on bank risks. Most authors do not control for selection bias or endogeneity in their studies, which undoubtedly leads to biased results. Therefore, it is urgent and interesting to examine the effects of FSIs on bank risks. Due to spillover and monitoring effects, FSIs are expected to reduce bank risks. According to the spillover effect view, FSIs provide not only external financing but also advanced management knowledge, which are presumably beneficial to investment recipients (Wei and Liu, 2006; Buckley et al., 2007; Spencer, 2008; Zhang and Li, 2010; Aggarwal et al., 2011; Zhu and Yang, 2016). Generally, FSIs are well-known, mature foreign financial institutions with financial management experience and technology. This advanced knowledge can be broadly spread to China's banks in the fields of corporate governance and risk management. The China Banking Regulatory Commission (CBRC) also recognizes that FSIs help Chinese banking institutions bolster their operational management, risk control, and corporate governance (CBRC 2010 Annual Report2). Aggarwal et al. (2011) suggest that foreign investors may export good corporate governance to host countries if they are from countries with strong shareholder protection measures. Hasan and Xie (2013) suggest that active involvement of FSIs in bank management has improved the corporate governance model of Chinese banks. Jia (2009) suggests that commercial banks with better corporate governance exhibit more prudent risk-taking behavior. Thus, we predict that FSIs may decrease bank risks. The monitoring effect view argues that FSIs can serve as outside monitors of target banks and improve banks' risk-taking behavior and risk management. For instance, monitors may prevent high-risk behavior and reduce moral hazards by preventing the most egregious forms of misbehavior (Tirole, 2001; Zhu and Yang, 2016). Additionally, the introduction of FSIs may attract attention from overseas regulatory institutions or media, which may also reduce banks' risk-taking behavior. Although FSIs have the ability to reduce bank risks according to the spillover and monitoring effects, why are FSIs willing to help China's banks to reduce bank risks? This is mainly due to the fact that the CBRC announced the rules for FSIs in five criteria3 at the end of 2005. These rules ensure that FSIs are long-term strategic investors (more than 3 years), not short-term financial investors, and transfer risk control experience and skills to their Chinese partners. Furthermore, FSIs hope that their investments would help them get long-term strategic returns, rather than shortterm financial returns. For instance, they hope to tap into an economy with a seemingly limitless growth potential and deliver greater access to the broader Chinese economy (Chang, 2013). Thus, FSIs would like to help China's partners, and access to Chinese economy. Therefore, based on spillover and monitoring effects, we predict that FSIs reduce bank risks. Using data from China's banks between 1995 and 2014, we combine the propensity score matching (PSM) and difference in differences (DID) approaches (we call this combination the “PSM-DID” approach) to investigate the effects of FSIs on bank risks and make several findings. First, introducing FSIs reduces bank risks, with the exception of insolvency risk. Second, the effects of FSIs are weaker in state-owned banks than in non-state-owned banks. Third, FSI-assigned directors and managers can further reduce bank capital risk, liquidity risk, and credit risk, and improve bank asset quality. And directors and managers have weaker effects in state-owned banks. Finally, our empirical results support the view that spillover effects work more than monitoring effects in the context of China's financial background. This article makes several contributions to the literature. First, we not only examine the effects of FSIs on bank risks but also question whether FSI exits from China's banks affect their partners. The existing studies mainly explored the effects of FSIs or foreign ownership on bank performance in China (Berger et al., 2005; Lensink et al., 2008; Berger et al., 2009; Lin and Zhang, 2009; Xu, 2011; Jiang et al., 2013; Sun et al., 2013). Some FSIs began to sell their stakes in Chinese banks in 2007. This setback has also spread to three of the big four state-owned and joint-stock banks. However, to the best of our knowledge, no study has explored the effects of FSI exits on their partners. Given this background, we question whether FSI exits affect Chinese banks. Second, due to the particularities of state ownership, we posit that state ownership may moderate the effects of FSIs. When we analyze the background of FSIs in Chinese banks, we notice a puzzling phenomenon. Some well-

1 For example, Zuo Dapei, a famous researcher in the Chinese Academy of Sciences even argues that foreign banks actually consider Chinese banks to be automated teller machines. 2 http://www.cbrc.gov.cn/EngdocView.do?docID=20110419222D1DDDE39BE80AFFEB3FF789309200. 3 The five criteria are as follows: an FSI should take stakes between 5% and 20%; the lock-up period for FSIs should be more than 3 years; FSIs are encouraged to assign directors to the board of directors of the banks in which they invest; FSIs should not invest in more than two banks with similar types; and FSIs should transfer their knowledge and network technology skills.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

149

known foreign investors began to sell their shares of Chinese state-owned banks but held their shares of joint-stock and city commercial banks. How can this phenomenon be explained? Some researchers argue that FSIs try to obtain large profits by selling their stakes in state-owned banks in China. In fact, FSIs obtain profits by selling all types of Chinese bank shares, regardless of whether they are state-owned or non-state-owned banks. As a result, we suggest that state-owned banks are more difficult for FSIs to change than non-state-owned banks, which make them unable to obtain strategic returns from state-owned banks and thus lead them to sell their stakes. Thus, we posit that state ownership may moderate the effects of FSIs. Third, this paper searches for the ways in which FSIs influence domestic banks and examines whether FSIs assigning directors and managers affect the risks of Chinese banks. This analysis provides empirical support for Chinese policymakers' welcoming FSIs to assign directors and managers to their partners. Finally, we use the PSM-DID approach to control for selection bias and endogeneity (Hillion and Vermaelen, 2004; Villalonga, 2004; Roberts and Whited, 2012), which allows us to obtain the true treatment effects of FSIs. The remainder of this paper is structured as follows. Section 2 presents the research background. Section 3 reviews some of the literature. Section 4 presents our sample, methods, and variables. Section 5 discusses the empirical results, Section 6 presents further analyses, and Section 7 concludes.

2. Foreign strategic investors in China's banking industry Berger et al. (2009) and Chen and Liao (2011) have provide good reviews of the Chinese banking industry and its reforms. Here, we review introducing FSIs and FSI exits from Chinese banks, separately.

2.1. Introducing foreign strategic investors to Chinese banks Ownership reforms in the Chinese banking industry were launched over a decade ago and included capital injections, restructuring ownerships, introducing FSIs, and undergoing initial public offerings. Introducing FSIs has played an important role in all stages of ownership reforms. Foreign banks are allowed to enter China to build four types of businesses: foreign bank branches, wholly owned foreign banks, joint ventures, and foreign strategic investments. Xu (2011) notes that foreign bank branches, wholly owned foreign banks, and joint ventures are defined as foreign banks operating in China and are subject to the particular rules and regulations governing foreign banks. Foreign strategic investments or foreign equity investments in Chinese banks are an important form of foreign entry. However, as foreign investors can only acquire minority ownership of domestic banks, these investment recipients remain Chinese banks and are subject to the rules governing Chinese banks. Before 2001, the regulations allowing FSIs to hold minority stakes in domestic banks were developing very slowly. During this time, the maximum total stakes that foreign investors were legally allowed to hold in a Chinese bank were 15% for a single investor and 20% for all foreign investors. At that time, only the ADB and the International Finance Corporation (IFC) were specifically approved by the People's Bank of China (PBOC) to invest in Chinese banks. In 1996, the ADB spent US$19 million to purchase 92.222 million shares, a 1.9% stake, of China Everbright Bank, a national joint-stock bank. This is the first case of FSI introduction into China's commercial banks. In 1999, the IFC invested US$22 million to purchase 100 million shares, a 5% stake, of the Bank of Shanghai, which is a city bank in which Shanghai's municipal government holds a 30% stake. That same year, the IFC purchased a 7% stake in China Everbright Bank. After China's entry into the WTO in 2001, especially after 2003, great progress was made in introducing FSIs as a result of the encouragement and promotion of regulators and the relaxation of regulatory policy. In 2001, the IFC purchased a 15% stake in the Bank of Nanjing, which is a city commercial bank. This was followed by the purchase of a 2% stake in the Bank of Shanghai by the IFC and the acquisition of an 8% stake in the Bank of Shanghai by the Hong Kong and Shanghai Banking Corporation Limited (HSBC). In 2003, supervision and management functions were devolved from the PBOC, and the CBRC was established. According to the CBRC's regulations, the maximum total stakes that foreign investors are legally allowed to hold in a Chinese bank are 20% for a single investor and 25% for all foreign investors. In 2005, the CBRC stated that foreign financial institutional investment stakes in state-owned Chinese banks cannot be less than 5% and that sales are forbidden for a period of 3 years to ensure long-term cooperation between FSIs and domestic banks. The CBRC also provides more detailed specifications for board member appointments and technology and network support, and for investing in no more than two Chinese banks with similar businesses. Examples of FSIs after 2003 include Citigroup's purchase of approximately 5% of the Shanghai Pudong Development Bank in January 2003 and a consortium (including Hang Seng Bank Ltd. and the IFC) purchase of a 24.98% stake in Industrial Bank in December 2003. During this stage, the most representative case is the investment made by Newbridge Capital (a U.S. investor group) in the Shenzhen Development Bank, which is considered to be a milestone for China's commercial banks and foreign financial institutions. In 2004, Newbridge Capital bought approximately 18% of the Shenzhen Development Bank. Due to the dispersed ownership of this Chinese bank, this was the first time that a foreign investor became the largest and controlling shareholder of a national domestic bank. Since China's accession to the WTO and ownership reforms, foreign acquisitions have prevailed in banking. Thus far, 43 of China's commercial banks had introduced FSIs by the end of 2014. Table A-1 lists the status of the Chinese commercial banks that have introduced FSIs. The big four state-owned banks, 11 joint-stock banks, and 28 city or rural commercial banks are on the list.

150

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

2.2. Foreign strategic investor exits from Chinese banks Contrary to the initial expectations of both FSIs and Chinese banks, some foreign investors began to sell their stakes in Chinese banks in 2007. This setback has also spread to three of the big four state-owned banks. In January 2005, the ADB obtained a 0.24% stake in the Bank of China (BOC). This was followed by the purchases of the Royal Bank of Scotland (RBS) and Fullerton Financial Holdings (FFH) in August 2005. The RBS acquired a 10% stake in the BOC for £1.7 billion and was the second-largest shareholder of the BOC from 2004 to 2009. FFH purchased a 5% stake in the BOC. In September 2005, when the United Bank of Switzerland (UBS) paid US$500 million for a 1.6% stake in the BOC, the Swiss bank hoped to cooperate with the Chinese bank in investment banking. The Bank of Tokyo Mitsubishi Ufj (BTMU) purchased a 0.19% stake in the BOC. However, in 2007, 2008, 2009, and 2012, respectively, FFH, the UBS, the RBS, and the ADB announced the sales of their entire stakes in the BOC. On December 31, 2008, the UBS announced the sale of its 3.4 billion H-shares of the BOC to institutional investors. By January 2009, the RBS had sold its entire stake in the BOC for US$2.34 billion. The pattern was the same for the China Construction Bank (CCB). The Bank of America (BOA) made its first investment in the CCB in 2005, paying US$3 billion for a 10% stake. The BOA poured another US$7 billion into the CCB in 2008, boosting its position to almost 20%. However, between January 2009 and November 2011, the BOA sold at least US$22.6 billion worth of shares, which had increased in value, leaving it with less than a 1% share. In April 2014, the BOA raised US$1.47 billion by selling its remaining stake in the CCB, ending an 8-year investment that generated paper profits more than five times the original cost. The Industrial and Commercial Bank of China (ICBC), China's biggest bank, tells the same story. In January 2006, the Goldman Sachs Group (GSG), Allianz AG, and American Express signed an agreement to buy a 10% stake in the ICBC for US$3.78 billion, and the GSG provided staff training, risk management assistance and internal controls, and corporate governance guidance. However, these foreign banks have sold all of their ICBC shares since 2009. The GSG bought a 7% stake in the ICBC for US$2.58 billion in April 2006 before the Chinese bank went public in Hong Kong. Through stock sales in 2011, 2012, and 2013, the GSG completed its exit from China's biggest lender. Of the big four state-owned banks, almost all foreign banks sold their stakes, with the exception that the BTMU maintains its 0.19% stake in the BOC, FFH holds its 7.15% stake in the CCB, and the Standard Charter Bank (SCB) holds a 0.37% stake in the Agricultural Bank of China (ABC). Such exits have also happened to joint-stock and city commercial banks. In 2007, the ADB and IFC unloaded the entirety of their stakes in China Everbright Bank. Newbridge Capital exited from the Shenzhen Development Bank. Citibank sold 506 million shares (2.71%) of the Shanghai Pudong Development Bank to institutional investors. The HSBC first bought a stake of 24 million shares in the Bank of Shanghai for US$63 million in 2001. In 2010, it paid another US$44 million for 24 million shares. However, in December 2013, the HSBC sold its entire stake in the Bank of Shanghai. The IFC sold a 20% stake in Changsha Commercial Bank, a 15% stake in Bank of Nanjing, and a 7% stake in the Bank of Shanghai. In fact, almost all foreign investments are profitable. For example, the BOA raised US$1.47 billion by selling its remaining stake in the CCB, generating paper profits more than five times the original cost. The GSG grossed US$10.1 billion from selling its ICBC stake after investing US$2.58 billion in 2006. According to Dealogic, U.S. banks acquired stakes of at least US$14.8 billion in Chinese banks from 2002 to 2010 and sold US$37.3 billion worth of those stakes from 2009 to 2013. The HSBC had acquired an 8% stake in the Bank of Shanghai in 2001 for US$63 million and sold all of its shares in 2013 for approximately US$645 million. Why did these foreign banks choose to exit Chinese banks? One reason for selling their stakes in Chinese banks is that international capital regulations, known as Basel III, which outlines the latest standards for bank capital, risk management, and liquidity, pressure lenders to pull out of minority investments. Due to Basel III, the next few years require banks to hold extra capital against minority stakes in other financial institutions, making such investments more expensive and less attractive for U.S. and European banks. The second reason may be that foreign investors do not gain the strategic returns they had anticipated in China. Chang (2013) argues that the BOA sold the bulk of its remaining CCB holdings in 2011 and the GSG unloaded another tranche of shares in the ICBC because they were frustrated that their large stakes were not helping them further their businesses in China. When the foreign banks invested in Chinese banks, they had hoped that the investments would help them tap into an economy with a seemingly limitless growth potential. However, some years later, they have found that they cannot gain a longterm foothold by making these investments in China. Holding stakes in banks has failed to deliver greater access to the broader Chinese economy. So, they decide to exit. We acknowledge that FSIs may have more incentives to gain short-term profits before exiting by encouraging high-risk behavior. However, FSIs do not have the decision-making power to conduct high-risk behavior as they are not ultimate controlling shareholders. This just responds to the reason that Tang Shuangning, the former vice-chairman of the CBRC, request state owners or domestic shareholders should keep ultimate controlling power after China's banks intruding FSIs on November 2, 2005.4 Third, there are growing concerns over the risk of bad loans. After 2008, the Chinese government loosened credit conditions, cut taxes, and embarked on a massive infrastructure spending program as part of a wide-ranging effort to offset adverse global economic conditions by boosting domestic demands. A stimulus package estimated at 4 trillion yuan (approximately US$570 billion) has been spent to finance programs in 10 major areas, such as low-income housing, rural infrastructure, water, electricity, transportation, and the environment, and commercial banks, especially state-owned banks, are required to serve these aims. Many loans have been granted to unviable projects and local governments, and the banks will eventually be forced to bear inevitable losses. We also find that ownership structure matters when foreign banks divest from their Chinese partners. The big four stateowned banks introduced nine foreign banks. However, almost all of these foreign banks have now sold their Chinese stakes.

4

http://www.gov.cn/ztzl/2005-11/03/content_90094.htm.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

151

The most interesting thing is that these foreign investors never sent directors to state-owned Chinese banks. The 12 joint-stock banks had introduced 22 foreign banks, 12 of which sold their shares of Chinese banks. However, of the 27 city commercial banks that have introduced 39 FSIs, only the IFC sold its shares of the Bank of Shanghai, the Bank of Nanjing, and the Bank of Changsha. From these facts, we suggest that when foreign banks invest in city commercial banks, they may obtain the strategic returns that they anticipate, as city commercial banks are less controlled by the central government. Political interventions make it difficult for foreign investors to change corporate governance in state-owned banks. 3. Literature review Banking has become increasingly globalized as a result of deregulation, advances in communications and technology, and more general economic integration. In particular, FSIs, or foreign owners, have increased sharply over the last few decades, a phenomenon that has led to many studies with mixed results. Here, we focus on the literature relevant to China. Other country studies are listed in Table 1. Foreign banks are allowed to enter China to build a commercial presence through four forms of entry: foreign bank branches, wholly owned foreign banks, joint ventures, and foreign strategic investments (Xu, 2011). Of the four types, wholly owned foreign banks and joint ventures are now quite rare. Participation as FSIs is the most important form of foreign ownership expansion or penetration in China. Similar studies to ours have primarily focused on the effects of FSIs on bank performance and risk in the research field of bank privatization, and have yielded mixed results. Lin and Zhang (2009) assess the effects of ownership reform on performance using a panel of Chinese banks from 1997 to 2004; their results suggest little performance change over both the short and the long term after a foreign acquisition. Shen et al. (2009) reveal that China's opening-up policy has clearly been effective in terms of increasing profits following the introduction of FSIs. However, for banks that have already introduced FSIs, releasing more shares to FSIs may not increase profits. García-Herrero et al. (2009) find that foreign equity did not affect profitability or pre-provision profit in Chinese banks from 1997 to 2004. Berger et al. (2009) find that minority foreign ownership is associated with significantly improved efficiency. Lu et al. (2010) investigate whether the mutual satisfaction of Chinese banks and FSIs in terms of their cooperation affects the performance of Chinese banks; their research represents a preliminary guide for Chinese policymakers. Lin (2011) studies the effects of foreign bank entry on domestic firms' access to bank credit using withincountry staggered geographic variations in foreign bank lending policies in China. The findings suggest that foreign bank lending helps alleviate the financial constraints of firms, especially of those that are less connected to the government (Lin, 2011). Xu (2011) examines the impact of foreign bank entry on China's banking performance using a spatially disaggregated measure of

Table 1 Effects of FSIs or foreign ownership investment in other countries. Author(s)

Country

Period

Empirical findings

Hasan and Marton (2003) Claessens and Laeven (2004) Martinez-Peria and Mody (2004) Choi and Hasan (2005) Fries and Taci (2005) Haber and Musacchio (2004) Bonin et al. (2005)

Hungary

1993–1998

50 countries

1994–2001

Foreign banks and banks with higher foreign bank ownership involvement were associated with lower inefficiency. The presence of more foreign banks can increase bank competition.

5 countries

1995–2001

Korea

1998–2002

15 Eastern European countries Mexico

1994–2001

6 countries

1994–2002

Berger et al. (2005)

Argentina

1993–1999

Detragiache and Gupta (2006) Micco et al. (2007)

Malaysia

1995–2001

179 countries

1995–2002

Lensink et al. (2008) 105 countries Gulamhussen and Portugal Guerreiro (2009)

1998–2003 1996–2004

Angkinand and Wihlborg (2010) Havrylchyk and Jurzyk (2011) Taboada (2011)

52 countries

1997–2003

11 Central and Eastern European countries 63 countries

1993–2005

Lee and Hsieh (2014)

27 Asian countries

1997–2004

1995, 2000, and 2005 1995–2009

The participation of foreign ownership can influence the spreads charged to borrowers and the process of financial intermediation. The foreign ownership level has a positive and statistically significant impact on stock market performance. Privatized banks with majority foreign ownership are most efficient; those with domestic ownership, least. Foreign ownership reduces credit, screen loans more intensively, and charges lower interest rate spreads. Foreign ownership improves efficiency. The importance of attracting a strategic foreign owner in the privatization process is confirmed. Foreign ownership is associated with statistically significantly lower profit efficiency than domestic ownership. Foreign non-regional banks had much less exposure to risky sectors (construction, real estate, and share purchases) than foreign regional banks and domestic banks. Foreign ownership located in developing countries improves performance. In industrial countries, no significant relationship between ownership and performance is observed. Foreign ownership negatively affects bank efficiency. Foreign equity reduces both total and operating costs, and foreign board membership reduces domestic banks' dependence on revenues from traditional areas of business and enhances the potential for generating revenues from non-traditional businesses. Foreign ownership is associated with greater risk taking. A positive impact of foreign bank ownership on the acquired banks' performance as well as on their market power. Increased foreign presence in the banking sector results in improvements in capital allocation in common law countries. The effect of increasing bank foreign ownership on stability and overall bank risk (Z-index) is significantly negative.

152

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

foreign bank presence and finds that foreign bank entry improves domestic banks' efficiency. Jiang et al. (2013) find that banks with minority foreign ownership are more profitable. Sun et al. (2013) investigate the effects of FSIs on bank efficiency in the context of regional economic development and suggest that strategic investors significantly increase efficiency in Chinese city commercial banks, although the effect of strategic investors on the efficiency of Chinese city commercial banks is negatively correlated with the level of regional economic development. Hasan and Xie (2013) focus on how FSIs affect Chinese bank performance, and find that active involvement of FSIs in bank management has changed the corporate governance model of Chinese banks from a control-based model to a market-oriented model and has promoted bank performance accordingly. A study by Wu et al. (2012) examines how FSIs affect earnings management and demonstrates that the banks with FSI experience improve financial report quality because they engage in less loss-avoidance earnings management. Wu et al. (2015) investigate whether FSIs influence the earnings smoothing (ES) of local Chinese banks. These two papers are similar, but Wu et al. (2012) focus on loss avoidance, whereas Wu et al. (2015) focus on ES and consider the influence of the number of FSIs.

4. Sample, methods, and variables 4.1. Sample and data We use income statement and balance sheet data for the period from 1995 to 2014 for China's commercial banks from the Fitch IBCA/Bureau van Dijk's Bank Scope database. The information for FSIs, state ownership, and listed status are obtained from the CBRC, the banks' public reports, and financial magazines using our own calculations. Starting with all 180 Chinese commercial banks, we apply the following sample selection process. First, we eliminate rural commercial banks, which differ considerably from state-owned, joint-stock, and city commercial banks in terms of governance mechanisms, business orientations, and other characteristics. Second, a sample bank should have at least six annual observations to ensure data stability. Thus, we do not include banks with less than six annual observations. Third, we exclude the observations that are missing requisite data. Fourth, we remove observations in the year that a bank introduced FSIs to help mitigate some of the short-term transitional costs of completing the governance change. Finally, to reduce the effects of outliers on our results, all continuous variables are winsorized at the 1% and 99% levels. Our basic sample is unbalanced and consists of 102 commercial banks and 1095 bank-year observations. Table 2 shows the distribution of observations.

4.2. Methodology This paper employs the PSM-DID approach to test the effects of FSIs on bank risks. First, we use PSM to match the control group (i.e., banks that have not introduced FSIs or “control banks”) with the treatment group (i.e., banks that have introduced FSIs or “treatment banks”). Then, we use DID models to explore the treatment effects of FSIs on bank risks.

Table 2 Distribution of observations. Year

FSI obs.

Non-FSI obs.

State obs.

Non-state obs.

Total obs.

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total

0 1 1 1 2 2 3 3 7 11 17 22 25 29 31 34 34 35 36 36 330

4 3 5 6 9 14 19 27 35 36 46 57 60 60 62 63 63 64 66 66 765

1 1 1 1 1 1 1 2 4 4 4 4 4 4 4 4 4 4 4 4 57

3 3 5 6 10 15 21 28 38 43 59 75 81 85 89 93 93 95 98 98 1038

4 4 6 7 11 16 22 30 42 47 63 79 85 89 93 97 97 99 102 102 1095

Note: FSI obs. contains the observations after a bank introduced FSIs. Non-FSI obs. includes observations before a bank introduced FSIs and all observations for banks that did not introduce FSIs during the sample period. State obs. includes observations from state-owned banks. Non-state obs. means observations from joint-stock and city commercial banks.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

153

4.2.1. Propensity score matching approach This paper uses the nearest neighbors matching method. Given a set of bank characteristics, i.e., the accounting ratios, the conditional probability of receiving treatment is estimated using a logistic function and a sample that contains FSI and Non-FSI banks.5 The FSI banks are ranked according to the estimated conditional probability, which is referred to as the “propensity score”. Each FSI bank is then matched to the single Non-FSI bank with the closest propensity score. The role of the score is to reduce the dimensionality of the matching problem. Matching on the propensity scores allows one to maximize the comparability of the treatment and control groups. The PSM method is used to generate the matching sample. This is conducted as follows. Let 1. t be the fiscal year before the banks introduced FSIs, which is defined as year-1. Given that banks first introduced FSIs in 1996, 1999, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2012, 2013 in our sample, t should equal 1995, 1998, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2011, and 2012, respectively; 2. i be an FSI bank or Non-FSI bank with i = 1, …, Nt, where Nt is the cross-sectional sample size in year t; Nt = FSIt + Non-FSIt, where FSIt and Non-FSIt are the number of FSI and Non-FSI banks in year t, respectively. Summing all of the years, the total number of FSI and Non-FSI banks equals FSI and Non-FSI, respectively; 3. Xits be the characteristic s observed for bank i in year t; with s = 1, …, 5. The characteristics are (1) assets (Sizeit), (2) return on assets (ROAit), (3) capital adequacy ratio (CARit), (4) non-performing loan ratio (NPLit), and (5) cost to income ratio (CTIit); 4. FSITREAT be a dummy variable equal to 1 for FSI banks and 0 for Non-FSI banks. The following steps generate the matching sample in fiscal year t starting with t = 19986: 5. Estimate the propensity to introduce, Pit, using the logit function: Pit = propensity score (FSITREATit = 1 / Xits), for i = 1, …, Nt; 6. Rank the estimated propensity scores for the FSI banks in ascending order; 7. Match each FSI bank to the single Non-FSI bank with the closest propensity score to form a sample of Non-FSIt nearest neighbors matching control banks; 8. Repeat steps (5) to (7) for each fiscal year t. We pool the estimated propensity scores across the fiscal years, t, and obtain a total sample of FSI banks and Non-FSI banks, with FSI equal to Non-FSI by construction. 4.2.2. Difference in differences approach To analyze the effects of FSIs on bank risks, we estimate the basic DID model below. The coefficient of Time_FSITREAT reflects the treatment effects of FSIs on bank risks in Model (1). Due to spillover and monitoring effects, we expect bank risks to decrease after introducing FSIs. Risks ¼ Constant þ a1 Time þ a2 FSITREAT þ a3  Time FSITREAT þ b  Control þ r  Year þ ε

ð1Þ

The second purpose of our study is to determine whether the effects of FSIs on bank risks differ between state-owned and non-state-owned banks. To examine this difference, we add the interaction variable Time_FSITREAT_State to Model (1) to form Model (2). Based on the ownership concentration, social, and corporate governance views of state-owned banks, we argue that the effects of FSIs on bank risks are weaker for state-owned banks than for non-state-owned banks. Risks ¼ Constant þ a1 Time þ a2 FSITREAT þ a3 Time FSITREAT þ a4  Time FSITREAT State þ b  Control þ r  Year þ ε ð2Þ In the above models, Risks are measured by Z-Score, CAR, Liquidity, LLP, and NPL; Control is a matrix of additional bank controls, containing State, List, Size, Loan growth, NIM, NII, and CTI; Time_FSITREAT is the interaction variable between Time and FSITREAT; Time_FSITREAT_State is the interaction variable between Time_FSITREAT and State; and ε is the disturbance term. All of the variables are defined or measured in Table 3. 4.3. Variables 4.3.1. Bank risks We consider various measures of bank risks that are commonly used in the literature. First, we use the Z-Score to measure insolvency risk. It reflects the average value of Z-Scoreit for the 4 years7 before or after a bank (or its matched bank) introducing FSIs. Following Ariss (2010) and Demirguc-Kunt and Huizinga (2010), Z-Scoreit = (ROAit + CARit) / SDROA, where ROAit is the rate of return on assets for bank i in year t, CARit is the capital-asset ratio for bank i in year 5

FSI banks are banks that have introduced FSIs; Non-FSI banks are banks that have not introduced FSIs. Due to missing data for 1995, we start with t = 1998. 7 Based on our data set, we use 4-year period to compute the average values. Some banks have fewer than 4 annual observations before and after foreign investment. We use as many year observations as possible to compute the average values for every variable. Additionally, we use 2- and 3-year period as robustness tests. The results are similar to our main findings. 6

154

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 3 Variable definitions. Symbol

Definition

Z-Score

Average value of Z-Scoreit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where Z-Scoreit is the return on assets plus the capital-asset ratio divided by the standard deviation of asset returns for bank i in year t Average value of CARit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where CARit is the ratio of book value of equity to total assets for bank i in year t Average value of Liquidityit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where Liquidityit is the ratio of liquid assets to the sum of deposits and short-term funding for bank i in year t Average value of LLPit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where LLPit is the ratio of loan loss provision to total loans for bank i in year t Average value of NPLit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where NPLit is the ratio of impaired (non-performing) loans to net loans for bank i in year t Dummy variable, equals 1 after banks (or their matched banks) introducing FSIs, and 0 otherwise Dummy variable, equals 1 if a bank has introduced FSIs and 0 otherwise Average value of FSIDit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where FSIDit equals 1 if FSIs assigning directors and managers to Chinese bank i in year t, and equal to 0 otherwise Dummy variable, equals 1 if a bank is a state-owned commercial bank and 0 otherwise Dummy variable, equals 1 if a banks went public and 0 otherwise Average value of Sizeit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where Sizeit is the natural logarithm of total assets for bank i in year t Average value of Loan growthit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where Loan growthit is the growth of total loans for bank i in year t Average value of NIMit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where NIMit is the ratio of net interest income to total earning assets for bank i in year t Average value of NIIit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where NIIit is the ratio of non-interest income to gross revenues for bank i in year t Average value of CTIit for the 4 years before or after a bank (or its matched bank) introducing FSIs, where CTIit is the cost to income ratio for bank i in year t Dummy variables, equals 1 if the banks refer to the corresponding year when they (or their matched bank) introduced FSIs and 0 otherwise

CAR Liquidity LLP NPL Time FSITREAT FSID State List Size Loan growth NIM NII CTI Year

t, and SDROA is an estimate of the standard deviation of the rate of ROA over the sample period. A higher Z-Score indicates that a bank is more stable, as it is inversely related to the bank's insolvency probability. Second, we adopt CAR as a proxy for capital risk (Shehzad et al., 2010). CAR reflects the average value of CARit for the 4 years before or after a bank (or its matched bank) introducing FSIs. CARit is the ratio of the book value of equity to total assets for bank i in year t. A higher CAR indicates less risk. Third, this article uses Liquidity as a proxy for liquidity risk (Imbierowicz and Rauch, 2014). Liquidity reflects the average value for Liquidityit for the 4 years before or after a bank (or its matched bank) introducing FSIs. Liquidityit is calculated as the ratio of liquid assets to the sum of deposits and short-term funding for bank i in year t. A higher liquidity value indicates less risk. Fourth, we employ LLP as a proxy for bank asset quality (Iannotta et al., 2007). LLP reflects the average value of LLPit for the 4 years before or after a bank (or its matched bank) introducing FSIs. LLPit is measured by the ratio of loan loss provision to total loans for bank i in year t. A higher LLP indicates higher risk. Finally, we use NPL to reflect bank credit risk (Shehzad et al., 2010; Imbierowicz and Rauch, 2014). NPL is the average value of NPLit for the 4 years before or after a bank (or its matched bank) introducing FSIs. NPLit is measured by the ratio of impaired (non-performing) loans to net loans for bank i in year t. A higher NPL indicates higher risk. 4.3.2. Introducing foreign strategic investors According to the DID approach, we use an interaction variable, Time_FSITREAT, to capture FSIs' treatment effects. Time is a dummy variable that equals 1 after a bank (or its matched bank) introduced FSIs and 0 before it (or its matched bank) introduced FSIs. FSITREAT is a dummy variable that equals 1 if banks have introduced FSIs and 0 otherwise. 4.3.3. Control variables First, we include the following variables to control for the effects of individual bank characteristics. Based on research by Iannotta et al. (2007), Lin and Zhang (2009), Nichols et al. (2009), Shehzad et al. (2010), and Barry et al. (2011), we use State, List, Size, Loan growth, NIM, NII, and CTI to control for individual bank characteristics. State, a dummy variable that equals 1 if a bank is state-owned and 0 otherwise, captures the effect of bank ownership. List is a dummy variable that equals 1 if a bank went public and 0 otherwise. Size, Loan growth, NIM, NII, and CTI are the average values of Sizeit, Loan growthit, NIMit, NIIit, and CTIit, respectively, for the 4 years before or after a bank (or its matched bank) introducing FSIs. Sizeit, the natural logarithm of the total assets for bank i in year t, controls for the effect of bank size. Loan growthit, the growth of the total loans for bank i in year t, is adopted as a proxy for bank growth opportunities. NIMit, the ratio of net interest income to total earning assets for bank i in year t, captures the profitability of interest business. NIIit, the ratio of non-interest income to gross revenues for bank i in year t, captures the effect of a bank's business orientation. CTIit, the cost to income ratio for bank i in year t, is used as a proxy for bank efficiency.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

155

Then, we employ the Year dummies to control for annual time differences, as China's banks introduced FSIs in different years. Each dummy variable equals 1 if the banks refer to the corresponding year when they (or their matched bank) introduced FSIs and 0 otherwise. 5. Empirical results 5.1. Descriptive statistics Using PSM8 we obtain a sample with 35 FSI banks and 35 Non-FSI banks. We then compute the average values of every variable used in the above models for the 4 years before or after a bank (or its matched bank) introducing FSIs. For every bank, we obtain two observations, the observation before treatment and the observation after treatment. Finally, we obtain a sample in which treatment group includes 35 banks and 70 observations, and the control group contains 35 banks and 70 observations. Table 4 reports the descriptive statistics for all of the variables. Table 5 presents the simple comparison between treatment banks and control banks. The findings suggest that banks that have introduced FSIs are associated with reductions in capital risk, liquidity risk, and credit risk, and improvements in asset quality above and beyond selection effects. 5.2. Effects of foreign strategic investors on bank risks. Table 6 reports the ordinary least squares (OLS) regression estimates for Model (1) with Z-Score, CAR, Liquidity, LLP, and NPL as the dependent variables. In the Z-Score model, Time_FSITREAT is positive but not significant, indicating that banks' insolvency risks do not decrease after introducing FSIs. This result occurs because commercial banks can easily obtain stable profits from traditional activities in the context of controlled interest rates, and their insolvency risk is very low and stable, which is difficult to change. In the CAR model, Time_FSITREAT is positive and significant at the 1% level, revealing that FSIs lower bank capital risk. In the Liquidity model, Time_FSITREAT is positive and significant at the 1% level, indicating that bank liquidity risk is reduced after introducing FSIs. In the LLP model, Time_FSITREAT is negative and significant at the 1% level, suggesting that FSIs improve bank asset quality. In the NPL model, Time_FSITREAT is negative and significant at the 1% level, revealing that bank credit risk is also decreased after introducing FSIs. In short, FSIs reduce bank capital risk, liquidity risk, and credit risk, and improve asset quality. These results are consistent with our expectations that the introduction of FSIs reduces the bank risks of their Chinese partners due to spillover and monitoring effects. That is, FSIs may produce substantial improvements in the corporate governance, technological advancement, and risk management of Chinese banks, thus reducing bank risk. Furthermore, FSIs serve as outside monitors supervising their partners' risk-taking behavior. Additionally, FSIs may reduce bank risk by mandating or encouraging banks to go public and list their shares on major stock exchanges, which may require domestic banks to improve their performance and reduce risk (Berger et al., 2009). 5.3. Interaction effects of foreign strategic investors and state ownership on bank risks Table 7 shows the OLS regression estimates for Model (2) with Z-Score, CAR, Liquidity, LLP, and NPL as the dependent variables. In this section, we focus on the Time_FSITREAT_State variable. Time_FSITREAT_State is negative but not significant in the Z-Score model, indicating that the effects of FSIs on insolvency risk do not differ between state-owned and non-state-owned banks. Time_FSITREAT_State is significantly negative in the CAR and Liquidity models and positive in the LLP and NPL models. That is, the effects of FSIs on capital risk, liquidity risk, asset quality, and credit risk are weaker in state-owned banks than in non-state-owned banks. These results explain the puzzling phenomenon that it is more difficult for FSIs to affect Chinese partners and get the strategic returns in state-owned banks, and respond to the research by Chang (2013). This pattern can be explained by the ownership concentration, social, and corporate governance views of state-owned banks. According to the ownership concentration view, state-owned banks have high ownership concentrations, which may hinder FSIs' ability to change existing risk behavior. The social view argues that one of the goals of state-owned banks is to serve social welfare maximization by correcting market failure and meeting the macroeconomics regulation (Sapienza, 2004; Clarke et al., 2005). That is, state-owned banks have to consider more social factors rather than economic ones (Brandt and Li, 2003; Cull and Xu, 2003; Firth et al., 2009), and have less motivation and enthusiasm to change their existing behavior for the economic purposes pursued by FSIs. According to the corporate governance view, state-owned banks generally have worse governance in China than non-state-owned banks (Berger et al., 2009; Jia, 2009; Lin and Zhang, 2009). Furthermore, Lensink et al. (2008) argue that good governance may weaken any detrimental effects of foreign ownership or strengthen any beneficial effects. Therefore, the beneficial effects of FSIs are stronger in banks with better governance (i.e., non-state-owned banks). 5.4. Robustness tests. First, Wu et al. (2015) note that the CBRC proposed the FSI concept to local banks at the end of 2005, and most local banks did not disclose their financial information before 2006. Thus, they use 2006 as their starting year (Wu et al., 2015). To ensure the 8

The logistic regression results for every year are not shown for reasons of brevity. They are available upon request from the authors.

156

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 4 Summary of descriptive statistics. Variables

Obs.

Mean

Std. dev.

Min

Max

Z-Score CAR Liquidity LLP NPL Time FSITREAT FSID State List Size Loan growth NIM NII CTI

140 140 140 140 140 140 140 140 140 140 140 140 140 140 140

21.542 0.059 0.299 0.009 0.038 0.500 0.500 0.414 0.057 0.229 10.776 0.282 0.032 0.187 0.422

12.326 0.033 0.175 0.007 0.049 0.504 0.503 0.483 0.228 0.348 2.417 0.216 0.013 0.118 0.153

3.536 0.014 0.077 0.000 0.001 0.000 0.000 0.000 0.000 0.000 2.227 -0.366 0.008 0.001 0.151

33.437 0.282 1.302 0.035 0.316 1.000 1.000 1.000 1.000 1.000 16.285 2.018 0.066 0.622 0.943

Note: See Table 3 for variable definitions and measurements.

robustness of our results, we also consider the banks that introduced FSIs during the period from 2006 to 2014 as treatment banks to test the effects of FSIs on bank risks. The results in Tables 8 and 9 confirm our main findings as discussed above. Second, we match each FSI bank to the single Non-FSI bank with the closest propensity score in the main analyses. To ensure the robustness of the results, we expand our sample, matching each FSI bank to two Non-FSI banks with the closest propensity scores (if one Non-FSI bank is matched to different FSI banks in the same year, this bank appears only once due to collinearity considerations). The other PSM steps are the same as those in Subsection 4.2.1. Finally, we obtain a sample in which the treatment group includes 35 banks and 70 observations and the control group includes 64 banks and 128 observations. Then, we use this new sample to re-estimate the empirical Models (1) and (2). The results in Tables 10 and 11 confirm our main findings as discussed above. Third, in the literature on bank privatization (Berger et al., 2005; Berger et al., 2009; Lin and Zhang, 2009; Zhu and Yang, 2016), authors often use selection variables to eliminate the selection bias in their models. To ensure the robustness of our results, this paper also employs similar methods to establish alternative Models (3) and (4), and to explore the effects of FSIs on bank risks using all 1095 observations. The results in Tables 12 and 13 confirm our main findings as discussed above. Risksit ¼ Constant þ a  Selectionit þb  FSIit þc  Controlit þ r  Year þ ε

ð3Þ

Risksit ¼ Constant þ a  Selectionit þb1 FSIit þb2 FSIit Stateit þc  Controlit þ r  Year þ ε

ð4Þ

In Models (3) and (4), Risksit are measured by Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit; Selectionit is a dummy variable that equals 1 if a bank introduced FSIs over the sample period and 0 otherwise; FSIit is a dummy variable that equals 1 beginning in the year following a bank's introduction of FSIs and equals 0 for the years before a bank introduced FSIs and for the banks that did

Table 5 The comparison results.

Treatment banks (1) Before introducing FSIs (2) After Introducing FSIs Difference A (2) − (1) t-Statistic Control banks (1) Before introducing FSIs (2) After introducing FSIs Difference B (2) – (1) t-Statistic Difference A – difference B t-Statistic

Obs.

Z-Score

CAR

Liquidity

LLP

NPL

35 35

21.758 23.216 1.458 1.236

0.054 0.075 0.021⁎⁎⁎ 4.325

0.291 0.334 0.043⁎⁎⁎ 5.386

0.010 0.007 −0.003⁎⁎⁎ −5.745

0.042 0.026 −0.016⁎⁎⁎ −6.395

35 35

20.016 21.178 1.162 0.926

0.051 0.058 0.007⁎ 1.712

0.277 0.294 0.017⁎⁎ 2.143

0.010 0.009 −0.001⁎⁎ -2.423

0.045 0.039 −0.006⁎⁎ -2.143

0.296 0.437

0.014⁎⁎⁎

0.026⁎⁎⁎ 3.783

–0.002⁎⁎⁎ –3.893

–0.010⁎⁎⁎ –4.273

Note: See Table 3 for the variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

3.175

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

157

Table 6 Effects of FSIs on bank risks. Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

22.636⁎⁎⁎

1.436⁎⁎⁎

0.768⁎⁎⁎

2.546⁎⁎⁎

(10.343) 0.216 (1.282) 0.408⁎⁎⁎

(11.657) 0.074 (1.212) 0.418⁎⁎⁎

(6.657) −0.405 (−0.384) −0.363⁎⁎⁎

0.876⁎⁎⁎ (10.454) −0.110 (−0.173) −0.379⁎⁎⁎

(3.360) 0.243⁎⁎⁎

(2.977) 0.224⁎⁎⁎

(−2.680) −0.169⁎⁎⁎

(−2.679) −0.181⁎⁎⁎

(2.677) −0.251 (−0.327) 0.044⁎⁎ (2.436) −0.022⁎⁎⁎

(2.684) 0.125 (1.337) 0.030⁎⁎ (2.433) −0.132⁎⁎

CTI

(−5.435) −0.062⁎⁎

(−4.537) 0.427 (0.928) 3.493 (0.436) −0.102 (−0.456) −0.653⁎⁎⁎

(−2.485) −1.254 (−0.467) −5.463 (−0.235) −0.128 (−1.543) −0.103⁎⁎

(−2.941) 0.004⁎⁎⁎ (2.846) −0.026⁎⁎ (−2.523) 0.007 (0.572) 0.102⁎ (1.874) −0.329⁎⁎⁎ (−9.434) 0.242 (0.657) 0.032⁎⁎⁎

(−2.827) 0.227⁎⁎ (2.316) −0.033⁎⁎ (−2.346) 0.017⁎⁎⁎

NII

(10.445) 0.157 (0.992) 0.425 (1.198) 0.215 (1.074) 0.003 (1.253) 0.021 (1.356) 0.006 (0.563) 0.052 (0.849) 0.305⁎⁎ (8.483) −0.214⁎⁎⁎

Year Adj.-R2 F N

(−2.324) Yes 0.535 27.324 140

(−5.655) Yes 0.348 33.205 140

(−2.354) Yes 0.518 23.421 140

(2.868) Yes 0.548 41.493 140

(3.675) Yes 0.342 35.396 140

Time FSITREAT Time_FSITREAT State List Size Loan growth NIM

(4.336) 0.403⁎⁎⁎ (2.796) −3.430⁎⁎⁎ (−5.546) 0.425 (0.235) 0.733⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at 1% level. ⁎⁎ Significance at 5% level. ⁎ Significance at the 10% level.

not introduce FSIs; Controlit is a matrix of additional bank control variables, such as Stateit, Listit, Sizeit, Loan growthit, NIMit, NIIit, and CTIit; and ε is the disturbance term. In Model (3), the coefficient of FSIit measures the effects of FSIs on bank risks. FSIit_Stateit in Model (4) reflects the different effects of FSIs on bank risks between state-owned banks and non-state-owned banks. All variables are measured in Table A-2. Finally, we check whether the effects of FSIs on bank risks differ before and after 2007. Since 2007, some FSIs have started selling their shares of Chinese banks. This may be due to some important shocks on FSIs or Chinese banks that occurred around that time, such as the global implementation of Basel III and the split-share reform in China. Therefore, there are some reasons to suspect that the effects of FSIs on bank risks differ before and after 2007. To assess this, we split our sample into banks introducing FSIs before 2007 and banks introducing FSIs in and after 2007, and perform separate regression estimations for each group. Table 14 reports the results. Panel A shows the results using banks that introduced FSIs before 2007 as the research sample. Panel B reports the results using banks that introduced FSIs in and after 2007 as the research sample. The results in the two panels are similar and consistent with our main findings. The Chow tests (Chow, 1960) also show that the effects of FSIs on bank risks are not different in every model in these two panels, suggesting that the effects of FSIs on bank risks do not differ before and after 2007. That is, our results do not seem to be driven by the shocks that occurred around 2007. Additionally, this paper also adds the interaction variable Time_FSITREAT_Shock to Model (1) and form Model (5) to verify this view. Table 15 shows the results. Time_FSITREAT_Shock is not significant in every model, which supports the view that there are no significant differences in the effects of FSIs on bank risks before and after 2007. Risks ¼ Constant þ a1 Time þ a2 FSITREAT þ a3 Time FSITREAT þ a4  Time FSITREAT Shock þ b  Control þ r  Year þ ε ð5Þ

In Model (5), Risks are measured by Z-Score, CAR, Liquidity, LLP, and NPL; Control is a matrix of additional bank controls, containing State, List, Size, Loan growth, NIM, NII, and CTI; Time_FSITREAT is the interaction variable between Time and FSITREAT; Shock is a dummy variable that equals 1 if a Chinese bank introduced FSIs before 2007 and 0 otherwise; Time_FSITREAT is the interaction variable between Time and FSITREAT; Time_FSITREAT_Shock is the interaction variable between Time_FSITREAT and Shock; Year is a matrix of year dummy variables; and ε is the disturbance term. All of the variables are defined or measured in Table 3.

158

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 7 Interaction effects between FSIs and state ownership on bank risks. Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

20.017⁎⁎⁎

1.018⁎⁎⁎

0.456⁎⁎⁎

2.278⁎⁎⁎

(10.245) 0.134 (0.879) 0.421 (1.143) 0.224 (1.215) −0.084 (−1.146) 0.002 (0.479) 0.014 (0.742) 0.001 (1.058) 0.036 (1.003) 0.221⁎⁎⁎

(4.326) 0.204 (1.127) 0.396⁎⁎⁎

(11.153) 0.071 (1.204) 0.414⁎⁎⁎

(12.451) −0.387 (−0.365) −0.354⁎⁎⁎

0.742⁎⁎⁎ (9.415) −0.104 (−0.167) −0.372⁎⁎⁎

(3.346) 0.285⁎⁎⁎ (2.785) −0.123⁎⁎ (−2.436) −0.103 (−0.446) 0.029⁎⁎⁎

(2.896) 0.264⁎⁎⁎ (2.856) −0.112⁎⁎⁎ (−2.647) 0.056 (1.251) 0.025⁎⁎

(−2.658) −0.218⁎⁎⁎ (−3.287) 0.094⁎⁎⁎ (2.683) 0.003⁎⁎ (2.075) −0.015⁎⁎⁎

(−2.646) −0.235⁎⁎⁎ (−3.218) 0.087⁎⁎⁎ (2.837) 0.256⁎⁎ (2.245) −0.016⁎⁎⁎

(2.946) −0.001⁎⁎ (−2.124) 0.365 (1.090) 3.236 (0.433) −0.028 (−0.563) −0.021⁎⁎⁎ (−2.597) Yes 0.352 34.386 140

(2.245) −0.112⁎⁎ (−2.491) −1.472 (−1.021) −5.189 (−0.536) −0.063 (−1.164) −0.054⁎⁎ (−2.345) Yes 0.525 24.837 140

(−2.893) 0.001 (0.661) 0.074⁎ (1.879) −0.457⁎⁎

(−2.886) 0.010⁎⁎⁎ (4.459) 0.164⁎⁎ (2.142) −3.638⁎⁎⁎

(−2.227) 0.024 (0.649) 0.024⁎⁎ (2.438) Yes 0.554 43.029 140

(−5.426) 0.334 (0.422) 0.546⁎⁎⁎ (2.674) Yes 0.348 37.143 140

Time FSITREAT Time_FSITREAT Time_FSITREAT_State State List Size Loan growth NIM NII CTI Year Adj.-R2 F N

(8.327) −0.218⁎⁎⁎ (−4.548) −0.017⁎⁎⁎ (−3.535) Yes 0.541 28.866 140

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

6. Further analyses 6.1. Effects of directors and managers assigned by foreign strategic investors on bank risks When we analyze the background of FSIs in Chinese banks, an interesting phenomenon emerges. The CBRC welcomes FSIs to appoint directors and managers who can share their managerial experience and knowledge to their counterparts in local banks (Wu et al., 2015). However, not all FSIs have followed this suggestion. For example, the BOA appointed one director to the CCB. The SCB held shares in the ABC but did not assign a director. This raises an interesting question: can FSI-assigned directors and managers (FSID) reduce bank risks? To explore this issue, we employ FSI banks as the research sample and establish Model (6). We also form Model (7) to examine whether the effects of FSID on bank risks differ between state-owned and non-state-owned banks. Risksit ¼ Constant þ a  FSIDit þb  Controlit þ r  Year þ ε

ð6Þ

Risksit ¼ Constant þ a1 FSIDit þa2 FSIDit Stateit þb  Controlit þ r  Year þ ε

ð7Þ

In Models (6) and (7), Risksit are measured by Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit; FSIDit is a dummy variable that equals 1 if FSIs assign directors and managers to Chinese bank i in year t and 0 otherwise; FSIDit_Stateit is the interaction variable between FSIDit and Stateit; Controlit is a matrix of additional bank control variables, containing Stateit, Listit, Sizeit, Loan growthit, NIMit, NIIit, and CTIit; and ε is the disturbance term. All variables are measured in Table A-2. Table 16 presents the OLS regression estimates for Model (6) with Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit as the dependent variables. In the Z-Scoreit model, FSIDit is positive but not significant. This indicates that FSIs assigning directors and managers do not affect insolvency risk. FSIDit is significantly positive in the CARit and Liquidityit models and negative in the LLPit and NPLit models, indicating that FSIs assigning directors and managers may reduce capital risk, liquidity risk, and credit risk, and improve asset quality. This occurs because FSIs can provide domestic banks with more direct and convenient risk management assistance and monitor the risk-taking behavior of local banks if they assign directors and managers. This is also why China's policymakers welcome FSIs to assign directors and managers for their Chinese partners. Furthermore, Table 17 shows that FSIDit_Stateit is

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

159

Table 8 Effects of FSIs on bank risks (2006 to 2014). Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

18.589⁎⁎⁎

2.685⁎⁎⁎

1.318⁎⁎⁎

2.632⁎⁎⁎

(11.425) 0.235 (1.436) 0.439⁎⁎⁎

(14.259) 0.094 (1.432) 0.435⁎⁎⁎

(12.258) −0.432 (−0.675) −0.395⁎⁎⁎

0.741⁎⁎⁎ (13.254) −0.143 (−0.376) −0.405⁎⁎⁎

(3.443) 0.261⁎⁎⁎

(3.245) 0.237⁎⁎⁎

(−2.756) −0.180⁎⁎⁎

(−2.743) −0.197⁎⁎⁎

(2.868) −0.056 (−0.895) 0.019⁎⁎ (2.093) −0.006⁎

(2.755) −0.009 (−0.638) 0.015⁎⁎⁎ (3.189) −0.005⁎⁎

CTI

(−9.854) −0.612⁎⁎⁎

(−1.873) 0.115 (0.327) 5.019 (1.004) −0.006 (−0.086) −0.534⁎⁎⁎

(−2.252) 0.226 (0.167) −2.145 (−0.982) −0.039 (−0.596) −0.142⁎⁎⁎

(−3.543) 0.003⁎⁎⁎ (3.184) −0.008⁎⁎⁎ (−3.092) 0.007 (1.176) 0.157⁎ (1.798) −2.685⁎⁎⁎ (−3.258) 0.031 (0.417) 0.106⁎⁎⁎

(−3.433) 0.074⁎⁎⁎ (3.539) −0.008⁎⁎ (−2.127) 0.004⁎⁎

NII

(10.259) 0.176 (1.253) 0.474 (1.231) 0.228 (1.241) 0.003 (0.698) 0.012 (1.246) 0.001 (0.265) 0.098 (0.387) 1.786⁎⁎⁎ (7.685) −0.019⁎⁎⁎

Year Adj.-R2 F N

(−7.254) Yes 0.545 27.896 76

(−6.854) Yes 0.351 33.865 76

(−2.658) Yes 0.530 23.973 76

(4.521) Yes 0.559 42.236 76

(6.128) Yes 0.356 35.854 76

Time FSITREAT Time_FSITREAT State List Size Loan growth NIM

(2.265) 0.049⁎⁎⁎ (3.095) −2.527⁎⁎⁎ (−6.985) −0.021 (−0.218) 0.512⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

significantly negative in the CARit and Liquidityit models and positive in the LLP and NPL models. That is, directors and managers assigned by FSIs have weaker effects on capital risk, liquidity risk, credit risk, and asset quality in state-owned banks than in nonstate-owned banks. This further supports the above findings that FSIs have weaker effects on state-owned banks. In addition, some of the FSI banks in our sample have more than one director and manager. For example, the Bank of Communications has two directors and managers (Mr. Peter Wong and Mr. Meilun Shi). The China Guangfa Bank has six directors (Mr. Michael Zink, Mr. R. Daniel Massey, Mr. Raymond Lim, Mr. Richard Stanley, Mr. Robert Morse, and Mr. Y. S. Wong). This raises the question: whether increasing the number of directors and managers (FSIDN) has stronger effects on bank risks. To test this question, we add FSIDN to Model (6) and establish Model (8). Risksit ¼ Constant þ a1 FSIDit þa2 FSIDNit þb  Controlit þ r  Year þ ε

ð8Þ

In Model (8), Risksit are measured by Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit; FSIDit is a dummy variable that equals 1 if FSIs assign directors and managers to Chinese bank i in year t and 0 otherwise; FSIDNit is the number of FSI-assigned directors and managers for bank i in year t; Controlit is a matrix of additional bank control variables, such as Stateit, Listit, Sizeit, Loan growthit, NIMit, NIIit, and CTIit; and ε is the disturbance term. All variables are measured in Table A-2. Table 18 presents the OLS regression estimates for Model (8) with Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit as the dependent variables. FSIDit and FSIDNit are not significant in the Z-Scoreit model, showing that FSI-assigned directors and managers do not affect insolvency risk. FSIDit and FSIDNit are all significantly positive in the CARit and Liquidityit models and negative in the LLPit and NPLit models, suggesting that FSIs assigning more directors and managers more strongly affects bank risks. This result is similar to the finding by Wu et al. (2015), who observe stronger FSI effects in banks with more FSI-assigned directors. This is mainly due to the fact that more directors and managers provide more management skills, corporate governance, and financial operational techniques to local banks. 6.2. Effects of foreign strategic investor exits on bank risks. Chinese policymakers and commercial banks expect FSIs to not only increase capital level and improve corporate governance, performance, and risk, but also to function as long-term strategic partners (Xu et al., 2006; Zhu et al., 2008; Zhang and Wang, 2011). However, contrary to their initial expectations, some FSIs began to sell their stakes in Chinese banks in 2007. This setback has also spread to three of the big four state-owned banks and to some joint-stock banks. Given this background, we question whether FSI exits from banks affect their Chinese partners. In the theory part, the paper mainly mentions about two approaches

160

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 9 Interaction effects between FSIs and state ownership on bank risks (2006 to 2014). Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

21.545⁎⁎⁎

1.845⁎⁎⁎

0.874⁎⁎⁎

2.579⁎⁎⁎

(9.652) 0.163 (1.132) 0.415 (1.102) 0.285 (1.325) −0.102 (−1.032) 0.001 (1.001) 0.009 (1.001) 0.003 (0.854) 0.054 (0.457) 0.352⁎⁎

(12.514) 0.223 (1.264) 0.404⁎⁎⁎

(12.657) 0.072 (1.224) 0.412⁎⁎⁎

(6.152) −0.417 (−0.537) −0.357⁎⁎⁎

0.864⁎⁎⁎ (12.548) −0.116 (−0.326) −0.368⁎⁎⁎

(3.154) 0.294⁎⁎⁎ (2.875) −0.119⁎⁎ (−2.523) −0.197 (−0.854) 0.037⁎⁎⁎

(2.938) 0.296⁎⁎⁎ (2.897) −0.103⁎⁎ (−2.237) 0.085 (1.128) 0.032⁎⁎⁎

(−2.631) −0.237⁎⁎⁎ (−3.768) 0.087⁎⁎⁎ (3.246) 0.002⁎⁎⁎ (2.954) −0.021⁎⁎⁎

(−2.664) −0.251⁎⁎⁎ (−3.585) 0.096⁎⁎⁎ (2.764) 0.154⁎⁎⁎ (2.684) −0.014⁎⁎⁎

(2.798) −0.011⁎⁎⁎ (−4.198) 0.582 (0.456) 3.142 (0.158) −0.125 (−0.584) −0.851⁎⁎⁎ (−5.954) Yes 0.361 35.352 76

(2.845) −0.085⁎⁎⁎ (−2.685) −1.258 (−0.457) −5.458 (−0.358) −0.114 (−1.127) −0.039⁎⁎⁎ (−2.851) Yes 0.542 24.757 76

(−2.854) 0.004 (0.624) 0.086⁎⁎ (2.154) −0.146⁎⁎⁎

(−2.648) 0.010⁎⁎⁎ (4.546) 0.581⁎⁎⁎ (6.215) −3.219⁎⁎⁎

(−9.118) 0.054 (0.854) 0.025⁎⁎⁎ (2.985) Yes 0.568 44.876 76

(−5.108) 0.521 (0.516) 0.548⁎⁎⁎ (3.578) Yes 0.367 37.857 76

Time FSITREAT Time_FSITREAT Time_FSITREAT_State State List Size Loan growth NIM NII CTI Year Adj.-R2 F N

(6.589) −0.201⁎⁎⁎ (−5.658) −0.021⁎⁎⁎ (−2.658) Yes 0.552 28.574 76

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level.

that FSIs help reduce bank risks. One is the spillover effect view, which suggests that FSIs provide advanced management knowledge that helps improve the risk management ability of commercial banks. The other is the monitoring effect view, which suggests that FSIs serve as outside monitors of target banks and improve the banks' risk-taking behavior and risk management. If the first channel works more, the effects of FSIs on bank risks may remain even after FSI exits, as knowledge has obtained during the FSIs' existence period. That is, FSI exits do not affect the bank risks of their partners. If the second channel works, bank risks will increase after FSI exits, as the monitors have left. To investigate this question, we use the PSM approach described in Subsection 4.2.1 to match control banks. In this section, the treatment banks are the banks from which some FSIs have exited and the control banks refer to banks from which no FSIs have exited. Before using the DID models, we compute the average value for every variable used in the DID models (see below) for the 4 years9 before or after the FSI exit for the sample matched by PSM. For every bank, we make two observations, the observation before treatment and the observation after treatment. Finally, we obtain a sample with 13 treatment banks (26 observations) and 13 control banks (26 observations). We estimate DID in Model (9) to examine whether FSI exits from China's banks affect their Chinese partners. Risks ¼ Constant þ a1 Time þ a2 FSIEXIT þ a3  Time FSIEXIT þ b  Control þ r  Year þ ε

ð9Þ

In Model (9), Risks are measured by Z-Score⁎, CAR⁎, Liquidity⁎, LLP⁎, and NPL⁎, which are the average values of Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit for the 4 years before or after an FSI exits from a bank (or its matched bank), where Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit could be seen in Subsection 4.3. Time is a dummy variable that equals 1 after an FSI exits a bank (or its matched bank) and 0 before it exits; FSIEXIT is a dummy variable that equals 1 if a bank experiences FSI exits and 0 otherwise; Time_FSIEXIT is the interaction variable between Time and FSIEXIT; Control is a matrix of additional bank controls, including State, List, Size, Loan growth, NIM, NII, and CTI; State, dummy variable, equals 1 if a bank is a state-owned commercial bank and 0 otherwise; List, dummy variable, equals 1 if a bank went public and 0 otherwise; Size, Loan growth, NIM, NII, and CTI are average values for Sizeit, Loan growthit, NIMit, NIIit, and CTIit for the 4 years before or after an FSI exits from a bank (or its matched bank), where Z-Scoreit, CARit, Liquidityit, LLPit, NPLit, Sizeit, Loan growthit, NIMit, NIIit, and CTIit could be seen in Subsection 4.3; Year is a matrix of year dummy variables; and ε is the disturbance term. 9 Based on our data, we use 4-year period. Some banks have less than 4 annual observations before or after FSI exits. We use as many year observations as possible to compute the average values for each variable.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

161

Table 10 Effects of FSIs on bank risks in an alternative sample. Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

21.358⁎⁎⁎

1.258⁎⁎⁎

0.487⁎⁎⁎

2.685⁎⁎⁎

(9.851) 0.162 (1.292) 0.443 (1.243) 0.227 (1.214) 0.001 (0.574) 0.006 (1.245) 0.001 (0.578) 0.051 (0.482) 0.548⁎⁎⁎ (6.851) −0.192⁎⁎⁎

(12.478) 0.224 (1.342) 0.412⁎⁎⁎

(12.854) 0.075 (1.332) 0.425⁎⁎⁎

(6.982) −0.412 (−0.765) −0.366⁎⁎⁎

0.487⁎⁎⁎ (9.258) −0.121 (−0.434) −0.383⁎⁎⁎

(3.675) 0.258⁎⁎⁎

(3.428) 0.236⁎⁎⁎

(−2.837) −0.183⁎⁎⁎

(−2.929) −0.195⁎⁎⁎

(2.869) −0.125 (−1.116) 0.026⁎⁎⁎ (2.856) −0.006⁎⁎⁎

(2.877) 0.078 (1.428) 0.027⁎⁎⁎ (2.849) −0.038⁎⁎⁎

(−3.263) 0.116⁎⁎⁎ (2.864) −0.009⁎⁎⁎ (−2.597) 0.006⁎⁎⁎

(−4.518) 0.547 (0.648) 3.454 (0.462) −0.157 (−0.478) −0.458⁎⁎⁎

(−2.682) −1.241 (−0.754) −5.821 (−0.851) −0.138 (−1.184) −0.037⁎⁎⁎

(−3.292) 0.001⁎⁎⁎ (2.924) −0.014⁎⁎⁎ (−2.915) 0.004 (0.428) 0.064⁎⁎ (2.148) −0.358⁎⁎⁎ (−9.458) 0.071 (0.548) 0.045⁎⁎⁎

(−5.458) Yes 0.354 35.879 198

(−2.842) Yes 0.524 25.394 198

(2.682) Yes 0.556 43.295 198

(3.254) Yes 0.349 37.484 198

Time FSITREAT Time_FSITREAT State List Size Loan growth NIM NII CTI Year Adj.-R2 F N

(−5.248) −0.021⁎⁎⁎ (−2.685) Yes 0.538 28.678 198

(4.682) 0.438⁎⁎⁎ (6.453) −3.157⁎⁎⁎ (−5.374) 0.596 (0.284) 0.458⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level.

Table 11 Interaction effects between FSIs and state ownership on bank risks in an alternative sample. Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

1.234⁎⁎⁎ (4.656) 0.216 (1.244) 0.408⁎⁎⁎ (3.778) 0.293⁎⁎⁎ (2.986) −0.118⁎⁎

0.523⁎⁎⁎ (11.598) 0.073 (1.274) 0.432⁎⁎⁎ (3.564) 0.284⁎⁎⁎ (2.907) −0.106⁎⁎

2.324⁎⁎⁎ (12.854) −0.407 (−0.738) −0.373⁎⁎⁎ (−2.967) −0.228⁎⁎⁎ (−3.346) 0.091⁎⁎

0.864⁎⁎⁎ (9.729) −0.116 (−0.428) −0.386⁎⁎⁎ (−3.283) −0.252⁎⁎⁎ (−3.436) 0.083⁎⁎⁎

(−2.335) −0.112 (−0.665) 0.032⁎⁎⁎ (3.285) −0.001⁎⁎

(−2.436) 0.064 (1.386) 0.031⁎⁎⁎ (2.657) −0.126⁎⁎⁎

(2.458) 0.004⁎⁎

(2.765) 0.272⁎⁎ (2.336) −0.019⁎⁎⁎ (−2.923) 0.013⁎⁎⁎

(−2.326) 0.378 (1.178) 3.272 (0.657) −0.034 (−0.744) −0.024⁎⁎⁎

(−2.656) −1.528 (−1.229) −5.276 (−0.669) −0.066 (−1.245) −0.056⁎⁎⁎

(2.327) −0.020⁎⁎⁎ (−3.291) 0.002 (0.86) 0.078⁎⁎

CTI

21.203⁎⁎⁎ (11.382) 0.157 (1.165) 0.439 (1.324) 0.265 (1.414) −0.079 (−1.156) 0.002 (0.546) 0.017 (0.853) 0.001 (1.253) 0.038 (1.219) 0.228⁎⁎⁎ (8.766) −0.231⁎⁎⁎ (−4.648) −0.019⁎⁎⁎

(2.182) −0.472⁎⁎ (−2.357) 0.027 (0.790) 0.027⁎⁎⁎

(4.684) 0.173⁎⁎ (2.436) −3.722⁎⁎⁎ (−5.574) 0.338 (0.467) 0.554⁎⁎⁎

Year Adj.-R2 F N

(−3.768) Yes 0.546 30.768 198

(−2.697) Yes 0.366 38.546 198

(−2.776) Yes 0.537 28.236 198

(2.787) Yes 0.566 46.264 198

(2.865) Yes 0.361 40.235 198

Time FSITREAT Time_FSITREAT Time_FSITREAT_State State List Size Loan growth NIM NII

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level.

162

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 12 Effects of FSIs on bank risks by using alternative models. Variables

Z-Scoreit

CARit

Liquidityit

LLPit

NPLit

Cons

15.237⁎⁎⁎

3.285⁎⁎⁎ (10.294) 0.485⁎⁎⁎ (5.106) 0.315⁎⁎⁎

1.325⁎⁎⁎ (10.238) 0.687⁎⁎⁎ (7.305) 0.289⁎⁎⁎

4.245⁎⁎⁎ (6.363) −0.507⁎⁎⁎ (−4.639) −0.257⁎⁎⁎

1.204⁎⁎⁎ (7.342) −0.593⁎⁎⁎ (−4.395) −0.253⁎⁎⁎ (−3.879) 0.215⁎⁎⁎

(3.284) −0.023⁎⁎⁎ (−3.726) 0.012 (0.673) 0.126⁎⁎

(3.261) −0.017⁎⁎⁎ (−3.295) 0.013⁎⁎⁎ (4.157) 0.548⁎⁎⁎

CTIit

(3.625) 0.143 (1.265) 0.043⁎⁎⁎ (3.523) −0.056⁎⁎⁎ (−3.473) −1.527 (−0.839) −5.547 (−0.737) −0.092 (−1.033) −0.031⁎⁎

(−3.845) 0.004⁎⁎⁎

(4.366) −0.123⁎⁎⁎ (−4.295) −0.014⁎⁎

(3.593) −0.325 (−1.436) 0.034⁎⁎⁎ (3.584) −0.011⁎⁎⁎ (−5.247) 0.683 (0.839) 3.449 (0.558) −0.095 (−0.392) −0.329⁎⁎⁎

(2.448) −0.326⁎⁎⁎ (−6.357) 0.063 (0.437) 0.036⁎⁎

(7.342) −2.115⁎⁎⁎ (−5.676) 0.473 (0.235) 0.329⁎⁎⁎

Year Adj.-R2 F N

(−2.326) Yes 0.452 24.235 1095

(−5.120) Yes 0.274 31.249 1095

(−2.532) Yes 0.443 21.338 1095

(2.325) Yes 0.469 37.483 1095

(2.893) Yes 0.273 31.435 1095

Selectionit FSIit Stateit Listit Sizeit Loan growthit NIMit NIIit

(8.382) 0.327 (1.243) 0.269 (1.354) 0.007 (0.903) 0.012 (1.420) 0.003 (0.856) 0.078 (0.759) 0.529⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table A-2 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level.

We first conduct comparison analyses. Table 19 presents the comparison results for treatment banks and control banks using the DID approach. The findings suggest that FSI exits have no significant effects on bank risks above and beyond selection effects. Then, the OLS regression is estimated for Model (9) with Z-Score⁎, CAR⁎, Liquidity⁎, LLP⁎, and NPL⁎ as the dependent variables. Table 13 Interaction effects between FSIs and state ownership on bank risks using alternative models. Variables

Z-Scoreit

CARit

Liquidityit

LLPit

NPLit

Cons

3.164⁎⁎⁎ (9.859) 0.479⁎⁎⁎ (4.839) 0.426⁎⁎⁎ (4.255) −0.164⁎⁎⁎

1.278⁎⁎⁎ (9.879) 0.682⁎⁎⁎ (6.893) 0.369⁎⁎⁎ (3.858) −0.158⁎⁎⁎

4.157⁎⁎⁎ (6.183) −0.498⁎⁎⁎ (−4.584) −0.362⁎⁎⁎ (−3.976) 0.175⁎⁎

1.159⁎⁎⁎ (6.983) −0.586⁎⁎⁎ (−4.264) −0.382⁎⁎⁎ (−4.285) 0.184⁎⁎

(−3.593) −0.335 (−1.364) 0.032⁎⁎⁎ (3.327) −0.009⁎⁎⁎

(−3.162) 0.137 (1.216) 0.038⁎⁎⁎ (3.337) −0.052⁎⁎⁎

(2.436) 0.004⁎⁎⁎

(2.275) 0.207⁎⁎⁎ (3.115) −0.015⁎⁎⁎ (−3.197) 0.012⁎⁎⁎

(−4.982) 0.669 (0.795) 3.436 (0.483) −0.092 (−0.475) −0.322⁎⁎⁎

(−3.264) −1.474 (−0.775) −5.534 (−0.649) −0.088 (−1.366) −0.027⁎⁎⁎

(3.176) −0.021⁎⁎⁎ (−3.265) 0.011 (0.595) 0.124⁎⁎

CTIit

15.204⁎⁎⁎ (7.936) 0.319 (1.169) 0.354 (1.536) −0.135 (−1.143) 0.008 (0.685) 0.011 (1.287) 0.003 (0.785) 0.074 (0.685) 0.523⁎⁎⁎ (4.287) −0.119⁎⁎⁎ (−4.657) −0.012⁎⁎

(2.365) −0.324⁎⁎⁎ (−6.143) 0.061 (0.675) 0.031⁎⁎⁎

(6.938) −2.113⁎⁎⁎ (−5.449) 0.472 (0.436) 0.307⁎⁎⁎

Year Adj.-R2 F N

(−2.258) Yes 0.456 25.658 1095

(−4.853) Yes 0.279 33.547 1095

(−2.657) Yes 0.447 22.798 1095

(2.686) Yes 0.479 38.698 1095

(3.295) Yes 0.284 32.879 1095

Selectionit FSIit FSIit_Stateit Stateit Listit Sizeit Loan growthit NIMit NIIit

(3.972) 0.543⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table A-2 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

163

Table 14 Effects of FSIs on bank risks (before and after 2007). Variables

Z-Score

Panel A: Banks introducing FSIs before 2007 Cons 22.438⁎⁎⁎ (12.384) Time 0.148 (0.837) FSITREAT 0.432 (1.321) Time_FSITREAT 0.213 (1.138) State 0.004 (1.246) List 0.025 (1.319) 0.004 Size (0.647) Loan growth 0.053 (0.647) NIM 0.308⁎⁎ (7.335) NII −0.207⁎⁎⁎

CAR

Liquidity

LLP

NPL

2.032⁎⁎⁎ (13.294) 0.208 (1.104) 0.427⁎⁎⁎

0.847⁎⁎⁎ (12.324) 0.072 (1.039) 0.438⁎⁎⁎

2.847⁎⁎⁎ (6.8373) −0.427 (−0.2843) −0.383⁎⁎⁎

0.902⁎⁎⁎ (12.382) −0.093 (−0.146) −0.392⁎⁎⁎

(3.836) 0.242⁎⁎⁎ (2.736) −0.247 (−0.326) 0.046⁎⁎

(3.374) 0.223⁎⁎⁎ (2.829) 0.126 (1.302) 0.039⁎⁎

(−2.776) −0.170⁎⁎⁎ (−2.904) 0.006⁎⁎⁎ (2.849) −0.031⁎⁎

(−2.822) −0.176⁎⁎⁎ (−2.839) 0.233⁎⁎ (2.327) −0.034⁎⁎

(2.448) −0.018⁎⁎⁎

(2.414) −0.131⁎⁎⁎ (−2.597) −1.256 (−0.524) −5.425 (−0.257) −0.108 (−1.427) −0.089⁎⁎

(−2.457) 0.007 (0.547) 0.134⁎ (1.740) −0.308⁎⁎⁎ (−8.658) 0.226 (0.548) 0.025⁎⁎⁎

(−2.323) 0.022⁎⁎⁎

CTI

(−5.454) −0.055⁎⁎

(−4.837) 0.432 (0.653) 3.435 (0.438) −0.091 (−0.318) −0.633⁎⁎⁎

Year Adj.-R2 F N

(−2.284) Yes 0.537 28.463 84

(−5.473) Yes 0.352 34.372 84

(−2.192) Yes 0.523 24.302 84

(2.685) Yes 0.556 42.935 84

(3.225) Yes 0.349 36.485 84

1.232⁎⁎⁎ (9.573) 0.198 (1.192) 0.416⁎⁎⁎

0.685⁎⁎⁎ (11.236) 0.046 (1.151) 0.427⁎⁎⁎

2.453⁎⁎⁎ (5.386) −0.382 (−0.275) −0.374⁎⁎⁎

0.769⁎⁎⁎ (10.029) −0.083 (−0.134) −0.388⁎⁎⁎

(3.575) 0.228⁎⁎ (2.552) −0.242 (−0.385) 0.036⁎⁎

(3.353) 0.216⁎⁎ (2.563) 0.116 (1.125) 0.032⁎⁎

(−2.882) −0.157⁎⁎⁎ (−2.756) 0.004⁎⁎⁎

(−2.836) −0.165⁎⁎⁎ (−2.648) 0.198⁎⁎

(2.574) −0.114⁎⁎ (−2.335) −1.236 (−0.486) −5.427 (−0.342) −0.109 (−1.375) −0.086⁎⁎ (−2.135) Yes 0.515 23.236 56

(2.192) −0.026⁎⁎ (−2.436) 0.013⁎⁎⁎ (4.352) 0.418⁎⁎ (2.564) −3.342⁎⁎⁎

(7.393) −0.204⁎⁎⁎ (−5.256) −0.053⁎⁎ (−2.164) Yes 0.534 26.975 56

(2.463) −0.018⁎⁎⁎ (−4.368) 0.425 (0.597) 3.352 (0.476) −0.087 (−0.318) −0.625⁎⁎⁎ (−5.473) Yes 0.345 33.267 56

(2.684) −0.024⁎⁎ (−2.634) 0.004 (0.538) 0.114⁎ (1.756) −0.289⁎⁎⁎ (−8.658) 0.226 (0.446) 0.022⁎⁎⁎ (2.658) Yes 0.545 40.495 56

(−5.443) 0.419 (0.154) 0.709⁎⁎⁎ (3.365) Yes 0.342 34.284 56

0.742 0.438

0.492 0.474

0.624 0.336

0.424 0.527

0.473 0.346

Panel B: Banks introducing FSIs in and after 2007 Cons 20.392 ⁎⁎⁎ (9.274) Time 0.137 (0.837) FSITREAT 0.432 (1.343) Time_FSITREAT 0.187 (0.839) State 0.005 (1.204) List 0.021 (1.338) 0.003 Size (0.478) Loan growth 0.052 (0.584) NIM 0.332⁎⁎ NII CTI Year Adj.-R2 F N Chow test F P

(4.547) 0.433⁎⁎⁎ (2.774) −3.354⁎⁎⁎ (−5.244) 0.416 (0.163) 0.708⁎⁎⁎

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% levels.

Table 20 presents the results. Time_FSIEXIT in every column is not significant, indicating that FSI exits do not affect the bank risks of their partners, which parallels our comparison analyses. According to these results, it seems the first channel (i.e., spillover effects) works more than the second one (i.e., monitoring effects) in the context of China's financial background.

164

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 15 Effects of FSI on bank risks (before and after 2007). Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

21.293⁎⁎⁎

1.395⁎⁎⁎

0.773⁎⁎⁎

2.553⁎⁎⁎

(10.153) 0.153 (0.948) 0.426 (1.204) 0.204 (1.012) 0.374 (0.372) 0.043 (0.436) 0.003 (1.224) 0.023 (1.296) 0.005 (0.537) 0.055 (0.754) 0.315⁎⁎

(10.547) 0.212 (1.219) 0.414⁎⁎⁎

(11.868) 0.073 (1.195) 0.422⁎⁎⁎

(6.446) −0.439 (−0.364) −0.367⁎⁎⁎

0.883⁎⁎⁎ (10.765) −0.101 (−0.165) −0.384⁎⁎⁎

(3.427) 0.235⁎⁎⁎

(3.194) 0.221⁎⁎⁎

(−2.732) −0.163⁎⁎⁎

(−2.702) −0.172⁎⁎⁎

(2.623) 0.272 (0.193) 0.032 (0.483) −0.245 (−0.318) 0.042⁎⁎ (2.347) −0.021⁎⁎⁎ (−4.475) 0.428 (0.684) 3.458 (0.536) −0.095 (−0.325) −0.647⁎⁎⁎ (−5.582) Yes 0.349 33.635 140

(2.664) 0.195 (0.478) 0.036 (0.525) 0.122 (1.265) 0.037⁎⁎ (2.376) −0.127⁎⁎ (−2.415) −1.247 (−0.527) −5.446 (−0.352) −0.116 (−1.436) −0.097⁎⁎ (−2.285) Yes 0.520 23.745 140

(−2.875) 0.327 (0.212) −0.045 (−0.572) 0.004⁎⁎⁎

(−2.784) 0.435 (0.284) −0.064 (−0.376) 0.221⁎⁎

(2.769) −0.027⁎⁎ (−2.436) 0.006 (0.642) 0.119⁎

(2.226) −0.031⁎⁎ (−2.246) 0.018⁎⁎⁎ (4.436) 0.421⁎⁎⁎

(1.796) −0.315⁎⁎⁎ (−8.854) 0.237 (0.583) 0.027⁎⁎⁎ (2.795) Yes 0.551 41.863 140

(2.685) −3.376⁎⁎⁎ (−5.562) 0.423 (0.186) 0.716⁎⁎⁎ (3.476) Yes 0.345 35.937 140

Time FSITREAT Time_FSITREAT Shock Time_FSITREAT_Shock State List Size Loan growth NIM NII CTI Year Adj.-R2 F N

(8.336) −0.211⁎⁎⁎ (−5.447) −0.058⁎⁎ (−2.301) Yes 0.536 27.558 140

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

Table 16 Effects of FSID on bank risks. Variables

Z-Scoreit

CARit

Liquidityit

LLPit

NPLit

Cons

6.453⁎⁎⁎ (6.453) 0.073 (0.475) 0.024 (1.325) 0.126 (1.324) 0.021 (1.354) 0.012 (1.124) 0.154⁎⁎⁎

3.425⁎⁎⁎ (3.245) 0.094⁎⁎

1.425⁎⁎⁎ (5.458) 0.072⁎⁎

2.231⁎⁎⁎ (6.473) −0.054⁎⁎

1.245⁎⁎⁎ (9.454) −0.079⁎⁎⁎

(2.175) −0.325⁎⁎

(−2.125) 0.013⁎⁎⁎

(−2.768) 0.176⁎⁎⁎

(5.467) −0.254⁎⁎ (−2.546) 0.012⁎ (1.754) 0.023⁎⁎

(3.543) −0.326⁎⁎⁎ (−2.768) 0.268⁎⁎⁎ (4.768) 0.135⁎

CTIit

(6.245) −0.209⁎⁎ (−2.174) −0.035⁎⁎

(−2.153) 0.463⁎⁎⁎ (2.767) −0.014⁎⁎ (−2.325) 0.143 (1.099) 2.242 (0.454) −0.011 (−0.317) −0.065⁎⁎

(2.227) 0.325 (0.657) 0.362⁎⁎ (2.336) −0.325⁎⁎⁎ (−2.657) −1.165 (−0.372) −3.524 (−0.554) −0.046 (−0.394) −0.098⁎⁎

(2.212) −0.523⁎⁎ (−2.021) 0.017 (0.574) 0.087⁎

(1.758) −3.214⁎⁎⁎ (−5.232) 0.254 (0.465) 0.764⁎⁎

Year Adj.-R2 F N

(−2.432) Yes 0.451 35.342 330

(−2.332) Yes 0.245 39.687 330

(−2.325) Yes 0.456 29.586 330

(1.905) Yes 0.448 58.394 330

(2.436) Yes 0.317 43.453 330

FSIDit Stateit Listit Sizeit Loan growthit NIMit NIIit

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table A-2 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

165

Table 17 Interaction effects between FSID and state ownership on bank risks. Variables

Z-Scoreit

CARit

Liquidityit

LLPit

NPLit

Cons

6.786⁎⁎⁎

3.653⁎⁎⁎

1.527⁎⁎⁎

2.354⁎⁎⁎

(6.669) 0.088 (0.676) −0.064 (−0.884) 0.026 (1.445) 0.121 (1.265) 0.024 (1.436) 0.008 (1.083) 0.157⁎⁎⁎

(3.547) 0.113⁎⁎⁎ (2.589) −0.094⁎⁎

(5.685) 0.143⁎⁎⁎ (2.658) −0.079⁎⁎

(6.684) −0.092⁎⁎ (−2.542) 0.067⁎⁎

1.365⁎⁎⁎ (9.965) −0.121⁎⁎⁎ (−2.986) 0.073⁎⁎⁎ (2.646) 0.182⁎⁎⁎ (3.657) −0.322⁎⁎⁎ (−2.637) 0.274⁎⁎⁎ (4.986) 0.127⁎

CTIit

(−2.512) 0.330 (0.896) 0.357⁎⁎ (2.226) −0.343⁎⁎⁎ (−2.785) −1.143 (−0.343) −3.535 (−0.587) −0.049 (−0.415) −0.093⁎⁎

(2.325) 0.014⁎⁎⁎ (5.663) −0.251⁎⁎ (−2.458) 0.017* (1.790) 0.021⁎⁎

(6.364) −0.213⁎⁎ (−2.215) −0.033⁎⁎

(−2.325) −0.332⁎⁎ (−2.342) 0.456⁎⁎⁎ (2.685) −0.018⁎⁎ (−2.543) 0.126 (1.046) 2.2453 (0.516) −0.013 (−0.354) −0.062⁎⁎

(2.176) −0.526⁎⁎ (−2.231) 0.018 (0.587) 0.085⁎

(1.754) −3.223⁎⁎⁎ (−5.336) 0.257 (0.487) 0.761⁎⁎

Year Adj.-R2 F N

(−2.353) Yes 0.458 37.315 330

(−2.225) Yes 0.253 42.435 330

(−2.264) Yes 0.461 31.356 330

(1.876) Yes 0.454 62.364 330

(2.352) Yes 0.323 45.364 330

FSIDit FSIDit_Stateit Stateit Listit Sizeit Loan growthit NIMit NIIit

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table A-2 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

Finally, we divide FSIs into two categories: remained FSIs (i.e., FSIs that remain in China's banks in our sample period) and exited FSIs (i.e., FSIs that exited from China's banks in our sample period) and examine their effects on bank risks. We argue that if spillover effects work more than monitoring effects, the effects of FSIs on bank risks would not differ between banks Table 18 Effects of FSIDN on bank risks. Variables

Z-Scoreit

CARit

Liquidityit

LLPit

NPLit

Cons

6.554⁎⁎⁎ (6.685) 0.061 (0.424) 0.026 (0.253) 0.022 (1.215) 0.133 (1.437) 0.015 (1.135) 0.013 (1.244) 0.151⁎⁎⁎

3.582⁎⁎⁎

1.524⁎⁎⁎

(3.337) 0.071⁎⁎ (2.098) 0.035⁎

(5.645) 0.060⁎⁎ (2.153) 0.025⁎⁎

2.365⁎⁎⁎ (6.754) −0.042⁎⁎ (−2.024) −0.019⁎

1.338⁎⁎⁎ (9.658) −0.071⁎⁎⁎ (−2.635) −0.024⁎⁎ (−2.023) 0.169⁎⁎⁎

(5.436) −0.261⁎⁎⁎ (−2.646) 0.010⁎ (1.694) 0.025⁎⁎

(3.132) −0.348⁎⁎⁎ (−2.975) 0.259⁎⁎⁎ (4.325) 0.140⁎

(2.436) −0.516⁎

CTIit

(1.982) 0.317 (0.548) 0.370⁎⁎ (2.427) −0.312⁎⁎ (−2.546) −1.169 (−0.574) −3.508 (−0.393) −0.049 (−0.427) −0.102⁎⁎

(−1.923) 0.012⁎⁎⁎

(5.384) −0.212⁎⁎ (−2.180) −0.037⁎⁎

(1.892) −0.321⁎⁎ (−2.012) 0.471⁎⁎⁎ (2.879) −0.011⁎⁎ (−2.265) 0.145 (1.327) 2.230 (0.265) −0.014 (−0.332) −0.068⁎⁎

(−1.893) 0.020 (0.616) 0.089⁎⁎

(1.845) −3.201⁎⁎⁎ (−4.635) 0.257 (0.523) 0.767⁎⁎

Year Adj.-R2 F N

(−2.464) Yes 0.454 36.453 330

(−2.437) Yes 0.248 40.231 330

(−2.547) Yes 0.451 30.273 330

(2.125) Yes 0.453 59.327 330

(2.552) Yes 0.322 45.203 330

FSIDit FSIDNit Stateit Listit Sizeit Loan growthit NIMit NIIit

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table A-2 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

166

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 19 The comparison results for FSI exits.

Treatment banks (1) Before FSI exits (2) After FSI exits Difference A (2) − (1) t-Statistic Control banks (1) Before FSI exits (2) After FSI exits Difference B (2) − (1) t-statistic

Obs.

Z-Score⁎

CAR⁎

Liquidity⁎

LLP⁎

NPL⁎

13 13

23.437 22.453 −0.984 −0.858

0.070 0.067 −0.003 −0.437

0.308 0.305 −0.003 −0.528

0.007 0.007 0.000 0.253

0.027 0.030 0.003 1.135

13 13

23.269 22.757 −1.112 −0.907

0.073 0.068 −0.005 −0.683

0.312 0.305 −0.007 −0.732

0.008 0.007 −0.001 −0.748

0.028 0.030 0.002 0.977

0.128 0.142

0.002 0.241

0.004 0.583

0.001 0.672

0.001 0.523

Difference A − difference B t-Statistic

Note: Z-Score⁎, CAR⁎, Liquidity⁎, LLP⁎, and NPL⁎ are average values for Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit for the 4 years before or after FSI exits from a bank (or its matched bank), where Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit could be seen in Section 4.3.

without or with exited FSIs, as knowledge has obtained during the FSIs' existence period. However, if monitoring effects work more than spillover effects, the effects of FSIs on bank risks would differ between the banks without and with exited FSIs as the monitors have left. To examine this view, we split our sample into banks without and with exited FSIs and perform separate regression estimations for each group. Table 21 reports the results. Panel A shows the results using banks without exited FSIs as a research sample. Panel B shows the results using banks with exited FSIs as a research sample. The results in these two panels are similar and consistent with our main findings. The Chow tests (Chow, 1960) show that the effects of FSIs on bank risks are not different in every model in these two panels, suggesting that the effects of FSIs on bank risks do not differ between banks without

Table 20 Effects of FSI exits on bank risks. Variables

Z-Score⁎

CAR⁎

Liquidity⁎

LLP⁎

NPL⁎

Cons

0.827⁎⁎⁎ (6.463) −0.067 (−0.432) −0.124 (−0.241) 0.075 (0.463) −0.232 (−0.536) 0.032⁎⁎ (2.325) −0.016⁎⁎⁎ (−4.124) 0.353 (0.467) 3.136 (0.345) −0.074 (−0.386) −0.492⁎⁎⁎

0.546⁎⁎⁎ (8.107) −0.036 (−0.235) −0.094 (−0.204) 0.021 (0.274) 0.118 (1.178) 0.021⁎⁎ (2.253) −0.115⁎⁎ (−2.023) −1.126 (−0.433) −5.029 (−0.175) −0.085 (−1.115) −0.084⁎⁎⁎

2.128⁎⁎⁎ (6.125) 0.153 (0.245) 0.103 (0.842) 0.052 (0.258) 0.002⁎⁎

0.563⁎⁎⁎ (6.372) 0.135 (0.353) 0.219 (0.547) 0.061 (0.194) 0.218⁎⁎

(2.537) −0.022⁎⁎ (−2.332) 0.004 (0.436) 0.084⁎

(2.114) −0.028⁎⁎ (−2.396) 0.011⁎⁎⁎ (3.353) 0.329⁎⁎

CTI

19.283⁎⁎⁎ (8.283) −0.087 (−0.345) −0.245 (−0.423) 0.092 (0.328) 0.004 (1.546) 0.012 (1.115) 0.003 (0.352) 0.037 (0.546) 0.254⁎⁎ (5.835) −0.182⁎⁎⁎ (−4.328) −0.048⁎⁎

(1.756) −0.302⁎⁎⁎ (−7.357) 0.153 (0.357) 0.026⁎⁎

(2.356) -3.143⁎⁎⁎ (−4.382) 0.375 (0.297) 0.485⁎⁎⁎

Year Adj.-R2 F N

(−2.103) Yes 0.513 25.342 52

(−4.237) Yes 0.325 30.493 52

(−3.535) Yes 0.502 21.029 52

(2.435) Yes 0.526 38.925 52

(3.124) Yes 0.319 32.874 52

Time FSIEXIT Time_FSIEXIT State List Size Loan growth NIM NII

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. Z-Score⁎, CAR⁎, Liquidity⁎, LLP⁎, and NPL⁎, which are the average values of Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit for the 4 years before or after an FSI exits from a bank (or its matched bank), where Z-Scoreit, CARit, Liquidityit, LLPit, and NPLit could be seen in Subsection 4.3. Time is a dummy variable that equals 1 after an FSI exits a bank (or its matched bank) and 0 before it exits; FSIEXIT is a dummy variable that equals 1 if a bank experiences FSI exits and 0 otherwise; Time_FSIEXIT is the interaction variable between Time and FSIEXIT; Control is a matrix of additional bank controls, including State, List, Size, Loan growth, NIM, NII, and CTI; State, dummy variable, equals 1 if a bank is a state-owned commercial bank and 0 otherwise; List, dummy variable, equals 1 if a bank went public and 0 otherwise; Size, Loan growth, NIM, NII, and CTI are average values for Sizeit, Loan growthit, NIMit, NIIit, and CTIit for the 4 years before or after an FSI exits from a bank (or its matched bank), where Z-Scoreit, CARit, Liquidityit, LLPit, NPLit, Sizeit, Loan growthit, NIMit, NIIit, and CTIit could be seen in Subsection 4.3; Year is a matrix of year dummy variables; and ε is the disturbance term. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

167

Table 21 Effects of FSIs on bank risks (sub sample). Variables

Z-Score

Panel A: Banks without exited FSIs Cons 24.352⁎⁎⁎ (13.242) Time 0.165 (1.103) FSITREAT 0.386 (1.120) Time_FSITREAT 0.225 (1.224) State 0.004 (1.324) List 0.018 (1.241) 0.005 Size (0.483) Loan growth 0.048 (0.684) NIM 0.228⁎⁎ (7.356) NII −0.217⁎⁎⁎

CAR

Liquidity

LLP

NPL

1.647⁎⁎⁎ (11.324) 0.223 (1.336) 0.392⁎⁎⁎

0.868⁎⁎⁎ (12.438) 0.078 (1.372) 0.403⁎⁎⁎

2.964⁎⁎⁎ (6.928) −0.421 (−0.446) −0.353⁎⁎⁎

1.428⁎⁎⁎ (11.274) −0.124 (−0.231) −0.371⁎⁎⁎

(3.326) 0.257⁎⁎⁎ (2.748) −0.257 (−0.426) 0.037⁎⁎

(2.926) 0.237⁎⁎⁎ (2.768) 0.133 (1.436) 0.025⁎⁎

(−2.628) −0.186⁎⁎⁎ (−3.293) 0.006⁎⁎⁎ (3.242) −0.023⁎⁎

(−2.634) −0.191⁎⁎⁎ (−2.970) 0.236⁎⁎ (2.525) −0.031⁎⁎

(2.452) -0.017⁎⁎⁎

(2.143) −0.115⁎⁎ (−2.254) −1.105 (−0.419) −4.642 (−0.164) −0.134 (−1.564) −0.105⁎⁎

(−2.436) 0.005 (0.463) 0.083⁎ (1.726) −0.245⁎⁎⁎ (−8.405) 0.246 (0.724) 0.036⁎⁎⁎

(−2.197) 0.014⁎⁎⁎ (4.265) 0.327⁎⁎ (2.523) −3.164⁎⁎⁎ (−5.134) 0.433 (0.353) 0.735⁎⁎⁎

CTI

(−5.562) −0.064⁎⁎

(−4.437) 0.372 (0.837) 3.153 (0.325) −0.107 (−0.525) −0.657⁎⁎⁎

Year Adj.-R2 F N

(−2.336) Yes 0.537 27.564 88

(−5.734) Yes 0.352 33.348 88

(−2.415) Yes 0.523 23.584 88

(2.906) Yes 0.554 41.904 88

(3.761) Yes 0.347 36.492 88

1.653⁎⁎⁎ (12.382) 0.227 (1.342) 0.401⁎⁎⁎

0.844⁎⁎⁎ (12.493) 0.080 (1.302) 0.409⁎⁎⁎

2.639⁎⁎⁎ (6.938) −0.415 (−0.437) −0.347⁎⁎

(3.218) 0.241⁎⁎⁎ (2.625) −0.245 (−0.293) 0.041⁎⁎

(2.723) 0.219⁎⁎⁎ (2.635) 0.122 (1.298) 0.026⁎⁎

(−2.532) −0.162 (−2.902) 0.003⁎⁎⁎

0.900⁎⁎⁎ (12.382) −0.116 (−0.224) −0.37954⁎⁎ (−2.354) −0.168⁎⁎⁎ (−2.748) 0.218⁎⁎

(2.395) -0.1302⁎⁎ (−2.334) −1.252 (−0.285) −4.823 (−0.192) −0.114 (−1.341) −0.073⁎⁎ (−2.179) Yes 0.513 23.154 52

(2.287) −0.031⁎⁎ (−2.294) 0.013⁎⁎⁎ (4.127) 0.329⁎⁎ (2.534) −3.243⁎⁎⁎

(7.382) −0.182⁎⁎⁎ (−5.243) −0.035⁎⁎ (−2.104) Yes 0.532 27.028 52

(2.379) -0.021⁎⁎⁎ (−4.294) 0.421 (0.627) 3.120 (0.365) −0.085 (−0.634) −0.436⁎⁎⁎ (−5.364) Yes 0.345 32.482 52

(2.829) −0.024⁎⁎ (−2.519) 0.005 (0.485) 0.084⁎ (1.723) −0.291⁎⁎⁎ (−8.384) 0.236 (0.765) 0.023⁎⁎⁎ (2.685) Yes 0.545 40.458 52

(−3.942) 0.447 (0.436) 0.602⁎⁎⁎ (3.245) Yes 0.341 35.153 52

0.874 0.231

0.538 0.385

0.729 0.274

0.492 0.472

1.039 0.205

Panel B: Banks with exited FSIs Cons 21.436⁎⁎⁎ (11.823) Time 0.162 (1.024) FSITREAT 0.421 (1.029) Time_FSITREAT 0.212 (1.001) State 0.002 (1.142) List 0.020 (1.283) 0.005 Size (0.435) Loan growth 0.054 (0.730) NIM 0.203⁎⁎ NII CTI Year Adj.-R2 F N Chow test F P

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

exited FSIs and banks with exited FSIs. These results support the view mentioned above that spillover effects work more than monitoring effects in the context of China's financial background. Furthermore, this paper adds the interaction variable Time_FSITREAT_FSIEXIT to Model (1) and form Model (10) to verify this view. Table 22 presents the results. Time_FSITREAT_FSIEXIT is not significant in every model, revealing that the effects of FSIs on bank risks do not differ between

168

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table 22 Effects of FSIs on bank risks (sub sample). Variables

Z-Score

CAR

Liquidity

LLP

NPL

Cons

24.284⁎⁎⁎

1.768⁎⁎⁎

2.54⁎⁎⁎

(13.242) 0.149 (0.857) 0.428 (1.283) 0.221 (1.210) 0.032 (0.528) 0.129 (0.476) 0.004 (1.185) 0.018 (1.106) 0.07 (0.732) 0.053 (0.956) 0.309⁎⁎

(11.420) 0.204 (1.143) 0.412⁎⁎⁎

0.879⁎⁎⁎ (12.403) 0.067 (1.105) 0.421⁎⁎⁎

(7.453) −0.384 (−0.326) −0.367⁎⁎⁎

0.995⁎⁎⁎ (11.283) −0.102 (−0.156) −0.384⁎⁎⁎

(3.485) 0.238⁎⁎⁎

(3.291) 0.227⁎⁎⁎

(−2.876) −0.163⁎⁎⁎

(−2.739) −0.176⁎⁎⁎

(2.834) 0.056 (0.837) −0.024 (−0.073) −0.246 (−0.284) 0.043⁎⁎

(2.758) 0.045 (0.634) −0.072 (0.063) 0.122 (1.293) 0.024⁎⁎

(−2.984) 0.184 (0.932) 0.329 (0.573) 0.004⁎⁎⁎

(−2.746) 0.074 (0.336) 0.184 (0.387) 0.216⁎⁎

(2.275) −0.024⁎⁎⁎ (−4.658) 0.434 (0.938) 3.497 (0.526) −0.089 (−0.320) −0.648⁎⁎⁎ (−5.879) Yes 0.351 33.476 140

(2.329) −0.136⁎⁎ (−2.543) −1.256 (−0.538) −5.465 (−0.305) −0.121 (−1.365) −0.101⁎⁎ (−2.438) Yes 0.522 23.546 140

(2.749) −0.022⁎⁎ (−2.483) 0.007 (0.648) 0.123⁎

(2.154) −0.031⁎⁎ (−2.236) 0.019⁎⁎⁎ (4.658) 0.416⁎⁎⁎

(1.940) −0.332⁎⁎⁎ (−9.784) 0.234 (0.457) 0.033⁎⁎⁎ (2.906) Yes 0.550 41.658 140

(2.895) −3.434⁎⁎⁎ (−5.659) 0.412 (0.198) 0.736⁎⁎⁎ (3.787) Yes 0.344 35.464 140

Time FSITREAT Time_FSITREAT FSIEXIT Time_FSITREAT_FSIEXIT State List Size Loan growth NIM NII CTI Year Adj.-R2 F N

(8.674) −0.208⁎⁎⁎ (−4.215) −0.064⁎⁎ (−2.443) Yes 0.537 27.547 140

Note: We estimate all regressions using OLS. t statistics (in parentheses) are corrected for heteroskedasticity following White's (1980) methodology. See Table 3 for variable definitions and measurements. ⁎⁎⁎ Significance at the 1% level. ⁎⁎ Significance at the 5% level. ⁎ Significance at the 10% level.

banks without and with exited FSIs. Obviously, these two kinds of FSIs have similar effects on bank risks. This is mainly due to the fact that the CBRC has announced the rules for FSIs at the end of 2003, formally defined in five criteria at the end of 2005, as mentioned above. These criteria urge these two kinds of FSIs to keep the same behavior from the beginning, and thus they have similar effects. Risks ¼ Constant þ a1 Time þ a2 FSITREAT þ a3 Time FSITREAT þ a4 FSITREAT þ a5  Time FSITREAT FSIEXIT þ b  Control þ r  Year þ ε

ð10Þ

In Model (10), Risks are measured by Z-Score, CAR, Liquidity, LLP, and NPL; Control is a matrix of additional bank controls, containing State, List, Size, Loan growth, NIM, NII, and CTI; Time_FSITREAT is the interaction variable between Time and FSITREAT; FSIEXIT is a dummy variable that equals 1 if a bank experiences FSI exits and 0 otherwise; Time_FSITREAT_FSIEXIT is the interaction variable between Time_FSITREAT and FSIEXIT; Year is a matrix of year dummy variables; and ε is the disturbance term. All of the variables are defined or measured in Table 3. 7. Conclusions. Over the past decade, China's banking industry has experienced extraordinary ownership reforms. Introducing FSIs has undoubtedly played an important role in the ownership reforms adopted by Chinese banks. Policymakers and commercial banks expected the introduction of FSIs to not only increase capital and improve corporate governance but also to provide long-term strategic partnerships. FSIs expected to obtain profit and gain a foothold in the Chinese economy. However, contrary to the expectations of FSIs, Chinese policymakers, and commercial banks, some FSIs began to sell their shares of Chinese banks in 2007. This setback has also spread to three of the big four state-owned and joint-stock banks. Most of these foreign investments were profitable, but what have Chinese banks obtained from this campaign? Using data from China's banks between 1995 and 2014, we employ the PSM-DID approach to investigate the effects of FSIs on bank risks, such as insolvency risk, capital risk, liquidity risk, credit risk, and asset quality. We find the following results. First, bank risks, except for insolvency risk, are reduced after introducing FSIs as a result of spillover and monitoring effects. Second, according to the ownership concentration, social, and corporate governance views of state-owned banks, the effects of FSIs on capital

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

169

Appendix A

Table A-1 The status of China's banks introducing FSIs. Chinese banks

FSIs

Acquisition year

Ownership shares of FSIs (%)

Exit time

Nature of Chinese banks

Bank of China

Royal Bank of Scotland Fullerton Financial Holdings Swiss bank Asian Development Bank The Bank of Tokyo-Mitsubishi UFJ Bank of America Bank of America Fullerton Financial Holdings Goldman Sachs Group Allianz Group American Express Company Standard Chartered Bank Hongkong Bank International Finance Corporation Asia Financial Holdings Citigroup

2005 2005 2005 2005 2006 2005 2008 2005 2006 2006 2006 2010 2004 2003 2004 2003

10 5 1.55 0.24 0.19 9.1 20 5.1 6.05 2.36 0.47 0.37 19.9 1.22 4.55 4.62

2009 2007 2008 2012 – 2009 2011 – 2010 2009 2011 – – – 2007 2012

State

2004 2004 2004 2004 2004 2004 1996 1999 2006 2006 2006 2007 2005 2006 2006 2006 2008 2005 1999 2001 2001

5 4 15.98 17.89 7 17.89 1.9 7 20 8 4.74 4.83 6.88 7.02 2.88 4.08 15.38 19.99 5 2 8

– 2012 – 2010 2010 2010 2007 2007 – 2013 – – 2007 – – – – – – 2011 –

2001 2005 2005

15 19.2 19.9

2005 – –

2005 2004 2004 2004 2005 2006 2005 2005 2005 2004 2006 2006 2006 2006 2007 2007 2008 2008 2009 2007 2008

5 2.5 2.5 10.68 19.92 5 19.2 10 3.3 20 12.2 19.9 7.99 17 19.99 4.98 20 4.98 19.99 19.99 19.99

– – – – – – – – – 2008 – –

China Construction Bank

Industrial and Commercial Bank of China Agricultural Bank of China The Bank of Communications China Minsheng Bank Shanghai Pudong Development Bank Industrial Bank

Deal TEDA Investment International Finance Corporation Hang Seng Bank Limited Shenzhen Development Bank U.S. Newbridge Capital Group GE Capital Ping An Bank U.S. Newbridge Capital Group China Everbright Bank Asian Development Bank International Finance Corporation China Guangfa Bank Citibank Waterhouse Investment IBM Credit CITIC Bank Banco Bilbao Vizcaya Argentaria Huaxia Bank Pangu Bank Deutsche Bank Germany and Luxembourg Companies Sal. Oppenheim Company Evergrowing Bank Singapore's United Overseas Bank Bohai Bank Standard Chartered Bank Bank of Shanghai International Finance Corporation International Finance Corporation Hong Kong and Shanghai Banking Corporation Limited Bank of Nanjing International Finance Corporation BNP Paribas Bank of Beijing China Merchants Securities Standard Chartered-ING BANK N.V International Finance Corporation Xi'an City Commercial Bank International Finance Corporation Bank of Nova Scotia Qilu Bank Commonwealth Bank of Australia Hangzhou City Commercial Bank Commonwealth Bank of Australia Asian Development Bank Nanchong City Commercial Bank BNP Paribas German Investment and Development Co. German Savings bank Changsha City Commercial Bank International Finance Corporation Bank of Ningbo Singapore's OCBC Bank Bank of Tianjin ANZ Bank Chongqing City Commercial U.S. Investment Fund Carlyle Bank Dah Sing Bank, Hong Kong Qingdao Bank Intesa Sanpaolo, Italy Rothschild Financial Group Holding Co. Yantai Bank Hang Seng Bank Wing Lung Bank Xiamen Bank Fubon Bank (Hong Kong) Ltd. Chengdu Bank Malaysia Hong Leong Bank Bank of Yingkou Malaysia CIMB Bank Group

– – – – – – – –

State

State

State Non-state Non-state Non-state Non-state

Non-state Non-state Non-state Non-state

Non-state Non-state

Non-state Non-state Non-state

Non-state Non-state

Non-state Non-state Non-state Non-state

Non-state Non-state Non-state Non-state Non-state Non-state Non-state Non-state Non-state

(continued on next page)

170

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Table A-1 (continued) Chinese banks

FSIs

Acquisition year

Ownership shares of FSIs (%)

Exit time

Nature of Chinese banks

Union Bank of Hangzhou

Rabobank International Finance Corporation International Financial Institutions Hana Bank Bank of East Asia International Finance Corporation International Finance Corporation

2006 2006 2005 2010 2003 2009 2008

10 5

Non-state

18 15 15 10

– – – – – – –

Non-state Non-state Non-state Non-state Non-state

ANZ Bank

2006

19.99

–

Non-state

Tian An Investment The Bocg Investment Co. HK CTS Wkland Investment Co.

2003 2010 2012 2013

10 11 19.99 16

– – – –

Non-state Non-state Non-state Non-state

Harbin Bank Bank of Jilin Shenzhen City Commercial Bank Deyang Bank Tianjin Binhai Rural Commercial Bank Shanghai Rural Commercial Bank Dalian Bank Bank of Ningxia Jiaozuo City Commercial Bank Huishang Bank

Note: These materials were collected from the banks' public reports, relevant newspapers, and financial magazines.

risk, liquidity risk, credit risk, and asset quality are weaker in state-owned banks than in non-state-owned banks. Third, FSIs assigning directors and managers reduces capital risk, liquidity risk, and credit risk, and improves bank asset quality, as foreign banks can provide domestic banks with more direct and convenient risk management assistance and risk behavior monitoring if FSIs assigned directors and managers to local banks. And directors and managers assigned by FSIs have weaker effects in state-owned banks. Finally, we find that bank risks do not significantly change after FSI exits. According to this result, it seems that spillover effects work more than monitoring effects in the context of China's financial background. We also find that the effects of FSIs on bank risks do not differ between banks without and with exited FSIs, further suggesting the view that spillover effects are more effective than monitoring effects. This article makes several contributions to the literature. First, we examine not only the effects of FSIs on bank risks but conduct further analyses to determine whether FSI exits from China's banks affect their partners. Second, due to the particularity of state ownership, we are the first to posit that state ownership moderates the effects of FSIs. Third, this paper examines the ways that FSIs influence domestic banks and explores whether FSI-assigned directors and managers affect the risks of Chinese banks. Finally, we are the first to use the PSM-DID approach to control for selection bias and endogeneity problems, allowing us to demonstrate the true treatment effects of FSIs. Acknowledgements We are grateful for Iftekhar Hasan, Cheng Li, Vikas Mehrotra, Hong Zhao, Mingming Zhou, and seminar participants at the Xi'an Jiaotong University and University of Alberta for their helpful comments. We are also thanks for Rui Ma and Qian Guo from Xi'an Jiaotong University for the data collection and processing. Maoyong Cheng gratefully acknowledges the financial Table A-2 Variable definitions. Symbol

Definition

Z-Scoreit CARit Liquidityit LLPit NPLit Selectionit

The return on assets plus the capital-asset ratio divided by the standard deviation of asset returns for bank i in year t The ratio of book value of equity to total assets for bank i in year t The ratio of liquid assets to the sum of deposits and short-term funding for bank i in year t The ratio of loan loss provision to total loans for bank i in year t The ratio of impaired (non-performing) loans to net loans for bank i in year t Dummy variable, equals 1 if a bank introduced FSIs over the sample period and 0 otherwise Dummy variable, equals 1 beginning in the second year following a bank introduced FSIs and equals 0 for the years before a bank introduced FSIs and the banks that did not introduce FSIs Dummy variable, equals 1 if FSIs assigning directors and managers to Chinese bank i in year t, and equal to 0 otherwise Dummy variable, equals 1 if a bank is a state-owned commercial bank and 0 otherwise Dummy variable, equals 1 if a banks went public and 0 otherwise The natural logarithm of total assets for bank i in year t The growth of total loans for bank i in year t

FSIit FSIDit Stateit Listit Sizeit Loan growthit NIMit NIIit CTIit Year

The ratio of net interest income to total earning assets for bank i in year t The ratio of non-interest income to gross revenues for bank i in year t The cost to income ratio for bank i in year t Dummy variables, equals 1 if the banks refer to the corresponding year when they (or their matched bank) introduced FSIs and 0 otherwise

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

171

support from the National Natural Science Foundation of China (Grant Numbers 71403202, 71472148, 71502138, and 71572144), China Postdoctoral Science Foundation (Grant Nos. 2015 M572532 and 2016 T90897), and Postdoctoral Science Foundation of Shaanxi Province. All errors are our own. References Aggarwal, R., Erel, I., Ferreira, M., Matos, P., 2011. Does governance travel around the world? Evidence from institutional investors. J. Financ. Econ. 100, 154–181. Angkinand, A., Wihlborg, C., 2010. Deposit insurance coverage, ownership, and banks' risk-taking in emerging markets. J. Int. Money Financ. 29, 252–274. Ariss, R.T., 2010. On the implications of market power in banking: evidence from developing countries. J. Bank. Financ. 34, 765–775. Barry, T.A., Lepetit, L., Tarazi, A., 2011. Ownership structure and risk in publicly held and privately owned banks. J. Bank. Financ. 35, 1327–1340. Berger, A.N., Clarke, G.R.G., Cull, R., Klapper, L., Udell, G.F., 2005. Corporate governance and bank performance: a joint analysis of the static, selection, and dynamic effects of domestic, foreign, and state ownership. J. Bank. Financ. 29, 2179–2221. Berger, A.N., Hasan, I., Zhou, M., 2009. Bank ownership and efficiency in China: What will happen in the world's largest nation? J. Bank. Financ. 33, 113–130. Bernanke, B., 1983. Nonmonetary effects of the financial crisis in the propagation of the great depression. Am. Econ. Rev. 73, 257–276. Bonin, J.P., Hasan, I., Wachtel, P., 2005. Privatization matters: bank efficiency in transition countries. J. Bank. Financ. 29, 2155–2178. Brandt, L., Li, H.B., 2003. Bank discrimination in transition economies: ideology, information or incentives? J. Comp. Econ. 31, 387–413. Buckley, P.J., Clegg, J., Wang, C., 2007. Is the relationship between inward FDI and spillover effects linear? An empirical examination of the case of China. J. Int. Bus. Stud. 38, 447–459. Calomiris, C., Mason, J., 1997. Contagion and bank failures during the Great Depression: the Chicago banking panic of June 1932. Am. Econ. Rev. 87, 863–884. Calomiris, C., Mason, J., 2003a. Consequences of US bank distress during the depression. Am. Econ. Rev. 93, 937–947. Calomiris, C., Mason, J., 2003b. Fundamentals, panics and bank distress during the depression. Am. Econ. Rev. 93, 1615–1647. Chang, G., 2013. Global Banks are ‘Divorcing’ China (http://www.forbes.com/sites/gordonchang/2013/04/21/global-banks-are-divorcing-china/#6ea5ce491e19). Chen, S., Liao, C., 2011. Are foreign banks more profitable than domestic banks? Home- and host-country effects of banking market structure, governance, and supervision. J. Bank. Financ. 35, 819–839. Choi, S., Hasan, I., 2005. Ownership, governance, and bank performance: Korean experience. Financ. Markets Inst. Inst. 14, 215–242. Chow, G.C., 1960. Tests of equality between sets of coefficients in two linear regressions. Econometrica 28, 591–605. Claessens, S., Laeven, L., 2004. What drives bank competition? Some international evidence. J. Money, Credit, Bank. 36, 563–583. Clarke, G.R.G., Crivelli, J.M., Cull, R., 2005. The direct and indirect impact of bank privatization and foreign entry on access to credit in Argentina’s provinces. J. Bank. Financ. 29, 5–29. Cull, R., Xu, L.C., 2003. Institutions, ownership, and finance: the determinants of profit reinvestment among Chinese firms. J. Financ. Econ. 77, 117–146. Demirguc-Kunt, A., Huizinga, H., 2010. Bank activity and funding strategies: the impact on risk and returns. J. Financ. Econ. 98, 626–650. Detragiache, E., Gupta, P., 2006. Foreign banks in emerging market crises: evidence from Malaysia. J. Financ. Stab. 2, 217–242. Firth, M., Lin, C., Liu, P., Wong, S.M.L., 2009. Inside the black box: bank credit allocation in China's private sector. J. Bank. Financ. 33, 1144–1155. Fries, S., Taci, A., 2005. Cost efficiency of banks in transition: evidence from 289 banks in 15 post-communist countries. J. Bank. Financ. 29, 55–81. García-Herrero, A., Gavilá, S., Santabárbara, D., 2009. What explains the low profitability of Chinese banks? J. Bank. Financ. 33, 2080–2092. Gulamhussen, M.A., Guerreiro, L., 2009. The influence of foreign equity and board membership on corporate strategy and internal cost management in Portuguese banks. Manag. Account. Res. 20, 6–17. Haber, S.H., Musacchio, A., 2004. Foreign Banks and the Mexican Economy, 1997–2004. Division of Research, Harvard Business School. Hasan, I., Marton, K., 2003. Development and efficiency of the banking sector in a transitional economy: Hungarian experience. J. Bank. Financ. 27, 2249–2271. Hasan, I., Xie, R., 2013. Foreign bank entry and bank corporate governance in China. Emerg. Mark. Financ. Trade 49, 4–18. Havrylchyk, O., Jurzyk, E., 2011. Inherited or earned? Performance of foreign banks in Central and Eastern Europe. J. Bank. Financ. 35, 1291–1302. Hillion, P., Vermaelen, T., 2004. Death spiral convertibles. J. Financ. Econ. 71, 381–415. Iannotta, G., Giacomo, N., Sironi, A., 2007. Ownership structure, risk and performance in the European banking industry. J. Bank. Financ. 31, 2127–2149. Imbierowicz, B., Rauch, C., 2014. The relationship between liquidity risk and credit risk in banks. J. Bank. Financ. 40, 242–256. Jia, C., 2009. The effect of ownership on the prudential behavior of banks—the case of China. J. Bank. Financ. 33, 77–87. Jiang, C., Yao, S., Feng, G., 2013. Bank ownership, privatization, and performance: evidence from a transition country. J. Bank. Financ. 37, 3364–3372. Keeley, M.C., 1990. Deposit insurance, risk, and market power in banking. Am. Econ. Rev. 80, 1183–1200. Lee, C.C., Hsieh, M.F., 2014. Bank reforms, foreign ownership, and financial stability. J. Int. Money Financ. 40, 204–224. Lensink, R., Meesters, A., Naaborg, I., 2008. Bank efficiency and foreign ownership: do good institutions matter? J. Bank. Financ. 32, 834–844. Lin, H., 2011. Foreign bank entry and firms' access to bank credit: evidence from China. J. Bank. Financ. 35, 1000–1010. Lin, X.C., Zhang, Y., 2009. Bank ownership reform and bank performance in China. J. Bank. Financ. 33, 20–29. Lu, C.H., Shen, C.H., Wu, M.W., Wu, Z.W., 2010. Cooperation satisfaction and performance: empirical evidence from Chinese banks and their foreign strategic investors. Chin. World. Econ. 18, 90–108. Martinez-Peria, M.S., Mody, A., 2004. How foreign participation and market concentration impact bank spreads: evidence from Latin America. J. Money, Credit, Bank. 36, 510–537. Micco, A., Panizza, U., Yanez, M., 2007. Bank ownership and performance. Does politics matter? J. Bank. Financ. 31, 219–241. Mohsni, S., Otchere, I., 2014. Risk taking behavior of privatized banks. J. Corp. Finan. 29, 122–142. Nichols, D., Wahlen, J., Wieland, M., 2009. Publicly-traded versus privately-held: implications for conditional conservatism in bank accounting. Rev. Acc. Stud. 14, 88–122. Roberts, M.R., Whited, T.M., 2012. Endogeneity in Empirical Corporate Finance (http://ssrn.com/abstract=1748604). Sapienza, P., 2004. The effect of government ownership on bank lending. J. Financ. Econ. 72, 357–384. Shehzad, C.T., Haan, J.D., Scholtens, B., 2010. The impact of bank ownership concentration on impaired loans and capital adequacy. J. Bank. Financ. 34, 399–408. Shen, C.H., Lu, C.H., Wu, M.W., 2009. Impact of foreign bank entry on the performance of Chinese banks. Chin. World. Econ. 17, 102–121. Spencer, J.W., 2008. The impact of multinational enterprise strategy on indigenous enterprises: horizontal spillovers and crowding out in developing countries. Acad. Manag. Rev. 33, 341–361. Sun, J., Harimaya, K., Yamori, N., 2013. Regional economic development, strategic investors, and efficiency of Chinese city commercial banks. J. Bank. Financ. 37, 1602–1611. Taboada, A.G., 2011. The impact of changes in bank ownership structure on the allocation of capital: international evidence. J. Bank. Financ. 35, 2528–2543. Tirole, J., 2001. Corporate governance. Econometrica 69, 1–35. Villalonga, B., 2004. Does diversification cause the “diversification discount”? Financ. Manag. 33, 5–27. Wei, Y., Liu, X., 2006. Productivity spillovers from R&D, exports and FDI in China's manufacturing sector. J. Int. Bus. Stud. 37, 544–557. White, H., 1980. A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica 48, 817–838. Wu, M.W., Shen, C.H., Lu, C.H., Chan, C.C., 2012. Impact of foreign strategic investors on earnings management in Chinese banks. Emerg. Mark. Financ. Trade 48, 115–133. Wu, M.W., Shen, C.H., Lu, C.H., 2015. Do more foreign strategic investors and more directors improve the earnings smoothing? The case of China. Int. Rev. Econ. Financ. 36, 3–16. Xu, Y., 2011. Towards a more accurate measure of foreign bank entry and its impact on domestic banking performance: the case of China. J. Bank. Financ. 35, 886–901. Xu, G.P., Ge, R.R., He, X.D., 2006. Research on the equilibrium price of equity transfer of state-owned commercial banks. Financ. Res. 3 16–2. (In Chinese).

172

M. Cheng et al. / Pacific-Basin Finance Journal 40 (2016) 147–172

Zhang, Y., Li, H., 2010. Innovation search of new ventures in a technology cluster: the role of ties with service intermediaries. Strateg. Manag. J. 31, 88–109. Zhang, J.H., Wang, P., 2011. Research on bank efficiency and determinants: an international comparison. Financ. Res. 5, 13–28 (In Chinese). Zhu, W., Yang, J., 2016. State ownership, cross-border acquisition, and risk-taking: evidence from China's banking industry. J. Bank. Financ. (Forthcoming). Zhu, Y., Zeng, Y., Li, P., He, J., 2008. Chinese banks introduce foreign strategic investors: Bachground, argues and review. Manage. World 1, 22–37 (In Chinese).