Do bank regulations affect board independence? A cross-country analysis

Journal of Banking & Finance 37 (2013) 2714–2732

Contents lists available at SciVerse ScienceDirect

Journal of Banking & Finance journal homepage: www.elsevier.com/locate/jbf

Do bank regulations affect board independence? A cross-country analysis Li Li a, Frank M. Song b,c,⇑ a

School of International Trade and Economics, University of International Business and Economics, China School of Economics and Finance, University of Hong Kong, China c School of Economics, Peking University, China b

a r t i c l e

i n f o

Article history: Received 13 April 2010 Accepted 20 March 2013 Available online 16 April 2013 JEL classification: G3 G28 L51 O16 Keywords: Bank regulation Internal governance Board independence

a b s t r a c t Based on the hand-collected board structure data of 277 listed banks across 55 countries, and the bank regulation and supervision database compiled by the World Bank, this paper provides the first crosscountry assessment of the impacts of bank regulations on board independence of banks. In line with Beck et al. (2006), we examine the effects of two types of regulation policies, the first involving the empowerment of supervisory agencies to monitor and discipline banks directly, and the second focusing on encouraging private monitoring of banks through requiring disclosure of more accurate and complete information. We find that empowering official supervisory agencies to discipline banks directly reduces board independence, but encouraging private sector monitoring of banks increases it. The findings suggest that the first type of regulations tends to crowd out the internal governance of banks, while the second crowds in it. We also find that the legal system with better investor rights protection and better contracts enforcement not only increases board independence but also enhances the crowding in effect of promoting private monitoring and decreases the crowding out effect of direct official supervision on board independence. Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction In almost all countries, banks are intensively regulated because they are vulnerable to systematic risks. Several cross-country studies have investigated the effects of the national regulation of banks (Barth et al., 2004, 2006; Beck et al., 2006; Laeven and Levine, 2009). These studies suggest that regulations which encourage private monitoring work best to promote bank development and economic growth; while those which empower direct official supervision have no positive effects on bank development, and sometimes even undermine financial stability. In addition to regulations, corporate governance is also important for banks, but its role has nevertheless been ignored by most cross-country studies on bank regulations (e.g. Barth et al., 2004, 2006; Beck et al., 2006). The first paper that studies the effects of national regulation on bank performance while also considers the influence of banks’ corporate governance structure is Laeven and Levine (2009), where it is shown that, depending on banks’ ownership structure, the same regulation policy can have different effects on their risk-taking.

⇑ Corresponding author. E-mail address: [email protected] (F.M. Song). 0378-4266/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jbankfin.2013.03.026

In contrast to Laeven and Levine (2009), who focus on the simultaneous effects of regulation and corporate governance on bank performance, we study how cross-country differences of bank regulation policies impact the internal governance arrangements of individual banks. Specifically, we investigate the impacts of two types of regulation polices on the board independence of banks. Consistent with Beck et al. (2006), the first type of regulation policy involves empowering supervisory agencies to monitor and discipline banks directly, while the second focuses on encouraging private sectors to monitor banks by requiring more accurate and timely information disclosure. These two types of regulation policies have different implications for the monitoring costs and monitoring benefits of shareholders, and should therefore affect the internal governance arrangements of banks differently – which may explain why the literature shows opposite effects from such two types of regulation policies on bank development and lending corruption (Barth et al., 2004; Beck et al., 2006). We use board independence as proxy of the internal governance of banks for two reasons. First, the degree of board independence is critical for the internal governance of firms. Outside directors monitor top managers better and play critical roles in discrete tasks such as the hiring and firing of the CEO, adopting anti-takeover devices and negotiating takeover premiums (for a useful survey, see John and Senbet, 1998; Shleifer and Vishny, 1997). The cross-country

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

study by Dahya et al. (2008) demonstrates that higher board independence is related to higher firm value and lower levels of related party transactions. Second, the board of directors is an even more important governance mechanism for banks than for non-financial firms because of the special characteristics of banking. The existence of intensive official regulation, such as deposit insurance, restrictions on ownership structures, and restrictions on banks’ entry and operations, reduces the effectiveness of other mechanisms in dealing with corporate governance problems (Billett et al., 1998; Levine, 2004). Moreover, unlike other industries, external governance mechanisms such as takeovers hardly exist in banking (Prowse, 1997; Levine, 2004); and the opaqueness of banking makes it more difficult to design incentive contracts for top managers. All these aspects emphasise the need for more effective monitoring by boards of directors in the banking area. Several cross-country studies have demonstrated the importance of board independence for banking (Andres and Vallelado, 2008; Li and Song, 2011). The literature on non-financial firms suggests that regulatory environments and policies play important roles in shaping the internal governance of firms (Kole and Lehn, 1999; Booth et al., 2002). Banks are more widely and intensively regulated than non-financial firms, and such intensive regulations are more likely to affect the internal governance arrangements of banks. Crawford et al. (1995), Hubbard and Palia (1995) and Becher et al. (2005) explore the issue from the perspective of deregulation in US banks during the 1990s. They show that deregulation leads to increased managerial discretion, and that banks respond to deregulation by improving internal monitoring through aligning the incentives of managers, directors and shareholders. In contrast to these studies, which investigate the dynamics of internal governance structure in respect of bank deregulation, we focus on the influence of cross-country differences in regulation policies on cross-country differences in banks’ board independence. To the best of our knowledge, we provide the first cross-country evidence on how different bank regulation policies affect the internal governance structure of banks, specifically their board independence. We develop our hypotheses by analysing the impacts of the two types of regulation policies on monitoring costs and monitoring benefits for bank shareholders. Direct supervision and intervention by official supervisory agencies decrease managerial discretion, which means lower monitoring benefits. The intervention of authorities also indicates higher monitoring costs for shareholders, and is therefore likely to reduce board independence. Regulation policies that focus on promoting private monitoring decrease the information asymmetry between managers and shareholders by forcing banks to disclose accurate and timely information, and thus reduce the monitoring costs of shareholders and increase board independence. However, promoting private monitoring could also reduce board independence, since increased transparency and market discipline monitor bank managers and thus reduce shareholders’ monitoring benefits. We employ hand-collected data on boards of directors in 2004– 2010 and the World Bank surveys II and III published in 2003 and 2007 on bank regulation and supervision to examine our hypotheses. The type of bank regulation policies that empowers direct official supervision by supervisory agencies is measured by official supervisory power, and the type of regulation policies that encourages private monitoring of banks is measured by private monitoring index. The degree of board independence is measured by the ratio of independent directors on the board. In robustness check we also use prompt corrective power and external ratings and credit monitoring as alternative measures of the two types of regulations respectively, and use an independent director as board chairman and the ratio of independent directors on the audit committee as alternative measures of board independence. The empirical results show a negative relationship between official supervisory power and the ratio of indepen-

2715

dent directors on the board, while a positive relationship between the private monitoring index and the ratio of independent directors on the board. The results support our hypothesis on bank regulation policies empowering direct official supervision. As for the hypothesis on regulation policies encouraging private monitoring, the results suggest that the effects of reduced monitoring costs dominate those of reduced monitoring benefits. Beyond the two main results, we find that the legal system with better investor rights protection as well as better contracts enforcement not only increases board independence but also enhances the positive effect of promoting private monitoring on board independence and decreases the negative effect of direct official supervision on board independence. An array of robustness tests supports the main results. We contribute to the literature on national regulation of banks. Most of the existing cross-country studies directly investigate the impacts of regulation policies on bank performance without exploring the channel concerned. We extend the existing studies by documenting one possible channel – internal governance and specifically the board of directors. Since board independence is critical not only for non-financial firms (Dahya et al., 2008) but also for banks (Andres and Vallelado, 2008; Li and Song, 2011), different regulation policies, which influence board independence differently, will affect bank performance in different ways. The board of directors acts as the link between regulation policies and bank performance. The results imply that the possible impacts of regulation on banks’ internal governance arrangements cannot be ignored when formulating regulation policies. We also contribute to policy considerations of bank regulation. The New Basel Accord (Basel II) sets up a bank regulation and supervision framework. The influential best-practice recommendations of Basel II are based on three pillars. Most countries around the world have implemented or plan to implement the Basel II guidelines. The debates on regulatory overhaul after the 2007 financial crisis have also emphasised all three of these pillars. However, not enough studies have provided evidence on whether these guidelines can improve bank development, facilitate capital allocation and reduce system risk (Barth et al., 2004, 2007; Beck et al., 2006). We add to the literature by supporting Pillar 3 and challenging Pillar 2, since the regulation practices emphasised by Pillar 2 substitute for the internal governance of banks, while those emphasised by Pillar 3 enhance it. In addition, we contribute to the literature on the determinants of board structure. Studies on such determinants agree that corporate board structure is endogenous to specific business environments and specific company characteristics. However, there is no consensus on which factors shape board structure (e.g., Boone et al., 2007; Guest, 2008; Linck et al., 2008; Andres et al., 2012). Moreover, prior studies seldom consider regulatory policies as one of the factors shaping board independence (Guest, 2008; Kim et al., 2007). By focusing on banks, we provide cross-country evidence on the determinants of board structure in a specific industry. We show that the regulation policies unique to the banking industry help shape the board structure of banks. Since regulatory policies are mostly external to banks, our research to some extent avoids the endogeneity problem that prior studies have had to deal with. The rest of this paper is organised as follows. In Section 2, we develop our testable hypotheses. Section 3 describes the data and variables. We test the hypotheses on how regulation practices affect board independence in Section 4, and perform robustness checks in section 5. Section 6 is given over to concluding remarks.

2. Bank regulations and board independence: hypotheses Banks, like other non-financial corporations, face potential conflicts of interest between management and shareholders. Managers

2716

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

pursue private benefits and try to maximise their own interests rather than those of shareholders. The board of directors is meant to perform the critical functions of monitoring and advising management, and to mitigate conflicts of interest. Outside directors are often thought to play the monitoring role better than insiders, and a higher level of board independence is suggested to allow for more effective monitoring of top managers (see John and Senbet, 1998, for survey). However, outside directors monitor managers at some cost, since they face information acquisition and processing costs in transforming their general expertise for the specific firm which they serve as a director. Shareholders make a tradeoff between monitoring costs and monitoring benefits (Harris & Raviv, 2008). Empirical studies on the determinants of board structure demonstrate that board independence is positively related to monitoring benefits and negatively related to monitoring costs (e.g., Boone et al., 2007; Kim et al., 2007; Linck et al., 2008). Banks differ from non-financial firms in that they are regulated and supervised intensively by governments. Such regulations aim to reduce the systematic risks, increase financial stability and protect the interests of diffused depositors. However, as suggested by Booth et al. (2002), Becher et al. (2005), Crawford et al. (1995) and Hubbard and Palia (1995), the existence of bank regulation alters the business environments of banks, changes the opportunities of managers to deviate from shareholders’ interests, and should therefore affect the internal governance structure of banks. Specifically, the existence of regulation and supervision changes the trade-off between monitoring benefits and monitoring costs, and should have an effect on the board independence of banks. There are two different types of regulation policies. The traditional regulation approach, reflecting the notion of the ‘supervisory power view’1, emphasises the role of direct regulation and intervention by supervisory agencies in promoting bank behaviour and avoiding banking crises. Influential international institutions, such as the Bank for International Settlements (BIS), International Monetary Fund (IMF) and World Bank, all encourage the development of powerful bank supervisory agencies with the authority to scrutinise and discipline banks. Basel II recommends regulatory practices that empower official supervisory agencies to discipline banks in Pillar 2. Alternatively, as suggested by the ‘private empowerment view’,2 rather than depend on direct disciplining and intervention, supervisory authorities can increase their reliance on market discipline by enhancing the abilities and incentives of private agents to overcome information costs. It has recently been stressed that discipline from market participants complements and supports the regulatory practices of banks (DeYoung et al., 2001). Pillar 3 of Basel II recommends regulatory practices that force banks to disclose accurate information to private sectors and encourages private monitoring of banks. These two types of regulatory policies have different implications for shareholders in monitoring bank managers. Empowering official supervisory agencies to scrutinise and discipline directly decreases managerial discretion and thus reduces shareholders’ incentive to monitor. Managerial discretion - that is, the decision-making latitude held by top managers - influences the incentive of shareholders to monitor managers. The net benefits of monitoring increase with managers’ opportunities to pursue private benefits, and decrease with its costs (Boone et al., 2007). Managers with less discretion have fewer opportunities to pursue

1 The ‘supervisory power view’ (Beck et al., 2006) suggests direct disciplining and monitoring of banks by supervisory agencies, since private agents frequently lack the incentives and capabilities to monitor powerful banks. 2 The ‘private empowerment view’ (Beck et al., 2006) argues that the responsibility of supervisory agencies should be to enhance the abilities and incentives of private agents to overcome information costs so that private agents can exert effective governance over banks.

private benefit, which in turn decreases the net benefits of monitoring and discourages shareholders from monitoring managers (Crawford et al., 1995; Hubbard and Palia, 1995; Booth et al., 2002; Becher et al., 2005). Under this type of regulation policies, powerful supervisory agencies have certain rights, such as to require external auditors to report any presumed involvement of senior managers in illicit activities, fraud or insider abuse, to remove or replace the management during bank restructuring and reorganisation, and to order a bank’s management to institute provisions to cover actual or potential losses. Supervisors have the right to review and supervise the capital adequacy and risk position of banks, and can intervene at an early stage to prevent capital inadequacy (BIS, 2004). The power of supervisory agencies to discipline and intervene directly largely constrains the decision-making latitude and autonomy of bank managers. Smith and Watts (1992) confirm that in regulated industries, such as banking, opportunities to invest in projects are severely limited. This limited managerial discretion leads to lower net benefits of monitoring (Crawford et al., 1995; Hubbard and Palia, 1995; Booth et al., 2002; Becher et al., 2005). Moreover, powerful supervisors even have the right to remove and replace directors and to supersede shareholder rights during bank restructuring, which entails increased monitoring costs for shareholders. Lower monitoring benefits and higher monitoring costs will reduce board independence (Boone et al., 2007; Kim et al., 2007; Linck et al., 2008). We therefore expect that regulation policies empowering official supervision will reduce internal monitoring, and thus is negatively related to board independence. Based on the arguments above, we have the following hypothesis. Hypothesis 1. The bank regulation policies that empower direct official supervision by supervisory agencies decrease board independence. Regulatory policies which focus on encouraging private monitoring of banks influence the monitoring incentives of shareholders in two opposing ways. On one hand, such policies reduce the monitoring costs of outside directors by forcing banks to disclose accurate and timely information. As Fama and Jensen (1983) argue, outside directors are less well informed about a firm’s projects. The effectiveness of monitoring by outsiders depends on the information they acquire to verify the quality of projects and managers (Raheja, 2005). That is to say, outside directors face the costs of acquiring and processing information when playing the monitoring role (Harris & Raviv, 2007), and higher information asymmetry indicates higher monitoring costs (Boone et al., 2007; Linck et al., 2008). Since banks are inherently opaque and have more serious information asymmetry than non-financial firms (Furfine, 2001; Morgan, 2002), the abilities of outside directors to monitor bank managers effectively depend crucially on the completeness and accuracy of the information they receive (Barth et al., 2007). Under this type of regulation policies, supervisors force bank managers to provide more complete and accurate information, thereby reducing information asymmetry between managers and outside directors. Pillar 3 of Basel II develops a set of disclosure requirements ‘which will allow market participants to assess key pieces of information on the scope of application, capital, risk exposures, risk assessment processes, and hence the capital adequacy of the institution’ (BIS, 2004). This type of regulations also require banks to disclose such information as off-balance-sheet items and risk management procedures, and to be audited by licensed external auditors and assessed by international credit rating agencies. All these disclosure requirements ensure outsiders acquire more complete and accurate information, which reduces monitoring costs and increases shareholders’ incentive to monitor.

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

On the other hand, regulatory policies which focus on encouraging private monitoring of banks reduce managers’ opportunities to pursue private benefit and thus reduce the monitoring incentive of shareholders. Bushman et al. (2004) claim that the transparency of firms’ operations and activities disciplines managers to act in shareholders’ interests, decreasing the demand on corporate governance systems to alleviate moral hazard problems. Market monitoring by external auditors and credit rating agencies also disciplines bank managers and constrains their abilities to acquire private benefits. Previous studies in banking show that there are substitute effects between external control devices and internal governance mechanisms and among different internal governance mechanisms (Belkhir, 2009b; Brickley and James, 1987; Schranz, 1993). Therefore, regulation policies encouraging private monitoring could also reduce the incentives of shareholders to monitor bank managers. The net effect of this type of regulatory policies on board independence depends on which of these two opposing aspects dominates. If the effect of reduced monitoring costs dominates, encouraging private monitoring will increase internal monitoring, and thus increase board independence. However, if the effect of reduced monitoring benefits dominates, encouraging private monitoring will decrease internal monitoring, and thus decrease board independence. Based on the arguments above, we have the following hypothesis. Hypothesis 2. The effect of the bank regulation policies that encourage private monitoring of banks on board independence is uncertain.

2717

3.2. Variables 3.2.1. Board independence The board independence measure is the dependent variable in our analysis. We use the ratio of independent directors on the board to measure it. In the robustness tests, we also use an independent director as the board chairman 3 and the ratio of independent directors on the audit committee 4 as alternative measures of board independence. The detailed definitions of the variables are presented in Table 1. Some banks in our sample have a two-tier board structure, in which case we collect information on the board whose main responsibility is to monitor the bank’s managers. And we include a dummy variable in the regressions to indicate whether the bank has a twotier board structure. Referring to Dahya et al. (2008), we develop our own criteria for ‘independence’ of directors and the criteria is provided in the definition of ratio of independent directors on the board in Table 1. In the case of banks that define ‘independent director’ according to the corporate governance guidelines in their annual reports, we use this self-reported classification as the starting point, and make adjustments based on our own criteria for ‘independence’. For those banks which do not define ‘independent director’, we identify independent directors manually according to our criteria. The summary statistics of ratio of independent directors on the board and other variables used in this paper is provided in Tables 2 and 3 presents the statistics of the ratio of independent directors on the board, an independent director as the board chairman, and the ratio of independent directors on the audit committee for each country. The mean values of the three variables are 46.92, 0.24 and 64.81 respectively.

3. Data and variables 3.1. The sample and data sources We use hand-collected data on boards of directors, and the bank regulation and supervision database compiled by World Bank to carry out our analyses. Following Beck et al. (2006), we employ official supervisory power and private monitoring index to measure the two types of bank regulation policies respectively. Official supervisory power and private monitoring index are derived from the World Bank surveys II and III on bank regulation and supervision in 152 countries. Data on board independence are hand-collected from banks’ annual reports and websites. The regulation and supervision database covers 152 countries around the world. Following Caprio et al. (2007), we collect the board structure data of the 10 largest listed commercial banks and BHC banks (as defined by total assets at the end of 2006) in countries that are covered by the database. The period for which we collect data is 2004 to 2010. Since many countries have fewer than 10 banks where we can obtain information on the board, and some banks do not provide the financial and/ or market information that we use as control variables, after dropping the countries with only one observation, the final sample covers 277 banks in 55 countries around the world. In addition to the data sources mentioned above, we rely on others to control for bank-level and country-level factors which might affect the board independence of banks, specifically the BankScope database to obtain financial and market factors. We obtain data on countries’ macro factors from the World Development Indicators (WDI) database, and the legal and institutional settings from La Porta et al. (1998). We build an unbalanced panel data of 1134 bank-year observations. The observations in each year of 2004–2010 are 17, 175, 207, 178, 192, 188, and 177 respectively. We conduct robustness check for sample without observations in 2004 and get similar results. Table 1 identifies the data sources, and provides definition and brief descriptions of the variables.

3.2.2. Official supervisory power and private monitoring index The two key independent variables in our study are the official supervisory power index and the private monitoring index. We use the former to measure the regulation policies which focus on empowering official supervision, and the latter to represent the regulation policies which focus on encouraging private monitoring. In robustness check, we also use prompt corrective power and external ratings and credit monitoring as alternative measures of the two types of regulation policies respectively. Official supervisory power indicates whether bank supervisory authorities can take specific actions against bank management, owners or auditors in both normal times and times of distress. It is constructed from 6 dummies that reflect the supervisory power of supervision agencies and are relevant to internal governance structure of banks. It ranges from 1 to 6, and higher values indicate greater authority for bank supervisors. Prompt corrective power reflects the power of supervisors to take ‘‘prompt, corrective action’’ as banks’ conditions deteriorate (Barth et al., 2006), higher values representing higher supervisory power. The private monitoring index reflects the degree to which bank supervisory agencies force banks to disclose more and better information to the public and induce private sector monitoring of banks. The index ranges from 4

3 We consider an independent director as board chairman better reflects the level of board independence than whether the CEO acts as the board chairman. For example, many banks in our sample have an executive director acting as board chairman. More often than not, the interests of executive directors are aligned with the CEO’s rather than the shareholders’. The Credit Lyonnais Securities Asia (CLSA) also includes an independent director as board chairman as a sub-item to construct corporate governance rankings for 495 companies in 25 countries. 4 The functions of the board of directors are carried out by board committees. As claimed by Klein (1998), the audit committee is one of such committees which play the role of monitoring management and alleviating the agency problem between managers and shareholders. The literature shows that a higher ratio of independent directors on the audit committee reflects better monitoring (e.g. Klein, 2002; Goh, 2009).

2718

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Table 1 Definitions and sources of variables. Variable Bank-level Ratio of independent directors on the board

Independent director as board chairman Ratio of independent directors on the audit committee Log(Total asset) Ln(Board size) PB ratio Capital-asset ratio Loan-asset ratio Two-tier board Cross-listing Market power Ln(No. of subsidiaries) Ownership concentration Loan loss provision ratio Liquidity ratio Cost to income ratio Country-level Official supervisory power

Private monitoring index

Capital regulatory index

Overall activities restrictiveness

Entry into banking requirements

Definition and source Number of independent directors on the board divided by board size times one hundred. A director is considered independent if if he is not (1) an employee of the bank (including manager, employee or representative of employees), (2) a major shareholder (5% or more) in the bank, (3) an employee or manager of a subsidiary of the bank, (4) a manager of a holding company of the bank, (5) an employee or manager of a bank’s major shareholder if that shareholder is a company, (6) a close relative of a bank manager or major shareholder of the bank, and (7) a manager of the bank in the preceding three years. Source: authors’ calculations based on annual reports and websites of banks Dummy variable. 1 indicates that an independent director acts as board chairman, 0 otherwise. Source: authors’ calculations based on annual reports and websites of banks Number of independent directors on the audit committee divided by audit committee size times one hundred. If a bank has not set up an audit committee, this variable is set as 0. Source: authors’ calculations based on annual reports and websites of banks Logarithm of total bank asset at book value. Source: Fitch IBCA and Bureau van Dijk – BankScope Natural logarithm of bank board size. Source: authors’ calculations based on annual reports and websites of banks. The ratio of the market value of equity to the book value of equity. Source: BankScope The ratio of bank equities to total assets at book value (%). Source: Fitch IBCA and Bureau van Dijk – BankScope The ratio of bank loans to total assets at book value (%). Source: Fitch IBCA and Bureau van Dijk – BankScope Dummy variable. 0 indicates the bank has a one-tier board system; 1 indicates a two-tier system. Source: authors’ calculations based on annual reports and websites of banks Dummy variable. 1 indicates a bank’s shares are traded either as a direct listing on a US stock exchange or as an American Depository Receipt (ADR); 0 otherwise. Source: authors’ calculations Ratio of individual bank assets to total banking system assets (%). Source: authors’ calculations using data from BankScope Natural logarithm of number of bank’s subsidiaries. Source: Fitch IBCA and Bureau van Dijk – BankScope Sum of the ownership (in percentage) held by five largest shareholders. Source: Fitch IBCA and Bureau van Dijk – BankScope, annual reports and websites of banks Ratio of the bank’s loan loss provisions to net interest income (%). Source: Fitch IBCA and Bureau van Dijk – BankScope Ratio of the bank’s liquid assets to liquid liabilities (%). Source: Fitch IBCA and Bureau van Dijk – BankScope Ratio of overheads to the sum of net interest revenue and other operation income (%).Source: Fitch IBCA and Bureau van Dijk – BankScope Official supervisory power is based on the following questions that are relevant to internal governance structure: 1. Does the supervisory agency have the right to meet external auditors to discuss their report without the approval of the bank? 2. Are auditors required by law to communicate directly to the supervisory agency any presumed involvement of bank directors or senior managers in illicit activities, fraud or insider abuse? 3. Can the supervisory authority force a bank to change its internal organisational structure? 4. Has this power been utilised in the last 5 years? 5. Can the supervisory agency order the bank’s directors or management to establish provisions to cover actual or potential losses? 6. Can the supervisory agency suspend the directors’ decision to distribute dividends? It sums up the 6 dummies and ranges from 1 (low power) to 6 (high power). Higher values indicate greater official supervisory power. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 This is based on the following questions: 1. Is an external audit a compulsory obligation for banks? 2. Are auditors licensed or certified? 3. What percentage of the top ten banks are rated by international credit rating agencies (e.g., Moody’s, Standard and Poor)? 4. How many of the top ten banks are rated by domestic credit rating agencies? 5. Is there an explicit deposit insurance protection system? 6. Were depositors wholly compensated (to the extent of legal protection) the last time a bank failed? 7. Does accrued, though unpaid, interest/principal enter the income statement while the loan is still non-performing? 8. Does accrued, though unpaid, interest/principal enter the income statement while the loan is still non-performing? 9. Are financial institutions required to produce consolidated accounts covering all bank and any non-bank financial subsidiaries? 10. Are bank directors legally liable if information disclosed is erroneous or misleading? 11. Is subordinated debt allowable (required) as part of capital? 12. Is subordinated debt required as part of regulatory capital? 13. Are off-balance-sheet items disclosed to the public? 14. Must banks disclose their risk management procedures to the public? 15. Are bank regulators/ supervisors required to make public formal enforcement actions, which include cease-and-desist orders and written agreements between a bank regulatory/supervisory body and a banking organisation? It ranges from 4 (low private monitoring) to 11 (high private monitoring). Higher values indicate more private monitoring. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 This is based on the following questions: 1. Is the minimum capital-asset ratio requirement risk-weighted in line with the Basel I guidelines? 2. Does the minimum ratio vary as a function of an individual bank’s credit risk? 3. Does the minimum ratio vary as a function of market risk? 4. Is the market value of loan losses not realised in accounting books deducted from the book value of capital before minimum capital adequacy is determined? 5. Are unrealised losses in securities portfolios deducted? 6. Are unrealised foreign exchange losses deducted? 7. Is the fraction of revaluation gains which is allowed as part of capital less than 0.75? 8. Are the sources of funds to be used as capital verified by the regulatory/supervisory authorities? 9. Can the initial disbursement or subsequent injections of capital be carried out with assets other than cash or government securities? 10. Can the initial disbursement of capital be carried out with borrowed funds? It ranges from 3 (low capital stringency) to 10 (high capital stringency). Higher values indicate greater capital stringency. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 The extent to which banks may engage in securities (underwriting, brokering, dealing and all aspects of the mutual fund industry), insurance (insurance underwriting and selling) and real estate (real estate investment, development and management) activities. In each case of securities, insurance or real estate activity, Unrestricted = 1 = full range of activities can be conducted directly in the bank; Permitted = 2 = full range of activities can be conducted, but some or all must be conducted in subsidiaries; Restricted = 3 = less than a full range of activities can be conducted in the bank or subsidiaries; and Prohibited = 4 = the activity cannot be conducted in either the bank or its subsidiaries. The overall activities restrictiveness index is the sum of restriction on the three activities. It ranges from 3 (low restrictiveness) to 12 (high restrictiveness). High values indicate greater restriction. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 The entry into banking requirement, which is a variable developed based on eight questions about whether various types of legal submissions are required to obtain a banking licence. Which of the following are legally required to be submitted before

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

2719

Table 1 (continued) Variable

Prompt corrective power

External ratings and credit monitoring

GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Rule of law Anti-director rights

Board independence requirement

Definition and source issuance of the banking licence? (1)Draft by-laws? (2) Intended organisation chart? (3) Financial projections for first 3 years? (4) Financial information on main potential shareholders? (5) Background/experience of future directors? (6) Background/ experience of future managers? (7) Sources of funds to be disbursed in the capitalisation of new bank? (8) Market differentiation intended for the new bank? The index ranges from 0 (low entry requirement) to 8 (high entry requirement). Higher values indicate greater stringency. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 Prompt corrective power is a subcomponent of official supervisory power, based on the following questions: 1. Can the supervisory authority force a bank to change its internal organizational structure? 2. Can the supervisory agency order the bank’s directors or management to constitute provisions to cover actual or potential losses? 3. Can the supervisory agency suspend the directors’ decision to distribute dividends? It ranges from 0 (low power) to 3 (high power). Higher values indicate greater prompt corrective power. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 External ratings and credit monitoring is a subcomponent of external governance index in Barth et al. (2006), based on the following questions: 1. Is subordinated debt allowable as part of capital?2. Is subordinated debt required as part of capital? 3. Do regulations require credit ratings for commercial banks? What percent of the top ten banks are rated by international credit rating agencies (e.g., Moody’s, Standard and Poor)? It ranges from 0 (low monitoring) to 4 (high monitoring). Higher values indicate more monitoring from external rating agencies and creditors. Source: World Bank Survey of Bank Regulation and Supervision, 2003, 2007; Barth et al., 2006 Real gross domestic product per capita (thousands of US$). Source: WDI database Annual real growth rate of gross domestic product (%). Source: WDI database Annual inflation rate (%, GDP deflator). Source: WDI database The median ratio of bank capital and reserves to total assets (%) of banks in the country. Source: WDI database The median ratio of the value of nonperforming loans divided by the total value of the loan portfolio (%) of banks in the country. Source: WDI database An indicator of the degree to which the country adheres to the rule of law (ranging from 0 to 6), averaged for 1990-1995. Source: International Country Risk Guide An index aggregating shareholder rights. The index is formed by adding 1 when: 1. the country allows shareholders to mail their proxy vote to the firm; 2. shareholders are not required to deposit their shares prior to the General Shareholders’ Meeting; 3. cumulative voting or proportional representation of minorities in the board of directors is allowed; 4. an oppressed minorities mechanism is in place; 5. The minimum percentage of share capital that entitles a shareholder to call for an Extraordinary Shareholders’ Meeting is less than or equal to 10% (the sample median); or 6. shareholders have preemptive rights that can only be waived by a shareholders’ vote. The index ranges from 0 to 6. Source: La Porta et al., 1998 Dummy variable, indicating whether a country recommends/requires minimum number of independent directors in its corporate governance guideline and code of best practice. Source: authors’ calculations from corporate governance guidelines and codes of all the countries

to 11, with higher values indicating a higher extent of empowering private oversight. The external ratings and credit monitoring indicates evaluations by external rating agencies and incentives for creditors of the bank to monitor bank performance, higher values representing higher private supervision. The World Bank has organised four surveys on bank regulation and supervision, with the results of surveys I, II, III, and IV published in 2000, 2003, 2007, and 2012 respectively. We use the results of the World Bank survey II on bank regulation and supervision to construct official supervisory power and private monitoring index and their alternatives for year 2004–2006, and use the results of survey III to construct corresponding measures for year 2007–2010, with the methodology following Barth et al. (2006). 3.2.3. Bank-specific control variables Our paper focuses on the influence of country-level regulation policies as determinants of bank-level governance, but we do not want to dismiss the potential importance of bank-specific factors. Based on the existing literature, we include log(total asset), capital-asset ratio, PB ratio, market power, and ln(no. of subsidiaries) to control for the differences in bank business structure (Hermalin and Weisbach, 1998; Denis and McConnell, 2003; Boone et al., 2007; Coles et al., 2007; Linck et al., 2008), include ln(board size), ownership concentration, two-tier board, and crosslisting to represent the effects of alternative governance mechanisms (Yermack, 1996; Shleifer and Vishny, 1997; Adams and Mehran, 2012; Kim et al., 2007; Belkhir, 2009a,b), and include loan-asset ratio, loan loss provision ratio, liquidity ratio and cost to income ratio to account for the special characteristics of banking firms with respect to non-financial firms (Caprio et al., 2007).

3.2.4. Country-specific control variables Our explanatory variables official supervisory power and private monitoring index are country-level variables. We control for other country-specific factors that might explain bank-level governance, including GDP per capita, GDP growth and inflation as proxies for the level of economic development and macro-economic environment (Doidge et al., 2007), banking capital-asset ratio and banking nonperforming loan as proxies for banking sector characteristics, capital regulatory index, overall activities restrictiveness index and entry into banking requirements index to account for bank regulation policies other than official supervisory power and the private monitoring index (Barth et al., 2006), rule of law and anti-director rights to proxy the legal environment of different countries (La Porta et al., 2000; Kim et al., 2007; Ferreira et al., 2011), and the board independence requirement to control for the differences in countries’ corporate governance guidelines and codes. The definitions and summary statistics of all the control variables are to be found in Tables 1 and 2. We check the correlations among the bank-level variables and country-level variables and find that multi-colinearity is not a serious problem. Most of the correlation coefficients are below 0.3, which allows us to include these variables simultaneously in the models. The correlation matrixes are reported in Panel B and Panel C of Table 2, and the variance inflation factors (VIFs) are provided in all the regression models. 4. Empirical results 4.1. Econometric model We use a panel data model to estimate how regulation policies that empower official supervision and encourage private monitoring affect board independence. Since our sample is a mixture of

2720

Table 2 Descriptive statistics and correlation matrixes. Obs

Mean

Std. dev.

Min

Max

Panel A: Descriptive statistics for variables Bank-level 1. Ratio of independent directors on the board 2. Independent director as board chairman 3. Ratio of independent directors on the audit committee 4. Log(Total asset) 5. ln(Board size) 6. PB ratio 7. Capital-asset ratio 8. Loan-asset ratio 9. Two-tier board 10. Cross-listing 11. Market power 12. ln(No. of subsidiaries) 13. Ownership concentration 14. Loan loss provision ratio 15. Liquidity ratio 16. Cost to income ratio

1134 1134 1134 1134 1134 1134 1134 1134 1134 1134 1134 1116 1134 1069 1121 1123

46.92 0.24 64.81 17.39 2.42 1.82 9.70 53.76 0.18 0.37 8.93 3.91 0.56 21.32 36.42 56.71

22.78 0.43 35.89 2.01 0.35 9.23 9.65 17.93 0.38 0.48 10.01 1.97 0.27 33.36 59.64 19.77

0.00 0.00 0.00 12.29 1.10 0.02 0.16 0.03 0.00 0.00 0.00 0.00 0.00 140.93 1.76 14.47

100.00 1.00 100.00 22.06 3.26 309.45 83.36 92.28 1.00 1.00 60.04 9.13 1.35 487.99 929.38 426.49

Country-level 17. Official supervisory power 18. Private monitoring 19. Capital regulatory index 20. Entry into banking requirements 21. Overall activities restrictiveness 22. Prompt corrective power 23. External ratings and credit monitoring 24. Rule of law 25. Anti-director rights 26. Board independence requirement 27. GDP per capita 28. GDP growth 29. Inflation 30. Banking capital-asset ratio 31. Banking non-performing loan

1134 1134 1134 1134 1134 1134 1134 1106 1002 1134 1134 1134 1134 1134 1134

4.66 8.63 5.99 7.20 6.93 2.54 1.65 4.29 3.28 0.62 16.82 3.32 4.01 7.89 3.60

1.01 1.31 1.56 1.20 1.94 0.71 0.59 1.35 1.43 0.48 13.23 3.93 4.59 2.77 2.83

2.00 5.00 2.00 0.00 3.00 0.00 1.00 1.64 0.00 0.00 0.58 17.95 23.46 2.70 0.20

6.00 11.00 9.00 8.00 12.00 3.00 4.00 6.00 5.00 1.00 56.39 14.76 28.14 14.50 19.70

1

2

3

Panel B: Correlation matrix for bank-level variables 1 1 2 0.450*** 1 3 0.448*** 0.110*** 1 4 0.298*** 0.033 0.245*** 5 0.025 0.126*** 0.194*** 6 0.053* 0.005 0.048 7 0.052* 0.050* 0.026 *** *** 8 0.091 0.128 0.042 9 0.062** 0.123*** 0.188*** 10 0.180*** 0.004 0.202*** 11 0.158*** 0.067** 0.126*** 0.049 0.273*** 12 0.323*** 13 0.313*** 0.137*** 0.236*** 14 0.014 0.037 0.001 15 0.01 0.045 0.033 16 0.012 0.014 0.002

4

5

6

7

8

9

10

11

12

13

14

15

16

1 0.489*** 0 0.416*** 0.047 0.085*** 0.545*** 0.433*** 0.778*** 0.335*** 0.086*** 0.058* 0.061**

1 0.028 0.190*** 0.007 0.251*** 0.211*** 0.182*** 0.439*** 0.198*** 0.044 0.050* 0.045

1 0.008 0 0.016 0.043 0.013 0.027 0.009 0 0.008 0.031

1 0.391*** 0.037 0.160*** 0.189*** 0.216*** 0.063** 0.126*** 0.525*** 0.047

1 0.103*** 0.014 0.041 0.094*** 0.047 0.122*** 0.434*** 0.205***

1 0.012 0.01 0.115*** 0.261*** 0.075** 0.009 0.159***

1 0.277*** 0.441*** 0.215*** 0.057* 0.005 0.028

1 0.336*** 0.028 0.033 0.043 0.132***

1 0.421*** 0.045 0.150*** 0.087***

1 0.022 0.042 0.051*

1 0.044 0.216***

1 0.191***

1

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Variable

1

This table reports descriptive statistics for variables used in the paper and the Pearson correlation matrix for bank-level and country-level variables in Panels A, B and C, respectively. Detailed variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

1 0.189*** 1 0.284*** 0.097*** 1 0.403*** 0.266*** 0.061** 1 0.342*** 0.452*** 0.362*** 0.539*** 1 0.238*** 0.03 0.111*** 0.231*** 0.164*** 1 0.126*** 0.180*** 0.119*** 0.051 0.061* 0.352*** 1 0.043 0.315*** 0.816*** 0.408*** 0.453*** 0.429*** 0.466*** 1 0.157*** 0.163*** 0.230*** 0.201*** 0.001 0.050* 0.125*** 0.154*** 1 0.126*** 0.154*** 0.173*** 0.094*** 0.089*** 0.139*** 0.035 0.144*** 0.094*** 1 0.110*** 0.121*** 0.279*** 0.102*** 0.244*** 0.370*** 0.171*** 0.256*** 0.061** 0.251***

30 29 28 27 26 25 24 23 22 21 20 19 18 17

Table 2 (continued)

Panel C: Correlation matrix for country-level variables 17 1 18 0.075** 1 19 0.192*** 0.04 1 20 0.071** 0.008 0.074** 1 21 0.017 0.064** 0.104*** 0.026 *** *** 22 0.713 0.012 0.126 0.177*** *** *** 23 0.315 0.184 0.009 0.133*** 24 0.205*** 0.157*** 0.217*** 0.103*** 25 0.116*** 0.299*** 0.136*** 0.157*** 26 0.295*** 0.048 0.151*** 0.263*** 27 0.203*** 0.231*** 0.175*** 0.086*** 28 0.104*** 0.159*** 0.147*** 0.071** 29 0.012 0.103*** 0.153*** 0.107*** 30 0.130*** 0.134*** 0.02 0.150*** 31 0.246*** 0.090*** 0.160*** 0.220***

31

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

2721

cross-sectional and time series data, employing panel data analysis allows us to control for unobservable and constant bank heterogeneity. The model estimated is as follows:

yit ¼ a þ xit b þ zi c þ ui þ kt þ v it ; where yit measures the board independence (ratio of independent directors on the board in main tests and an independent director as the board chairman and ratio of independent directors on the audit committee in the robustness check) of bank i in year t; xit stands for a vector of all the time-variant variables, including our explanatory variables (official supervision power, private monitoring index and the alternative explanatory measures in the robustness check – prompt corrective power and external ratings and credit monitoring), the other regulation and supervision variables (capital regulatory index, overall activities restrictiveness index, and entry into banking requirements index), bank-level control variables (ln(board size), log(total asset), capital-asset ratio, loanasset ratio, PB ratio, market power, cross-listing, two-tier board, ownership concentration, loan loss provision ratio, liquidity ratio, and cost to income ratio) and some of country-level control variables (GDP per capita, GDP growth, inflation, banking capital-asset ratio, and banking non-performing loan); zi represents the timeinvariant variables for bank i, including some of the country-level control variables (rule of law, anti-director rights, and board independence requirement) and some of the bank-level control variables (ln(no. of subsidiaries)); ui and kt are the bank-specific and time-specific effects for year t; vit is the residual error term. We employ generalised least square (GLS) estimation with bank random effects (RE) and robust standard errors allowing for clustering within countries in our regressions for ratio of independent directors on the board and ratio of independent directors on the audit committee. The GLS RE estimator is robust to first-order autoregressive (AR(1)) disturbances within unbalanced panel data and cross-sectional correlation and/or heteroscedasticity (if any) across panels. We allow for clustering within countries to allow for possible correlation of errors across banks within the same country. We use panel probit estimation with random effects in regressions for an independent director as the board chairman since it is a binary variable. We also provide pooled OLS and probit estimation with robust standard errors allowing for clustering within countries in several tables. Year dummies are included to account for year effects. GLS RE rather than Fixed-Effects (FEs) estimators are used for the following reasons. First, our aim is to capture how cross-country variation of regulation policies affects cross-country variation of bank board independence. Since regulation policies are in general relatively stable, fixed effects estimation will wipe out the effects which we are focusing on. Second, technically, we have only two groups of values (values for year 2004–2006 from survey II and values for year 2007–2010 from survey III) with low variation for each of our explanatory variables official supervision power and the private monitoring index and other regulation and supervision variables, which are not appropriate to be estimated with FE regression, as they would be absorbed in ‘time-demeaning’ or ‘within transformation’ processes of the variables in the FE model. Third, for large ‘N’ and fixed small ‘T’, which is the case for the panel data set (observations on 277 banks over 6 years) in this paper, FE estimation is inconsistent (Baltagi, 2005). Fourth, FE estimation would lead to a great loss of degrees of freedom in the case of a large ‘N’ (Baltagi, 2005). Finally, FE estimates could aggravate the problem of multi-collinearity if solved with least squares dummy variables (Baltagi, 2005). Thus, we use GLS RE rather than FE estimation here. Breusch and Pagan’s (1979) Lagrange multiplier tests support the random-effects specification, which rejects the null hypothesis that errors are independent within banks.

Country name

Country name

Jordan Korea Republic of Latvia Lebanon Lithuania Luxembourg Malaysia Malta Mauritius Mexico Netherlands Norway Oman Philippines Poland Romania Saudi Arabia Singapore Slovakia South Africa Spain Sweden

25 44 16 11 21 2 7 12 7 9 3 6 34 6 14 12 44 51 55 4 4 55 18 15 28 21 44 Obs

47 7 6 11 5 6 52 7 2 4 17 4 8 40 12 6 17 18 15 47 35 17

Ratio of independent directors in board

An independent director as board chairman

Ratio of independent directors in audit committee

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

30.44 82.14 36.78 25.05 29.05 33.33 78.53 49.07 30.59 14.38 38.43 25.00 57.52 47.85 46.88 25.97 42.86 24.56 35.16 61.67 57.78 48.25 51.69 66.07 75.58 65.46 22.85

5.83 9.37 20.89 12.81 14.02 0.00 26.33 15.62 7.81 9.32 2.72 4.30 18.76 18.86 15.69 20.38 20.05 14.17 11.63 3.33 2.57 9.93 12.72 9.80 15.79 23.18 18.52

22.22 55.56 11.76 12.00 11.11 33.33 21.05 18.18 21.43 0.00 35.29 22.22 11.11 37.50 20.00 0.00 14.29 0.00 15.79 60.00 55.56 33.33 30.00 45.45 40.00 28.57 0.00

40.00 91.67 68.42 44.44 55.56 33.33 94.12 66.67 41.18 27.27 40.00 33.33 75.00 85.71 62.50 60.00 100.00 66.67 56.25 66.67 60.00 70.00 75.00 83.33 93.33 100.00 58.33

0.00 0.98 0.31 0.27 0.14 0.50 0.86 0.00 0.00 0.00 0.00 0.00 0.82 0.17 0.29 0.00 0.41 0.00 0.00 1.00 0.00 0.11 0.17 0.80 0.21 0.19 0.00

0.00 0.15 0.48 0.47 0.36 0.71 0.38 0.00 0.00 0.00 0.00 0.00 0.39 0.41 0.47 0.00 0.50 0.00 0.00 0.00 0.00 0.31 0.38 0.41 0.42 0.40 0.00

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

0.00 1.00 1.00 1.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 1.00 0.00 1.00 0.00 0.00 1.00 0.00 1.00 1.00 1.00 1.00 1.00 0.00

89.33 90.81 37.40 50.17 43.49 0.00 100.00 83.33 42.04 10.19 66.67 38.89 33.58 83.33 63.10 43.45 39.19 50.95 83.69 91.67 66.67 43.46 82.96 94.22 64.29 80.27 60.72

15.87 17.02 37.84 12.51 36.52 0.00 0.00 30.15 28.75 15.47 11.55 13.61 43.64 40.82 25.26 28.20 36.83 29.36 15.05 16.67 0.00 22.76 26.59 12.31 48.80 24.23 27.06

66.67 30.77 0.00 33.33 0.00 0.00 100.00 0.00 0.00 0.00 60.00 33.33 0.00 0.00 25.00 0.00 0.00 0.00 60.00 66.67 66.67 16.67 0.00 66.67 0.00 0.00 0.00

100.00 100.00 100.00 66.67 100.00 0.00 100.00 100.00 60.00 33.33 80.00 66.67 100.00 100.00 100.00 75.00 100.00 100.00 100.00 100.00 66.67 100.00 100.00 100.00 100.00 100.00 100.00

Ratio of independent directors in board

Independent director as board chairman

Ratio of independent directors in audit committee

Mean

SD

Min.

Max.

Mean

SD

Min.

Max.

35.63 58.94 38.39 50.99 15.67 37.53 47.31 43.54 58.48 59.17 91.05 66.25 58.06 24.15 41.48 30.76 70.00 64.82 29.05 54.25 50.00 49.10

11.36 17.29 21.95 9.82 15.07 9.72 11.12 12.73 7.29 8.77 10.24 17.50 14.28 6.39 8.43 10.92 17.32 8.36 19.69 11.12 14.45 11.63

7.69 31.25 0.00 33.33 0.00 29.17 28.57 33.33 53.33 50.00 70.00 40.00 44.44 12.50 30.77 9.09 60.00 45.45 0.00 37.50 30.00 30.77

63.64 81.82 57.14 58.33 33.33 50.00 77.78 57.14 63.64 66.67 100.00 75.00 85.71 33.33 50.00 36.36 100.00 81.82 50.00 75.00 83.33 70.00

0.15 0.14 0.00 0.00 0.00 0.17 0.17 0.43 1.00 0.00 1.00 0.75 0.50 0.03 0.75 0.00 1.00 0.00 0.27 0.36 0.00 0.29

0.36 0.38 0.00 0.00 0.00 0.41 0.38 0.53 0.00 0.00 0.00 0.50 0.53 0.16 0.45 0.00 0.00 0.00 0.46 0.49 0.00 0.47

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00

1.00 1.00 0.00 0.00 0.00 1.00 1.00 1.00 1.00 0.00 1.00 1.00 1.00 1.00 1.00 0.00 1.00 0.00 1.00 1.00 0.00 1.00

Mean 33.30 40.00 0.00 60.61 10.00 66.67 79.49 81.90 0.00 25.00 92.84 25.00 48.13 49.09 39.58 33.33 88.24 89.63 18.57 66.65 88.67 78.43

SD 28.77 50.33 0.00 12.96 13.69 51.64 16.03 13.72 0.00 50.00 10.10 50.00 26.98 26.53 22.41 0.00 16.42 13.99 21.69 33.20 16.66 23.40

Min. 0.00 0.00 0.00 33.33 0.00 0.00 50.00 66.67 0.00 0.00 75.00 0.00 0.00 0.00 0.00 33.33 66.67 66.67 0.00 0.00 33.33 33.33

Max. 75.00 100.00 0.00 75.00 25.00 100.00 100.00 100.00 0.00 100.00 100.00 100.00 80.00 100.00 80.00 33.33 100.00 100.00 50.00 100.00 100.00 100.00

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Argentina Australia Austria Belgium Brazil Bulgaria Canada Chile China Croatia Cyprus Czech Republic Denmark Egypt Finland France Germany Greece HongKong Hungary Iceland India Indonesia Ireland Israel Italy Japan

Obs

2722

Table 3 Statistics on the ratio of independent directors on the board, independent director as board chairman and ratio of independent directors on the audit committee (per country, 2004–2010).

22.22 0.00 44.60 53.03 36.89 3.31 35.89 92.59 100.00 41.15 37.50 80.25 99.28 64.81 1.00 1.00 1.00 0.00 1.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00

This table gives the statistics of the ratio of independent directors on the board, independent director as the board chairman, and ratio of independent directors on the audit committee by country for a sample of 1134 bank-year observations. Detailed variable definitions and sources are given in Table 1.

100.00 100.00 100.00 75.00 100.00 100.00 100.00 0.53 0.38 0.30 0.00 0.35 0.32 0.43 0.44 0.17 0.09 0.00 0.14 0.12 0.24 92.31 60.00 42.86 33.33 76.92 93.33 100.00 50.00 23.53 0.00 33.33 27.27 66.67 0.00 16.15 9.71 13.85 0.00 13.02 7.91 22.78 9 46 32 2 51 43 1134 Switzerland Thailand Turkey United Arab Emirates United Kingdom United States of America Total

68.03 44.29 16.37 33.33 54.59 82.27 46.92

SD Max. Min. Max. Min. SD

Ratio of independent directors in board

Mean

Obs Country name

Table 3 (continued)

33.33 100.00 0.00 0.00 0.00 83.33 0.00

SD

Min.

Ratio of independent directors in audit committee

Mean

Independent director as board chairman

Mean

Max.

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

2723

4.2. Univariate analysis In the following section, we examine how regulation policies which empower official supervision and encourage private monitoring affect bank board independence. Before turning to the regression results, we first summarise the association between measures of board independence and indicators of the two types of bank regulations, using univariate analysis. The results are presented in Table 4. We divide the observations into banks with above-median and below-median values of official supervisory power /private monitoring index, and compare the mean value of board independence measures of low official supervisory power /private monitoring index group with that of high official supervisory power /private monitoring index group. We conduct the univariate analysis for observations in each year and for the total sample. For the total sample, low official supervisory power group has higher values of ratio of independent directors on the board (49.02), an independent director as the board chairman (0.32), and ratio of independent directors on the audit committee (66.91) than high official supervisory power group (with values of 45.39, 0.18, 63.27 respectively), and the differences are significant at the 1%, 1%, and 10% level respectively. The differences of mean values of board independence measures between low official supervisory power group and high official supervisory power group are mostly positive for year by year analysis; however the differences are sometimes not significant. On the contrary, low private monitoring index groups have lower values of board independence measures (with the values of 39.28, 0.15, and 54.47 for the total sample) than high private monitoring index groups (52.74, 0.31, and 72.67 respectively for the total sample), and the differences are significant for both the total sample and the year by year analysis. The results of univariate analysis suggest that official supervisory power is negatively related to board independence measures, while the private monitoring index is positively related to board independence measures. 4.3. Regression results 4.3.1. Bank regulation policies and board independence To assess exactly how direct official supervision and promoting private sector monitoring affect board independence, we regress the ratio of independent directors on the board on official supervisory power and private monitoring index when controlling for an array of bank-specific traits and country-specific characteristics. The results are presented in Table 5. In these regressions, we control for an array of bank-level traits (log(total assets), ln(board size), PB ratio, capital-asset ratio, loan-asset ratio, two-tier board and cross-listing), country-level variables (GDP per capita, GDP growth, inflation, board independence requirement, banking capital-asset ratio and banking non-performing loan) and year dummies. Unless otherwise explained, we control for these variables for all the regressions in this paper. Columns 1–3 of Table 5 present the panel data GLS estimators with bank random effects and robust standard errors allowing for country clusters, while columns 4–6 report the OLS estimators with robust standard errors allowing for country clusters. The results show that powerful official supervision reduces bank board independence. The coefficients of official supervisory power on ratio of independent directors on the board are negative and statistically significant at the 5% level in panel data model specifications and at 10% level in OLS model specifications, which supports our hypothesis 1. The results are consistent with Beck et al. (2006) in challenging the ‘supervisory power

2724

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Table 4 Bank regulation policies and board independence: univariate analysis. Ratio of independent directors on the board

Year 2004 Year 2005 Year 2006 Year 2007 Year 2008 Year 2009 Year 2010 Total

An independent director as board chairman

Ratio of independent directors on the audit committee

Low-High OSP

Low-High PM

Low-High OSP

Low-High PM

Low-High OSP

Low-High PM

21.07* [0.072] 2.09 [0.531] 0.07 [0.981] 2.64 [0.474] 3.05 [0.389] 3.40 [0.342] 9.14** [0.012] 3.63*** [0.009]

3.70 [0.793] 7.18** [0.030] 12.85*** [0.000] 15.24*** [0.000] 15.43*** [0.000] 14.51*** [0.000] 18.20*** [0.000] 13.47*** [0.000]

0.76*** [0.001] 0.11* [0.089] 0.06 [0.335] 0.11* [0.100] 0.15** [0.022] 0.16** [0.016] 0.21*** [0.002] 0.14*** [0.000]

0.27 [0.304] 0.02 [0.757] 0.18*** [0.002] 0.21*** [0.000] 0.23*** [0.000] 0.17*** [0.005] 0.24*** [0.000] 0.16*** [0.000]

3.33 [0.881] 4.23 [0.459] 10.52** [0.024] 1.78 [0.756] 0.47 [0.930] 1.74 [0.742] 4.36 [0.402] 3.64* [0.089]

51.90** [0.030] 23.11*** [0.000] 22.37*** [0.000] 14.769** [0.008] 13.64*** [0.009] 12.76** [0.013] 20.91*** [0.000] 18.2*** [0.000]

Obs. of Low/High OSP

Obs. of Low/High PM

7/10

7/10

84/91

88/87

91/116

100/107

63/115

71/107

76/116

78/114

75/113

75/113

82/95

71/106

478/656

490/644

This table presents the results of univariate analysis on bank board independence and regulation policy measures for each year and the total sample. For official supervisory power (OSP) and private monitoring index (PM), the differences of mean values of Ratio of independent directors on the board, an independent directors as the board chairman, and Ratio of independent directors on the audit committee between the groups with below-median regulation index values and above-median regulation index values are reported with their p-values. The table also provides the numbers of observations in low/high OSP groups and low/high PM groups for each year and the total sample. Detailed variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

view’.Regulations encouraging private monitoring increase board independence, as indicated by the positive and significant coefficients of the private monitoring index in all the model specifications at the 5% level. The positive coefficients suggest that the effects of reduced monitoring costs dominate the reduced benefits resulting from increased discipline for hypothesis 2. The finding that encouraging private monitoring promotes internal governance supports the ‘private empowerment view’ of Beck et al. (2006). We are examining the impacts of country-level regulation policies on individual banks and endogeneity problem is less of a concern, since it seems unlikely that an individual bank’s corporate governance structure will influence nationwide banking regulation policies. The opposite effects of official supervisory power and the private monitoring index on board independence provide a possible explanation for previous research results (e.g. Barth et al., 2004; Beck et al., 2006), which show that direct discipline and supervision from supervisory agencies is generally related to bad bank performance, while promoting private sectors to monitor banks improves bank performance. Our results suggest that the different impacts of the two types of regulation policies on bank performance may derive from their different implications for bank internal governance. The positive coefficients on log(total asset) indicate that the increased need for advice (Boone et al., 2007) dominates the increased monitoring costs (Linck et al., 2008) when bank size increases, which leads to a more independent board. However, this effect is significant only in pooled OLS regressions. Board size tends to be negatively related with board independence suggested by the negative and significant coefficients on ln(board size) in columns 1– 3. The negative and significant coefficients on PB ratio in columns 4–6 are consistent with the notion that growth firms displaying high information asymmetry are costly to monitor (Coles et al., 2007; Linck et al., 2008), however the effects disappear when using panel data regression. Banks whose shares are traded either by direct listing on a US stock exchange or as American Depository Receipts (ADRs) are expected to have a better governance structure, which is evidenced by the positive and significant coefficients on cross-listing in columns 1–3.

4.3.2. Controlling for more country-level variables Although we control for an array of bank-level and country-level factors in Table 5, there may be concerns that official supervision power and the private monitoring index are proxies for other country-specific traits. We therefore control for other country-specific factors that might explain, or subsume, the effects of official supervision power and the private monitoring index on ratio of independent directors on the board. As the proxy for Pillar 1 of Basel II, we add the capital regulatory index to assess the effects of the three Basel II Pillars simultaneously. We also include overall activities restrictiveness and entry into banking requirements to reflect the possible influence from bank operation restrictions and domestic and foreign bank entry restrictions. Researchers have shown that the country-level institutional environment, especially the quality of the legal system and the legal protection for shareholders, plays an important role in firm-level corporate governance (Kim et al., 2007; La Porta et al., 1998, 2000). We include rule of law and anti-director rights and their interaction to control for the legal environment of different countries. The regression results are presented in columns 1–6 in Table 6. Previous results on official supervisory power and the private monitoring index remain valid even when we control for these countryspecific factors. Thus, official supervisory power and the private monitoring index do not seem to be general proxies for country regulatory and institutional environments. The positive coefficients on anti-director rights, rule of law and rule of law⁄anti-director rights suggest that a good legal environment which strongly protects shareholder rights and a high quality legal system which enforces contracts effectively tend to induce shareholders to exert good internal governance, though the coefficients on rule of law are not significant. The results are consistent with Dahya et al. (2008) and Kim et al. (2007). The former find a positive relationship between the product of shareholder rights and rule of law with board independence in non-financial firms across 22 countries, while the latter demonstrate a positive relationship between shareholder rights and board independence in 14 European countries.

2725

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732 Table 5 Bank regulation policies and board independence. Panel regression (1) Official supervisory power

ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy Country cluster Number of banks Number of countries Observations R-squared Adjusted R-squared VIF Breusch–Pagan (v2) Breusch-Pagan (p-value)

(3)

(4) 3.413* [0.070]

0.871** [0.045] 1.119 [0.266] 6.761** [0.040] 0.002 [0.481] 0.038 [0.712] 0.019 [0.745] 2.797 [0.663] 1.669 [0.727] 4.383* [0.081] 0.192 [0.389] 0.110 [0.597] 0.069 [0.185] 0.955** [0.041] 0.692** [0.038] Yes Yes 277 55 1134 0.162 – 1.54 1340 [0.000]

1.576*** [0.008] 1.341*** [0.006] 1.126 [0.256] 6.801** [0.034] 0.003 [0.359] 0.048 [0.631] 0.020 [0.729] 3.159 [0.628] 2.503 [0.585] 4.298* [0.085] 0.172 [0.433] 0.137 [0.512] 0.073 [0.157] 0.986** [0.035] 0.616* [0.055] Yes Yes 277 55 1134 0.190 – 1.54 1318 [0.000]

1.058** [0.021]

Private monitoring Log(Total asset)

OLS regression (2)

1.337 [0.193] 6.978** [0.034] 0.002 [0.482] 0.043 [0.667] 0.020 [0.731] 3.319 [0.620] 2.361 [0.629] 4.348* [0.070] 0.176 [0.436] 0.119 [0.569] 0.082 [0.126] 0.846* [0.062] 0.615* [0.050] Yes Yes 277 55 1134 0.144 – 1.56 1498 [0.000]

4.075*** [0.002] 11.268 [0.127] 0.107*** [0.000] 0.328** [0.029] 0.133 [0.198] 3.878 [0.482] 6.598 [0.185] 1.147 [0.697] 0.181 [0.546] 0.700 [0.169] 0.205 [0.511] 0.603 [0.614] 0.805 [0.255] Yes Yes 277 55 1134 0.211 0.197 4.31 – –

(5)

(6)

5.130*** [0.000] 3.217** [0.012] 9.487 [0.185] 0.110*** [0.000] 0.212 [0.151] 0.118 [0.259] 2.615 [0.600] 4.212 [0.352] 1.124 [0.678] 0.083 [0.764] 0.678 [0.213] 0.252 [0.344] 0.767 [0.500] 1.215 [0.107] Yes Yes 277 55 1134 0.271 0.258 4.3 – –

3.251* [0.080] 5.075*** [0.000] 3.350*** [0.007] 10.098 [0.150] 0.099*** [0.000] 0.215 [0.153] 0.105 [0.297] 2.962 [0.554] 6.511 [0.137] 1.508 [0.582] 0.092 [0.737] 0.847 [0.126] 0.142 [0.613] 0.543 [0.614] 0.924 [0.201] Yes Yes 277 55 1134 0.287 0.274 4.17 – –

This table examines the impacts of the regulation policies empowering official supervision and those encouraging private monitoring on board independence. The dependent variable is the ratio of independent directors on the board. Columns 1–3 are estimated using panel data with GLS bank random-effects and robust standard errors allowing for clustering within countries. Columns 4–6 are estimated using OLS regression with robust standard errors allowing for clustering within countries. Variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

4.3.3. Controlling for more bank-level variables Studies of corporate boards declare some factors that affect private benefits of managers and monitoring costs to be determinants of board independence (Boone et al., 2007; Linck et al., 2008). Although we focus on the influence of country-level regulation policies in the shaping of bank-level governance, we do not want to ignore the potential importance of bank-specific factors. In addition to the bank-level control variables in Table 5, based on existing literature, we include market power, ln(no. of subsidiaries) and ownership concentration to capture some of these factors, with the results reported in columns 7, 9, 10, and 11 in Table 6. Furthermore, since banks differ from non-financial firms in many aspects, we also include loan loss provision ratio, liquidity ratio, and cost to income ratio in columns 8–11 in Table 6 to account for bankspecific characteristics. The results on official supervisory power and private monitoring index do not change when controlling for the factors which could determine board structure and the bank-specific factors associated with a bank’s capital adequacy, asset quality and liquidity. The results in Tables 5 and 6 suggest that the first type of regulations tends to crowd out the internal governance of banks, while the sec-

ond crowds in it. The negative coefficients on ownership concentration are consistent with Caprio et al. (2007) and Kim et al. (2007) which suggest a substitution between concentrated ownership and board independence, however, the coefficients are not significant in Table 6.

4.3.4. Bank regulation policies and board independence: the effects of rule of law and anti-director rights We include rule of law, anti-director rights, and rule of law⁄antidirector rights in Table 6 to control for the possible effects of country-level legal environment on board independence. In Table 7, we further explore the effects of legal environment by checking its impacts on the relationship between bank regulation policies and board independence. We divide our sample into groups with below median rule of law and above median rule of law and examine the relationship between bank regulation policies and board independence for the two subsamples, with the results reported in columns 1–2. We conduct similar regressions for anti-director rights and rule of lawanti-director rights and report the results in columns 3–4 and 5–6 respectively.

2726

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Table 6 Bank regulation policies and board independence: controlling for further country-level variables and bank-level variables. (1) Official supervisory power Private monitoring Capital regulatory index

(2) ***

1.849 [0.003] 1.459*** [0.003] 0.693** [0.029]

Overall activities restrictiveness Entry into banking requirements Rule of law

(3) ***

1.841 [0.001] 1.425*** [0.003] 0.509 [0.104] 0.754 [0.222] 0.469 [0.211]

(4) ***

2.065 [0.000] 1.457*** [0.010]

(5) ***

2.037 [0.000] 1.403*** [0.003]

5.221 [0.267] 3.483** [0.027]

Anti-director rights Rule of law*Anti-director rights Market power

(6) ***

2.425 [0.000] 1.598*** [0.003] 0.611* [0.062] 0.977 [0.142] 0.259 [0.468] 4.216 [0.398] 3.307** [0.019]

1.060*** [0.005]

(7) ***

2.407 [0.000] 1.550*** [0.000] 0.627** [0.047] 0.89 [0.128] 0.288 [0.430]

(8) ***

1.608 [0.007] 1.508*** [0.001]

0.121 [0.536] 0.157 [0.855] 5.513 [0.154]

Ownership concentration Loan loss provision ratio Liquidity ratio Cost to income ratio

Ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy Country cluster Number of countries Number of banks Observations R-squared VIF Breusch–Pagan (v2) Breusch–Pagan (p-value)

1.338 [0.039] 1.440*** [0.002]

1.253 [0.199] 7.038** [0.028] 0.004 [0.333] 0.064 [0.503] 0.033 [0.567] 3.181 [0.621] 2.329 [0.608] 4.411* [0.074] 0.192 [0.386] 0.151 [0.477] 0.060 [0.228] 1.001** [0.028] 0.500 [0.146] Yes Yes 55 277 1134 0.208 1.55 1344.96 [0.000]

1.231 [0.205] 6.696** [0.033] 0.004 [0.317] 0.058 [0.537] 0.023 [0.669] 2.892 [0.649] 2.069 [0.651] 4.444* [0.064] 0.229 [0.323] 0.146 [0.502] 0.068 [0.195] 1.027** [0.020] 0.551 [0.100] Yes Yes 55 277 1134 0.228 1.59 1330.06 [0.000]

1.447 [0.133] 9.159*** [0.009] 0.002 [0.538] 0.051 [0.641] 0.047 [0.485] 9.127 [0.283] 6.461 [0.260] 3.127 [0.211] 0.314 [0.515] 0.043 [0.729] 0.209*** [0.000] 0.873 [0.130] 0.156 [0.613] Yes Yes 36 236 1002 0.244 2.19 1091.68 [0.000]

1.613* [0.085] 9.146*** [0.009] 0.002 [0.451] 0.056 [0.610] 0.041 [0.543] 12.281 [0.121] 7.054 [0.187] 2.701 [0.266] 0.204 [0.422] 0.013 [0.912] 0.226*** [0.000] 1.058** [0.025] 0.187 [0.530] Yes Yes 36 236 1002 0.25 1.79 1270.69 [0.000]

(10) **

1.362 [0.037] 1.491*** [0.002]

(11) ***

2.150 [0.000] 1.791*** [0.001] 0.532 [0.165] 0.921 [0.238] 0.110 [0.747] 4.187 [0.395] 3.681** [0.014]

0.977*** [0.005]

ln(No. of subsidiaries)

Log(Total asset)

(9) **

1.701* [0.081] 9.046** [0.010] 0.003 [0.460] 0.068 [0.513] 0.049 [0.431] 8.888 [0.287] 5.390 [0.374] 3.250 [0.174] 0.160 [0.767] 0.044 [0.745] 0.214*** [0.001] 0.949* [0.087] 0.034 [0.916] Yes Yes 36 236 1002 0.281 2.26 1116.77 [0.000]

1.811** [0.040] 9.063*** [0.009] 0.003 [0.408] 0.072 [0.487] 0.045 [0.468] 11.381 [0.151] 6.343 [0.228] 2.884 [0.218] 0.103 [0.713] 0.025 [0.843] 0.226*** [0.000] 1.086*** [0.010] 0.056 [0.857] Yes Yes 36 236 1002 0.286 1.9 1307.91 [0.000]

1.092 [0.421] 7.546** [0.021] 0.003 [0.368] 0.061 [0.534] 0.026 [0.672] 2.214 [0.750] 2.305 [0.603] 3.615 [0.133] 0.163 [0.432] 0.121 [0.577] 0.073 [0.170] 1.038** [0.044] 0.587* [0.070] Yes Yes 55 273 1116 0.225 1.8 1239.56 [0.000]

0.004 [0.794] 0.007 [0.345] 0.010 [0.777] 1.171 [0.226] 7.312** [0.026] 0.006 [0.271] 0.218 [0.192] 0.023 [0.696] 3.810 [0.553] 2.637 [0.583] 3.808 [0.108] 0.224 [0.319] 0.156 [0.474] 0.085 [0.119] 0.929* [0.073] 0.540* [0.098] Yes Yes 54 262 1064 0.186 1.56 1254.26 [0.000]

0.055 [0.754] 0.669 [0.424] 5.230 [0.178] 0.013 [0.434] 0.006 [0.381] 0.005 [0.890] 1.669 [0.218] 8.250*** [0.010] 0.005 [0.388] 0.225 [0.164] 0.034 [0.582] 2.924 [0.674] 2.281 [0.632] 3.408 [0.143] 0.219 [0.310] 0.128 [0.569] 0.082 [0.139] 0.986* [0.076] 0.448 [0.163] Yes Yes 54 258 1046 0.206 1.86 1164.61 [0.000]

0.253 [0.337] 0.515 [0.590] 5.256 [0.212] 0.008 [0.620] 0.006 [0.411] 0.007 [0.828] 1.295 [0.414] 9.268*** [0.008] 0.004 [0.395] 0.239 [0.110] 0.051 [0.424] 9.534 [0.200] 6.246 [0.314] 1.778 [0.438] 0.080 [0.880] 0.069 [0.627] 0.228*** [0.001] 0.885 [0.176] 0.267 [0.401] Yes Yes 37 221 922 0.305 2.43 1026.27 [0.000]

2.173*** [0.000] 1.782*** [0.000] 0.541 [0.153] 0.846 [0.237] 0.142 [0.681]

1.004*** [0.008] 0.272 [0.286] 0.515 [0.600] 5.193 [0.211] 0.008 [0.626] 0.005 [0.486] 0.006 [0.843] 1.332 [0.353] 9.235*** [0.007] 0.005 [0.357] 0.23 [0.121] 0.046 [0.476] 11.622 [0.113] 7.013 [0.211] 1.526 [0.502] 0.03 [0.912] 0.053 [0.692] 0.240*** [0.000] 1.063* [0.052] 0.235 [0.447] Yes Yes 37 221 922 0.312 2.17 1143.28 [0.000]

This table examines the impacts of regulation policies on board independence when controlling for more country-level and bank-level characteristics. The dependent variable is the ratio of independent directors on the board. All the regressions are estimated using panel data with GLS bank random-effects and robust standard errors allowing for clustering within countries. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

Previous results on official supervisory power and the private monitoring index still hold for all the subsamples. Compare the coefficients of official supervisory power and private monitoring in-

dex between low and high rule of law groups in columns 1–2, we find that a higher quality legal system which enforces contracts more effectively enhances the effects of regulation policies on

2727

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732 Table 7 Bank regulation policies and board independence: the effects of rule of law and anti-director rights.

Official supervisory power Private monitoring Log(Total asset) ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy Country cluster Number of banks Number of countries Observations R-squared VIF Breusch–Pagan (v2) Breusch–Pagan (p-value)

Low Rule of law (1)

High Rule of law (2)

Low Anti-director rights (3)

High Anti-director rights (4)

Low Rule of law *Antidirector rights (5)

High Rule of law *Antidirector rights (6)

1.500** [0.011] 1.096* [0.070] 0.469 [0.723] 11.022*** [0.007] 0.003 [0.426] 0.057 [0.552] 0.059 [0.484] 3.435 [0.406] 12.146** [0.017] 0.096 [0.982] 0.173 [0.783] 0.127 [0.379] 0.071 [0.232] 0.617 [0.175] 0.322 [0.173] Yes Yes 132 25 564 0.195 1.58 569.03 [0.000]

3.264** [0.039] 2.619** [0.010] 3.011*** [0.008] 6.400 [0.189] 1.338 [0.192] 0.336** [0.040] 0.114 [0.247] 12.381 [0.105] 2.349 [0.751] 5.311 [0.112] 0.074 [0.884] 0.203 [0.262] 0.075 [0.765] 0.619 [0.521] 0.212 [0.805] Yes Yes 136 25 542 0.253 1.7 413.29 [0.000]

3.519*** [0.001] 3.009*** [0.004] 2.308** [0.011] 12.585*** [0.000] 0.004 [0.418] 0.051 [0.663] 0.043 [0.771] 16.297** [0.035] 3.898 [0.490] 1.979 [0.527] 0.062 [0.827] 0.238 [0.458] 0.185 [0.239] 0.410 [0.622] 0.401 [0.290] Yes Yes 97 15 363 0.358 2.61 257.89 [0.000]

1.606 [0.151] 2.637** [0.016] 1.337 [0.254] 6.644 [0.222] 0.193 [0.189] 0.084 [0.540] 0.100* [0.092] 17.633*** [0.003] 9.551 [0.151] 9.863*** [0.005] 0.079 [0.822] 0.052 [0.721] 0.224** [0.019] 0.846 [0.257] 0.123 [0.774] Yes Yes 139 21 639 0.274 1.93 799.63 [0.000]

2.947*** [0.000] 2.100*** [0.008] 1.485 [0.170] 12.244*** [0.000] 0.004 [0.295] 0.014 [0.911] 0.028 [0.842] 12.175 [0.168] 3.512 [0.560] 2.661 [0.328] 0.094 [0.695] 0.071 [0.687] 0.213* [0.066] 1.031** [0.040] 0.566* [0.065] Yes Yes 127 19 507 0.298 2.03 409.18 [0.000]

2.712*** [0.009] 3.747*** [0.000] 2.099* [0.075] 6.142 [0.298] 0.496*** [0.007] 0.125 [0.389] 0.115* [0.056] 27.132*** [0.000] 4.003 [0.528] 11.031*** [0.002] 0.375 [0.367] 0.062 [0.790] 0.081 [0.534] 0.459 [0.628] 0.226 [0.605] Yes Yes 109 17 495 0.271 2.02 586.45 [0.000]

This table reports the robustness test results of the regulation policies empowering official supervision and those empowering private monitoring on board independence, by checking the effects of rule of law and anti-director rights. The dependent variable is the ratio of independent directors on the board. All the regressions are estimated using panel data with GLS bank random-effects and robust standard errors allowing for clustering within countries. We split our sample into groups with low rule of law and high rule of law in columns 1–2, groups with low anti-director rights and high anti-director rights in columns 3–4, and groups with low rule of law*anti-director rights and high rule of law*anti-director rights in columns 5–6. Variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

board independence, represented by the increased coefficient on private monitoring index and the decreased coefficient on official supervisory power from low to high rule of law group. However, the coefficients in columns 3–4 demonstrate that a better legal environment which better protects shareholder rights attenuate the effects of regulation policies on board independence, represented by the decreased coefficient on private monitoring index and the increased coefficient on official supervisory power from low to high anti-director rights group. The coefficients of 5–6 show that a legal system with high rule of law⁄anti-director rights which protects shareholder rights strongly as well as enforces contracts effectively will enhance the positive effects of empowering private monitoring and slightly attenuate the negative effects of empowering direct official supervision on board independence. Results in Table 7 suggest that better investor rights protection decreases the crowding out effect of direct official supervision on internal governance but also decreases the crowding in effect of promoting private monitoring on internal governance. On the contrary, better contracts enforcement increases the crowding in effect of promoting private monitoring on internal governance but

also increases the crowding out effect of direct official supervision on internal governance. Columns 5–6 in Table 7 and columns 4, 6, and 11 in Table 6 together suggest the legal system with better investor rights protection as well as better contracts enforcement will not only induce internal governance itself but also increase the crowding in effect of promoting private monitoring and decrease the crowding out effect of direct official supervision on internal governance.

5. Additional robustness tests 5.1. Redefining variables In this subsection, we examine the sensitivity of our results to alteration in the definition of some key variables. First, we examine the relation between the two types of regulation policies and board independence by using an independent director as the board chairman and ratio of independent directors on the audit committee to replace ratio of independent directors on the board as measures of

2728

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Table 8 Redefining variables: alternative variables for board independence. An independent director as board chairman

Official supervisory power Private monitoring Log(Total asset) Ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy/country cluster No. of banks/no. of countries Observations R-squared/pseudo R2/adjusted R2 Log likelihood/log pseudo likelihood VIF Breusch–Pagan (v2) Breusch–Pagan (p-value) LR test of rho = 0 (Chibar2) LR test of rho = 0 (p-value)

Ratio of independent directors on the audit committee

Panel probit (1)

Probit (2)

Panel logit (3)

Logit (4)

Panel (5)

OLS (6)

0.018 [0.921] 0.485** [0.020] 0.394** [0.031] 1.060* [0.096] 0.011 [0.381] 0.025 [0.419] 0.037*** [0.009] 1.157 [0.162] 0.803 [0.264] 0.302 [0.554] 0.024 [0.269] 0.024 [0.615] 0.002 [0.911] 0.315** [0.011] 0.129* [0.073] Yes/ No 277/55 1134 –/–/– –/–296.105 1.54 – – 471.26 [0.000]

0.152* [0.092] 0.187*** [0.004] 0.019 [0.779] 0.464 [0.142] 0.002 [0.514] 0.002 [0.886] 0.008 [0.231] 0.415 [0.112] 0.067 [0.817] 0.041 [0.838] 0.012 [0.245] 0.033 [0.309] 0.005 [0.736] 0.093 [0.107] 0.086** [0.033] Yes/ Yes 277/55 1134 –/0.140/– –543.319/– 1.54 – – – –

0.012 [0.971] 0.977*** [0.001] 0.777** [0.039] 1.907 [0.136] 0.018 [0.360] 0.060 [0.394] 0.079*** [0.002] 2.448** [0.045] 2.069 [0.101] 0.375 [0.719] 0.054 [0.222] 0.044 [0.625] 0.005 [0.907] 0.634*** [0.003] 0.237* [0.064] Yes/ No 277/55 1134 –/–/– –/–299.59 1.54 – – 464.69 [0.000]

0.269* [0.100] 0.347*** [0.004] 0.015 [0.894] 0.781 [0.163] 0.004 [0.396] 0.000 [0.988] 0.015 [0.198] 0.756* [0.095] 0.122 [0.823] 0.050 [0.884] 0.023 [0.248] 0.065 [0.271] 0.012 [0.694] 0.166 [0.134] 0.160** [0.038] Yes/ Yes 277/55 1134 –/0.140/– –543.28/– 1.54 – – – –

2.299 [0.224] 2.625** [0.033] 2.000 [0.208] 2.665 [0.621] 0.005 [0.434] 0.172 [0.134] 0.024 [0.801] 8.088 [0.281] 7.903 [0.249] 8.145 [0.117] 0.140 [0.625] 0.005 [0.984] 0.085 [0.515] 0.298 [0.721] 0.216 [0.645] Yes/ Yes 277/55 1134 0.149/–/– –/– 1.54 965.71 [0.000] – –

2.606 [0.366] 6.291*** [0.001] 2.584 [0.268] 9.082 [0.267] 0.180*** [0.000] 0.251 [0.455] 0.099 [0.413] 9.922 [0.160] 4.937 [0.425] 4.947 [0.425] 0.263 [0.290] 0.512 [0.491] 0.040 [0.936] 1.313 [0.188] 0.689 [0.481] Yes/ Yes 277/55 1134 0.189/–/0.174 –/– 1.54 – – – –

This table presents the robustness regression results using alternative variables for dependent variable. In columns 1–4, we use an independent director as board chairman as the alternative measure of board independence. Columns 1 and 3 are estimated using panel data probit and logit model with bank random effects respectively. Column 2 and 4 are estimated using probit and logit model respectively with robust standard errors allowing for clustering within countries. In columns 5 and 6, we use ratio of independent directors on the audit committee as the alternative measure of board independence. Column 5 is estimated using panel data with GLS bank random-effects and robust standard errors allowing for clustering within countries. Column 6 is estimated using OLS regression with robust standard errors allowing for clustering within countries. Variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

board independence. The regression results are shown in Table 8. For an independent director as the board chairman, we provide the panel probit and panel logit model with random effects regressions in columns 1 and 2, and pooled probit and logit model with country clustering in columns 3 and 4. For ratio of independent directors on the audit committee, we provide the panel regression with random effects allowing country clustering in column 5 and pooled OLS regression with country clustering in column 6. The results on the private monitoring index are similar to earlier findings. The results on official supervisory power are similar to earlier results in the pooled probit and logit models (columns 2 and 4) using an independent director as the board chairman as dependent variable. When using ratio of independent directors on the audit committee as dependent variable, the coefficients on official supervisory power and private monitoring index are not significant although the signs are consistent with previous results.

Second, we alter explanatory variables, using prompt corrective power and external ratings and credit monitoring to replace official supervisory power and the private monitoring index respectively, and examine the effects of the two bank regulation policies on ratio of independent directors on the board, an independent director as the board chairman, and ratio of independent directors on the audit committee. The regression results are presented in Table 9. The results are generally consistent with previous findings except that the coefficients of prompt corrective power on ratio of independent directors on the audit committee in columns 7 and 9 are positive but insignificant. Third, we transform our main explanatory variables (official supervisory power and the private monitoring index) into dummies to conduct the robustness check. As Table 10 shows, we first divide the sample into two groups with above-median and below-median values of official supervisory power and private monitoring index respectively, and generate a dummy variable for each of the two

2729

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732 Table 9 Redefining variables: alternative variables for official supervisory power and private monitoring index.

Prompt corrective power

Ratio of independent directors on the board

An independent director as board chairman

Ratio of independent directors on the audit committee

(1)

(6)

(7)

0.268* [0.090] 0.225 [0.260] 0.004 [0.951] 0.561* [0.052] 0.002 [0.539] 0.001 [0.924] 0.011* [0.083] 0.208 [0.393] 0.086 [0.768] 0.015 [0.945] 0.011 [0.315] 0.031 [0.312] 0.003 [0.825] 0.106* [0.075] 0.084* [0.054] Yes Yes 277 55 1134 – 0.124 553.63 1.55 –

1.162 [0.704]

ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy Country cluster Number of banks Number of countries Observations R-squared Pseudo R-squared Log pseudo likelihood VIF Breusch–Pagan (v2)

(3)

(4) 0.275* [0.071]

1.837 [0.052] 1.170 [0.257] 6.780** [0.043] 0.002 [0.453] 0.030 [0.769] 0.013 [0.821] 2.685 [0.685] 2.183 [0.655] 4.410* [0.066] 0.181 [0.432] 0.084 [0.685] 0.078 [0.138] 0.889** [0.047] 0.703** [0.030] Yes Yes 277 55 1134 0.141 – – 1.56 1526.22

1.517 [0.108] 1.493* [0.092] 1.250 [0.216] 6.995** [0.034] 0.002 [0.472] 0.030 [0.768] 0.016 [0.782] 2.486 [0.705] 2.628 [0.592] 4.390* [0.058] 0.175 [0.448] 0.097 [0.641] 0.078 [0.143] 0.823* [0.071] 0.685** [0.034] Yes Yes 277 55 1134 0.150 – – 1.55 1512.93

*

External ratings and credit monitoring Log(Total asset)

(2)

1.812* [0.067]

1.334 [0.183] 7.103** [0.030] 0.002 [0.535] 0.036 [0.735] 0.020 [0.726] 2.781 [0.671] 2.371 [0.634] 4.374* [0.061] 0.181 [0.426] 0.113 [0.583] 0.077 [0.150] 0.785* [0.085] 0.657** [0.040] Yes Yes 277 55 1134 0.144 – – 1.55 1534.27

(5)

0.015 [0.830] 0.570** [0.049] 0.001 [0.653] 0.003 [0.799] 0.011* [0.087] 0.282 [0.229] 0.045 [0.882] 0.042 [0.840] 0.009 [0.366] 0.030 [0.335] 0.005 [0.693] 0.096 [0.118] 0.085* [0.063] Yes Yes 277 55 1134 – 0.118 557.79 1.55 ––

0.237 [0.233] 0.011 [0.871] 0.459 [0.135] 0.002 [0.604] 0.000 [0.969] 0.009 [0.143] 0.287 [0.295] 0.005 [0.987] 0.021 [0.920] 0.012 [0.325] 0.048 [0.155] 0.009 [0.533] 0.116** [0.040] 0.097** [0.028] Yes Yes 277 55 1134 – 0.11 562.77 1.56

(8)

(9)

*

2.315 [0.152] 2.277 [0.686] 0.008 [0.211] 0.170 [0.100] 0.017 [0.867] 8.473 [0.266] 9.374 [0.193] 8.033 [0.151] 0.174 [0.539] 0.076 [0.760] 0.093 [0.427] 0.278 [0.759] 0.109 [0.809] Yes Yes 277 55 1134 0.115 – – 1.55 1090.61

4.813 [0.061] 2.108 [0.175] 2.319 [0.668] 0.006 [0.363] 0.144 [0.194] 0.026 [0.801] 9.884 [0.170] 7.913 [0.265] 7.786 [0.167] 0.141 [0.620] 0.109 [0.666] 0.105 [0.418] 0.329 [0.690] 0.058 [0.892] Yes Yes 277 55 1134 0.130 – – 1.56 1031.96

2.095 [0.422] 5.278** [0.033] 2.017 [0.200] 2.688 [0.621] 0.006 [0.330] 0.144 [0.194] 0.030 [0.771] 9.639 [0.180] 8.519 [0.241] 7.736 [0.172] 0.146 [0.596] 0.134 [0.591] 0.102 [0.424] 0.438 [0.594] 0.023 [0.956] Yes Yes 277 55 1134 0.133 – – 1.55 1028.31

This table presents the robustness regression results using alternative variables for explanatory variables. We use prompt corrective power and external ratings and credit monitoring replacing official supervisory power and private monitoring index respectively to measure the two types of regulation policies. The dependent variables in columns 1–3, 4–6, 7–9 are ratio of independent director on the board, an independent director as board chairman, and ratio of independent directors on the audit committee respectively. Columns 1–3 and 7–9 are estimated using panel data with GLS bank random-effects and robust standard errors allowing for clustering within countries. Columns 4–6 are estimated using probit model with robust standard errors allowing for clustering within countries. Variable definitions and sources are given in Table 1.  Significance at the 1%. ** Significance at the 5%. * Significance at the 10%.

indices. For above-median value group, the dummy variable has the value of 1, and for below-median value group, the dummy variable has the value of 0. We regress the three board independence measures on the two dummies instead of the original variables. The regression results are in rows 1, 4, and 7 of Table 10. Using the same method, we transform the two indices into dummies in thirds and include the upper third as dummy and show the regression results in rows 2, 5, and 8. We then include the lower third as dummy and show the results in rows 3, 6, and 9. The results are consistent with previous findings. Higher official supervisory power is related to lower board independence, but a higher private monitoring index is related to higher board independence. The results support our previous findings on the relation between the two types of regulation policy and board independence. 5.2. Altering sample We further examine the sensitivity of our results by altering sample, investigating the impacts of official supervisory power and

private monitoring on ratio of independent directors on the board year by year. Table 11 reports the year by year regression results. All the regressions are estimated using OLS estimators with robust standard errors. Except for the coefficients of official supervisory power in year 2005 and year 2006, the results are consistent with our previous findings that official supervisory power decreases the ratio of independent directors on the board while private monitoring increases board independence ratio. Many countries set up ‘corporate governance guidelines and codes of best practice’ for corporations. There are cross-country differences in the requirements for board composition in these guidelines and codes. In previous regressions, we include a dummy variable (board independence requirement), which indicates whether a country has such requirements, to control for the possible effects of corporate governance guidelines. We also conduct robustness tests by splitting our sample into countries that have no requirements/recommendations on the minimum number or fraction of independent directors and those do have such requirements/recommendations. The results on official supervisory power

2730

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Table 10 Redefining variables: transforming official supervisory power and private monitoring index into dummies.

OSP High dummy PM High dummy

Ratio of independent directors on the board

An independent director as board chairman

Ratio of independent directors on the audit committee

(1)

(4)

(7)

(2)

PM Upper third dummy OSP Lower third dummy PM Lower third dummy

ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Year dummy Country cluster Number of banks Number of countries Observations R-squared/pseudo R-squared Log pseudo likelihood VIF

(5)

3.326*** [0.000] 8.255*** [0.001] 0.097*** [0.000] 0.244*** [0.000] 0.119*** [0.002] 3.239* [0.069] 6.280*** [0.000] 1.143 [0.470] 0.138* [0.066] 0.789*** [0.001] 0.255* [0.095] 0.471 [0.113] 1.163*** [0.000] Yes No 277 55 1134 0.240 – 1.56

(6)

0.368*** [0.000] 0.535*** [0.000] 7.934*** [0.000] 15.665*** [0.000]

OSP Upper third dummy

Log(Total asset)

(3)

4.114*** [0.002] 10.694*** [0.000]

3.682*** [0.000] 14.209*** [0.000] 0.095*** [0.000] 0.211*** [0.002] 0.098*** [0.006] 2.363 [0.177] 4.368*** [0.002] 2.380 [0.111] 0.079 [0.266] 0.720*** [0.004] 0.311** [0.040] 1.006*** [0.000] 0.689** [0.033] Yes No 277 55 1134 0.318 – 1.52

0.015 [0.674] 0.397*** [0.009] 0.003 [0.346] 0.000 [0.931] 0.009*** [0.001] 0.441*** [0.000] 0.137 [0.242] 0.029 [0.797] 0.011* [0.080] 0.028 [0.155] 0.007 [0.619] 0.088*** [0.000] 0.095*** [0.000] Yes No 277 55 1134 0.1415 533.64 1.56

(9)

-8.539*** [0.001] 10.232*** [0.000]

0.012 [0.922] 0.624*** [0.000] 3.490*** [0.009] 9.582*** [0.000] 3.468*** [0.000] 9.177*** [0.000] 0.117*** [0.000] 0.270*** [0.000] 0.127*** [0.001] 3.739** [0.034] 5.590*** [0.000] 0.648 [0.687] 0.160** [0.038] 0.741*** [0.003] 0.241 [0.141] 0.772*** [0.010] 1.009*** [0.000] Yes No 277 55 1134 0.221 – 1.55

(8)

3.693* [0.099] 16.862*** [0.000]

0.007 [0.837] 0.657*** [0.000] 0.002 [0.542] 0.001 [0.852] 0.010*** [0.000] 0.446*** [0.000] 0.060 [0.592] 0.060 [0.596] 0.012** [0.041] 0.036* [0.057] 0.013 [0.392] 0.123*** [0.000] 0.091*** [0.000] Yes No 277 55 1134 0.1422 534.67 1.52

0.332*** [0.001] 0.382*** [0.002] 0.004 [0.905] 0.447*** [0.002] 0.002 [0.588] 0.002 [0.608] 0.010*** [0.000] 0.402*** [0.001] 0.105 [0.366] 0.016 [0.885] 0.009 [0.131] 0.032* [0.091] 0.008 [0.611] 0.098*** [0.000] 0.084*** [0.001] Yes No 277 55 1134 0.1256 543.31 1.55

2.345*** [0.006] 11.901*** [0.001] 0.168*** [0.000] 0.254* [0.098] 0.103* [0.098] 10.203*** [0.001] 4.668* [0.057] 4.710* [0.068] 0.303*** [0.006] 0.520 [0.194] 0.072 [0.771] 1.505*** [0.000] 0.416 [0.417] Yes No 277 55 1134 0.182 – 1.56

3.214*** [0.000] 5.304 [0.139] 0.179*** [0.000] 0.304* [0.058] 0.102 [0.100] 9.914*** [0.002] 6.220*** [0.009] 5.395** [0.035] 0.305*** [0.006] 0.329 [0.411] 0.122 [0.637] 0.972** [0.027] 0.975* [0.059] Yes No 277 55 1134 0.166 – 1.52

3.329 [0.127] 19.883*** [0.000] 2.310*** [0.005] 11.171*** [0.002] 0.200*** [0.000] 0.262* [0.077] 0.105* [0.087] 11.053*** [0.000] 5.519** [0.020] 3.848 [0.134] 0.327*** [0.003] 0.520 [0.197] 0.029 [0.906] 1.014** [0.015] 0.675 [0.162] Yes No 277 55 1134 0.182 – 1.55

This table reports the robustness test results of the regressions using transformed dummies of official supervisory power and private monitoring index. Columns 1, 4, and 7 employ official supervisory power high dummy and private monitoring high dummy as explanatory variables. Columns 2, 5, and 8 employ official supervisory power upper third dummy and private monitoring upper third dummy as explanatory variables. Columns 3, 6, and 9 employ official supervisory power lower third dummy and private monitoring lower third dummy as explanatory variables. Columns 1–3 and 7–9 are estimated using OLS regression with robust standard errors. Columns 4–6 are estimated using probit regression with robust standard errors. We also estimate all the regressions with country clustering in unreported table, with the results of private monitoring index unchanged and some of the results of official supervisory power weaken. Variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

and the private monitoring index still hold in the two subsamples. We have not reported the results here for reasons of brevity, but they are available from the authors. 6. Conclusions This paper examines how regulation policies which focus on empowering official supervision and those which focus on encouraging private monitoring affect the board independence of banks. We use hand-collected board data of 277 banks in 55 countries over the period 2004–2010 and the World Bank cross-country surveys II and III on bank regulation and supervision.

We reach two main conclusions. First, empowering official supervision reduces board independence. Second, encouraging private monitoring increases board independence. The two conclusions provide a possible explanation for previous research results (e.g. Barth et al., 2004; Beck et al., 2006), which show that direct discipline and supervision from supervisory agencies is generally related to bad bank performance, while promoting private sectors to monitor banks improves bank performance. Beyond these two important results, we find that the legal system with better investor rights protection as well as better contracts enforcement not only increases board independence itself but also enhances the crowding in effect of promoting private

2731

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732 Table 11 Altering sample: bank regulation policies and board independence (year by year).

Official supervisory power Private monitoring Log(Total asset) Ln(Board size) PB ratio Capital-asset ratio Loans-asset ratio Two-tier board Board independence requirement Cross-listing GDP per capita GDP growth Inflation Banking capital-asset ratio Banking non-performing loan Number of banks Number of countries Observations R-squared VIF

Year 2005 (1)

Year 2006 (2)

Year 2007 (3)

Year 2008 (4)

Year 2009 (5)

Year 2010 (6)

1.574 [0.265] 4.716*** [0.000] 3.039** [0.026] 10.970* [0.084] 2.190 [0.125] 0.127 [0.369] 0.120 [0.205] 1.797 [0.699] 3.390 [0.428] 0.014 [0.997] 0.096 [0.662] 0.352 [0.640] 0.617 [0.316] 0.909 [0.278] 1.036 [0.135] 175 42 175 0.287 2.05

0.873 [0.445] 5.628*** [0.000] 1.921 [0.119] 7.829 [0.184] 0.482 [0.307] 0.122 [0.296] 0.157** [0.037] 11.266*** [0.004] 10.664*** [0.003] 4.316 [0.240] 0.140 [0.522] 1.878** [0.014] 1.624*** [0.004] 1.474** [0.032] 0.283 [0.705] 207 41 207 0.360 1.89

5.610*** [0.003] 6.031*** [0.000] 3.832*** [0.004] 10.504* [0.083] 0.116*** [0.000] 0.256 [0.174] 0.158* [0.087] 5.951 [0.155] 11.724*** [0.005] 3.254 [0.429] 0.075 [0.668] 2.031** [0.015] 0.392 [0.520] 0.183 [0.836] 0.271 [0.660] 178 41 178 0.318 1.88

4.454** [0.022] 7.878*** [0.000] 4.214*** [0.001] 6.768 [0.258] 0.680 [0.640] 0.385* [0.079] 0.123 [0.144] 4.270 [0.316] 10.139*** [0.005] 0.791 [0.819] 0.318* [0.054] 4.243*** [0.000] 0.656** [0.016] 0.198 [0.792] 1.179 [0.211] 192 43 192 0.414 1.79

3.594** [0.032] 6.112*** [0.000] 3.818*** [0.001] 5.263 [0.356] 2.425 [0.195] 0.416** [0.034] 0.081 [0.352] 7.276 [0.115] 3.944 [0.288] 1.923 [0.585] 0.391** [0.027] 2.452*** [0.000] 0.723*** [0.006] 0.347 [0.590] 0.017 [0.980] 188 44 188 0.398 1.63

3.233* [0.078] 3.969*** [0.001] 5.082*** [0.000] 10.106* [0.061] 1.765 [0.225] 0.185 [0.411] 0.053 [0.656] 0.562 [0.915] 0.615 [0.865] 0.363 [0.927] 0.054 [0.795] 1.681** [0.027] 0.875* [0.066] 0.379 [0.604] 2.128*** [0.007] 177 41 177 0.409 1.89

This table reports the robustness test results of the regulation policies empowering official supervision and those empowering private monitoring on board independence, by examining the impacts of official supervisory power and private monitoring on the ratio of independent directors on the board year by year. The dependent variable is the ratio of independent directors on the board. All the regressions are estimated using OLS estimators with robust standard errors. We also estimate all the regressions with country clustering in unreported table, with the results of private monitoring index unchanged and several of the results of official supervisory power weaken. Variable definitions and sources are given in Table 1. * Significance at the 10%. ** Significance at the 5%. *** Significance at the 1%.

monitoring and decreases the crowding out effect of direct official supervision on board independence, which is consistent with Kim et al. (2007) and La Porta et al. (1998, 2000) that the country-level legal environment, especially the quality of contracts enforcement and the protection for shareholders, plays an important role in firm-level corporate governance. Our results have policy implications. After the 2007 financial crises, many policy makers and economists proposed more stringent capital requirements, more powerful official supervision and more transparency, reflecting the ideas of the three Pillars of Basel II. Our results tentatively support more transparency, since it induces the internal governance of banks. Our results challenge more powerful official supervision since it reduces the monitoring exerted by shareholders. Our paper calls for more attention to the role of corporate governance when formulating regulation policies or overhauling the regulatory framework of banks. Laeven and Levine (2009) emphasise the role of corporate governance by showing that the same regulation policy can have different effects on bank risk-taking according to a bank’s corporate governance structure. Our results further suggest that different regulation policies can affect bank corporate governance structure very differently. Policy makers should take into account micro-level responses when formulating regulation policies.

Acknowledgements This paper is supported by NSFC under Grant Nos. 71003021 and 71203246. For their helpful comments and suggestions, we should like to thank an anonymous reviewer, Chen Lin, Xianming Zhou, Huiyan Qiu, Joseph P.H. Fan, Xinzhong Xu, Luc Laeven, and seminar participants in the 2009 Journal of Corporate Finance Special Conference on Emerging Markets, the economics and finance workshops at the University of Hong Kong and the University and International Business and Economics. References Adams, R.B., Mehran, H., 2012. Bank board structure and performance: Evidence for large bank holding companies. Journal of Financial Intermediation 21, 243– 267. Andres, P., Vallelado, E., 2008. Corporate governance in banking: the role of the board of directors. Journal of Banking and Finance 32, 2570–2580. Andres, P., Romero-Merino, M.E., Santamaría, M., Vallelado, E., 2012. Boards determinants in banking industry. An international perspective. Managerial and Decision Economics 33 (3), 147–158. Baltagi, B.H., 2005. In: Econometric Analysis of Panel Data, 3rd ed. Wiley, New York, Chichester, Toronto and Brisbane. Barth, J.R., Caprio, G., Levine, R., 2004. Bank supervision and regulation: what works best? Journal of Financial Intermediation 2, 205–248. Barth, J.R., Caprio, G., Levine, R., 2006. Rethinking Bank Supervision and Regulation: Until Angels Govern. Cambridge University Press, Cambridge, UK.

2732

L. Li, F.M. Song / Journal of Banking & Finance 37 (2013) 2714–2732

Barth, J.R., Hai, J., Hartarska, V., Phumiwasana, T., 2007. A cross-country analysis of bank performance: the role of external governance. In: Gup, B. (Ed.), Corporate Governance in Banking: An International Perspective. Edward Elgar Publishers, Northampton, England. Becher, D.A., Campbell, T.L., Frye, M.B., 2005. Incentive compensation for bank directors: the impact of deregulation. Journal of Business 78, 1753–1777. Beck, T., Demirguc-Kunt, A., Levine, R., 2006. Bank supervision and corruption in lending. Journal of Monetary Economics 53, 2131–2163. Belkhir, M., 2009a. Board of directors’ size and performance in the banking industry. International Journal of Managerial Finance 5 (2), 201–221. Belkhir, M., 2009b. Ownership structure, board structure and firm performance: evidence from banking. Applied Financial Economics 19 (19), 1581–1593. Billett, M.T., Garfinkel, J.A., O’Neal, E.S., 1998. The cost of market versus regulatory discipline in banking. Journal of Financial Economic 3, 333–358. BIS, 2004. International Convergence of Capital Measurement and Capital Standards: A Revised Framework. Discussion Paper. Basel Committee on Banking Supervision, Basel. Boone, A., Field, L., Karpoff, J., Raheja, C., 2007. The determinants of corporate board size and composition: an empirical analysis. Journal of Financial Economics 85, 66–101. Booth, J.R., Cornett, M.M., Tehranian, T., 2002. Boards of directors, ownership, and regulation. Journal of Banking and Finance 26, 1973–1996. Brickley, J.A., James, C.M., 1987. The takeover market, corporate board composition, and ownership structure: the case of banking. Journal of Law and Economics 30 (1), 161–180. Breusch, T.S., Pagan, A.R., 1979. A simple test for heteroscedasticity and random coefficient variation. Econometrica 47, 1287–1294. Bushman, R., Chen, Q., Engel, E., Smith, A., 2004. Financial accounting information, organization complexity, and corporate governance. Journal of Accounting and Economics 37, 167–201. Caprio, G., Laeven, L., Levine, R., 2007. Governance and banks valuations. Journal of Financial Intermediation 16, 584–617. Coles, J., Daniel, N., Naveen, L., 2007. Boards: does one size fit all? Journal of Financial Economics 87, 329–356. Crawford, A.J., Ezzell, J.R., Miles, J.A., 1995. Bank CEO pay-performance relations and the effects of deregulation. Journal of Business 68, 231–256. Dahya, J., Dimitrov, O., McConnell, J.J., 2008. Dominant shareholders, corporate boards, and corporate value: a cross-country analysis. Journal of Financial Economics 87, 73–100. Denis, D., McConnell, J.J., 2003. International corporate governance. Journal of Financial and Quantitative Analysis 38, 1–36. DeYoung, R., Flannery, M., Lang, W., Sorescu, S., 2001. The information content of bank exam ratings and subordinated debt prices. Journal of Money, Credit and Banking 33, 900–925. Doidge, C., Karolyi, G., Stulz, R., 2007. Why do country characteristics matter so much for corporate governance? Journal of Financial Economics 86, 1–39. Fama, E., Jensen, M., 1983. Separation of ownership and control. Journal of Law and Economics 26, 301–325. Ferreira, D., Kirchmaier, T., Metzger, D., 2011. Boards of Banks. ECGI-Finance Working Paper. Furfine, C.H., 2001. Banks as monitors of other banks: evidence from the overnight federal funds market. Journal of Business 74, 33–57.

Goh, B.W., 2009. Audit committees, boards of directors, and remediation of material weaknesses in internal control. Contemporary Accounting Research 26 (2), 549–579. Guest, P.M., 2008. The determinants of board size and composition: Evidence from the UK. Journal of Corporate Finance 14 (1), 51–72. Harris, M., Raviv, A., 2008. A theory of board control and size. Review of Financial Studies 21, 1797–1832. Hermalin, B., Weisbach, M., 1998. Endogenously chosen boards of directors and their monitoring of the CEO. American Economic Review 88, 96–118. Hubbard, R.G., Palia, D., 1995. Executive pay and performance: evidence from the U.S. banking industry. Journal of Financial Economics 39, 105–130. John, K., Senbet, L.W., 1998. Corporate governance and board effectiveness. Journal of Banking and Finance 22 (4), 371–403. Kim, K.A., Kitsabunnarat-Chatjuthamard, P., Nofsinger, J.R., 2007. Large shareholders, board independence, and minority shareholder rights: evidence from Europe. Journal of Corporate Finance 13 (5), 859–880. Klein, A., 1998. Firm performance and board committee structure. Journal of Law and Economics 41, 275–299. Klein, A., 2002. Audit committee, board of director characteristics, and earnings management. Journal of Accounting and Economics 33, 375–400. Kole, S.R., Lehn, K.M., 1999. Deregulation and the adaptation of governance structure: the case of the U.S. airline industry. Journal of Financial Economics 52, 79–117. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 1998. Law and finance. Journal of Political Economy 106, 1113–1155. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Vishny, R., 2000. Investor protection and corporate governance. Journal of Financial Economics 58, 3–27. Laeven, L., Levine, R., 2009. Corporate governance, regulation, and bank risk taking. Journal of Financial Economics 93, 259–275. Levine, R.E., 2004. The Corporate Governance of Banks: A Concise Discussion of Concepts and Evidence. Unpublished Working Paper. World Bank Policy Research Working Paper No. 3404. Li, L., Song, F.M., 2011. Corporate Governance and Bank Valuation: A Cross-country Analysis. Unpublished Working Paper. University of Hong Kong, HK. Linck, J., Netter, J., Yang, T., 2008. The determinants of board structure. Journal of Financial Economics 87, 308–328. Morgan, D., 2002. Rating banks: risk and uncertainty in an opaque industry. American Economic Review 92, 874–888. Prowse, S., 1997. Corporate control in commercial banks. Journal of Financial Research 20, 509–527. Raheja, C., 2005. Determinants of board size and composition: a theory of corporate boards. Journal of Financial and Quantitative Analysis 40, 283–306. Schranz, M.S., 1993. Takeovers improve firm performance: evidence from the banking industry. Journal of Political Economy 101 (2), 299–326. Shleifer, A., Vishny, R.W., 1997. A survey of corporate governance. Journal of Finance 52, 737–783. Smith, C., Watts, W., 1992. The investment opportunity set and corporate financing, dividend, and compensation policies. Journal of Financial Economics 32, 263–292. Yermack, D., 1996. Higher valuation of companies with a small board of directors. Journal of Financial Economics 40, 185–212.