The Role of Empowered Minority Shareholders on Bank Liquidity Creation

The Role of Empowered Minority Shareholders on Bank Liquidity Creation by Violeta Díaz 1 and Ying Huang 2 1 New Mexico State University, Finance Department, MSC 3FIN, P.O. Box 30001, Las Cruces, NM 88003, USA Phone: 575-646-2954, Email: diazv@nmsu.edu. 2 University of Manitoba, I.H. Asper School of Business, Department of Accounting and Finance, 181 Freedman Crescent, Winnipeg, MB R3T 5V4, Canada Phone: 204-272-1503, Email: ying.huang@umanitoba.ca 1

The Role of Empowered Minority Shareholders on Bank Liquidity Creation Abstract We examine the impact of internal governance on bank liquidity creation. Stronger minority shareholder protection in banks internal governance and more shareholder-friendly boards significantly increase bank liquidity created from on-balance sheet activities. However, more stringent executive compensation rules and practices are associated with a significant reduction in liquidity created from the on and off-balance sheet activities. Berger et al. (2012) demonstrate that regulatory interventions cause a reduction in liquidity creation. We show that the comparative power of minority shareholders can have interactive effects on liquidity creation after controlling for government intervention during the most recent financial crisis. 2

1. Introduction Banks play a pivotal role in the economy by performing two major tasks: creating liquidity and transforming risk. While risk transformation and its associated risk taking behavior have been examined in numerous empirical studies focusing on the linkage between banks risk-taking behavior, the ownership structure in governance, and regulation (Saunders, Strock, and Travlos, 1990; Esty, 1998; John, Litov, and Yeung, 2008; Caprio, Laeven, and Levine, 2007, Barth, Caprio, and Levine, 2002; Laevan and Levin, 2009), liquidity creation has been neglected and empirical studies on this topic are scant. This research is motivated by the need to explore the role of bank liquidity creation in the years leading to the sub-prime mortgage crisis, and during the great recession. During this time, banks faced liquidity problems, losses and failures and investments banks became bank holding companies in order to gain access to alternative liquidity sources. Furthermore, the relationship between the empowered minority shareholders and bank liquidity creation is unknown. However, this relationship has significant implications from economic and political stand points. The political procedures on the empowerment of minority shareholders may lead to changes in bank liquidity creation and have a significant impact on the stability of banks and the entire financial system. Mehran, Morrison, and Shapiro (2011) argue that corporate governance may be especially weak for banks and financial institutions due to the multiplicity of stakeholders (insured and uninsured depositors, the deposit insurance company, bond holders, subordinate debt-holders and hybrid securities holders), and the complexity of banks operations. In this study, we empirically examine the impact of internal governance on liquidity created from on- and off-balance-sheet activities respectively. Our empirical results show that types of liquidity creation vary by the empowerment of minority shareholders in the internal governance of banks, the independence of the board, and the strength of executive compensation rules and practices. 3

Moreover the moral hazard created by the perception of too-big-to-fail institutions may have led boards to encourage risk-taking as they knew that taxpayers rather than stakeholders would pay their big losses. Further, Beltratti and Stulz (2012) present empirical evidence that poses a challenge to a conventional view that poor bank governance was a major cause of the banks failure and the consequent sub-prime crisis. They show that there is no evidence banks with shareholder-friendly board performed better or were less risky during the crisis period than other banks. Thus, it is not clear what role the internal governance plays in banks risk taking activities, liquidity creation, and their subsequent failure. No previous research evaluates the interactive effects of the power of minority shareholders on liquidity creation. To our knowledge, we are the first study to show that bank liquidity creation is dependent upon the comparative power of minority shareholders. Our analysis bridges the gap in the relationship between bank liquidity creation, and governance structure in terms of minority shareholder protections in contrast to the ownership structure in banks. We confirm the results of ownership structure and risk taking behavior by Laeven and Levin (2009) by providing evidence that ties liquidity creation with empowered minority shareholders. We use liquidity creation measures developed by Berger and Bouwman (2012) to test the role of internal governance on liquidity creation controlling for risk-taking activities by using a GMM dynamic model, which accounts for endogeneity between the risk-taking activities and liquidity creation. We then form our two testable predictions. First, banks with empowered minority shareholders create more liquidity. More specifically, banks with better minority shareholders protection and more shareholder-friendly boards create higher liquidity relative to those with weaker minority shareholders protection and boards. Second, we explore the impact of stringent management and executive s compensation rules and practices on liquidity creation. We find that it leads to a reduction in liquidity created from off-balance-sheet activities. We further investigate the intertwining effect of government intervention and internal governance on 4

liquidity by using a proxy dummy variable. We define a crisis dummy variable which takes a value of 1 starting 2007 (Beltratti and Stulz, 2012) and 0 otherwise. Our results on how internal governance impact the liquidity creation still hold after controlling for the effect of government intervention and regulation during the recent financial crisis. The remainder of this paper is organized as follows. Section 2 presents the literature review and the development of our empirical hypotheses. Section 3 describes the data, sample, and methodology. Section 4 discusses the main results based on liquidity creation, stability, and governance measures, and Section 5 concludes. 2. Background and Literature Review 2.1. Liquidity Creation and Internal Governance Theoretical studies on banks activities associated with liquidity creation were pioneered several decades ago by Bryant (1980) and Diamond and Dybvig (1983). In these studies, the authors propose theories that explain a bank s role as liquidity creator by funding illiquid assets with liquid liabilities. However, until recently, empirical studies analyzing liquidity creation and the function of banks were very limited. Particularly, the role of liquidity creation leading to and during the recent sub-prime mortgage financial crisis is not clear. Our results indicate that after a continuous increase in liquidity creation since 2003, liquidity creation experienced a sharp decrease in 2007. In the midst of a financial crisis, the reduced liquidity brings an overall adverse effect not only to the entire financial system, but also to real sectors. Conversely, according to Acharya and Naqvi (2012), when there is excessive liquidity, the banking sector might initiate or aggravate an asset bubble which could potentially increase risk-taking moral hazard, and the vulnerability of the banking sector, setting off a financial crisis. As a result, liquidity creation and excessive risk-taking go hand in hand. It is expected that excessive risk-taking activities can be effectively curbed by strong shareholder protection and governance. The first papers to systematically study the corporate 5

governance of banks are Caprio and Levine (2002), Macey and O Hara (2003), Levine (2004). They review the major governance concepts for corporations in general and then discuss two special attributes of banks that make them special: greater opaqueness and greater government regulation. They start defining the corporate governance problem in terms of how holders of equity and debt influence managers to act in the best interests of the providers of capital. Previous studies in banks governance and risk taking behavior have focused on examination of ownership structure and its effect on risk taking. Saunders, Strok, and Travlos (1990) show that banks controlled by owners exhibit higher risk taking behavior relative to banks controlled by managers with small shareholdings. Laeven and Levine (2009) provide empirical evidence that bank risk is generally higher in banks that have large owners with substantial cash flow rights. They further document that owners seek to compensate for the utility loss from capital regulations and activity restrictions by increasing bank risk. However, the controlling owners choose actions to maximize their welfare rather than the welfare of the minority shareholders (Aggarwal, Erel, Stulz, and Williamson, 2008). Further, these authors show that minority shareholders benefit from governance improvements and do so partly at the expense of controlling shareholders. Thus, as a supplementary measure, minority shareholders protection, also serves as a rivaling force to large owners, and is an ideal measure for assessing the interacting effect of risk staking and role of internal governance in liquidity creation. The internal governance measure we used is a firm-level governance index: Corporate Governance Quotient (CGQ) provided by ISS (Institutional Shareholder Services) 1. The index assesses the investment in increasing power of minority shareholders relative to other BHCs in financial industry. By delegating stronger minority shareholder protection, the controlling 1 The minimum standard is provided and described in ISS Corporate Governance Quotient- CGQ Best Practices Manual published December, 2008. 6

shareholder reduces his ability to extract private benefits and therefore incurs private costs (Aggarwal, Erel, Stulz, and Williamson, 2008; Aggarwal, Erel, Ferreira, and Matos, 2011). Berger and Bouwman (2012) show that liquidity creation is linked to higher market to book values and price earnings ratios. Further, according to De Andres and Vallelado (2008), bank board composition and size are related to its ability to monitor and advise efficiently. These characteristics prove to be more efficient in monitoring an advising to create more value. We include these characteristics in our analyses to investigate the impact of minority shareholder protection and to bridge the gaps in bank liquidity creation research. As a result, we conjecture that banks with stronger minority shareholder protection and shareholder-friendly boards tend to create more liquidity than those with weaker minority shareholder protection and less independent boards as put forth by our first hypothesis. H1: More empowered minority shareholders and more shareholder-friendly boards increase bank liquidity creation. Moreover, strategic decisions by managers may be driven by their motivation to increase their bonuses, short-term stock price movements, or shareholders short-run interests rather than stakeholders long-run interests. And these strategic decisions involve largely off-balance-sheet activities such as taking positions in forwards, futures, swaps, and other derivatives, and mortgage servicing contracts which are highly risky in nature (Saunders and Cornett, 2013). Core, Holthausen, and Larcker (1999) indicate that CEOs earn greater compensation for firms with greater agency problems and with less effective governance structures. Thus, if the executive compensation plans and practices are strictly governed and monitored by shareholders, strategic decisions associated with risky off-balance-sheet activities are expected to decline. This is due to the fact that eventually these off-balance-sheet activities shape the earnings and performance measures on the balance sheet and consequently affect CEOs pay for performance. 7

Bryant (1980) and Diamond and Dybvig (1983) propose that most liquidity creation takes place from on-balance sheet activities. However, other theoretical studies suggest that banks can also create liquidity off the balance sheet (Holmstrom and Tirole, 1998; Kashyap, Rajan, and Stein, 2002). In fact, Berger and Bouwman (2012) show that half of the liquidity created by banks is from off-balance sheet activities. Thus, it is important to differentiate liquidity created on the balance sheet from off the balance sheet. This conjecture completes our second hypothesis as following. H2: More stringent executive compensation rules and practices decrease bank liquidity creation from off-balance-sheet activities. 2.2. Empowered Minority Shareholders, Liquidity Creation, and Regulation Overlay Due to their business nature, banks and financial institutions are heavily regulated. When banks are troubled, authorities often engage in regulatory interventions and provide capital support to resuscitate the bank and to reduce risk-taking activities (Bhattacharya, Boot, and Thakor,1998; Diamond and Rajan, 2005; Oshinsky and Olin, 2005; Philippon and Schnabl, 2013; Hoshi and Kashyap, 2010). Furthermore, during 2007-2009, a considerable number of banks failed, while many others had to be rescued by regulatory interventions and capital support. For example, TARP was established in the fall of 2008, and eventually included 13 programs implemented by the U.S. Treasury (Contessi and El-Ghazaly, 2011). The Capital Purchase Program (CPP) was implemented soon after, which enabled a $145 billion capital injection to nine major banks. The objectives of authorities are twofold; to reduce risk taking activities by reducing liquidity creation on one hand, while on the other hand to enable banks to lend more and to increase liquidity (Berger, Bowman, Kick, Schaeck, 2010). Berger, Bowman, Kick, and Schaeck (2012) find that capital support and regulatory interventions are associated with significant reductions in liquidity creation. We are interested in the role of empowered minority 8

shareholders on liquidity creation under the mediating effect of government intervention. We expect an interactive effect of internal governance on liquidity creation after controlling for the array of new regulation during the crisis period. 3. Data 3.1 Sample We construct the liquidity creation measures by following the three step procedure developed by Berger and Bouwman (2012). In the first step, balance sheet items: assets, liabilities, equity, and off balance sheet items are classified as liquid, semi-liquid, or illiquid based on their ease, cost, and time for customers to obtain liquid funds from the bank. The offbalance sheet guarantees and derivatives are classified consistently with on-balance sheet items functionality 2. In the second step, we assign weights to all bank activities consistent with liquidity creation theory. Positive weights are applied to both illiquid assets and liquid liabilities, thus liquidity is created when a liquid liability is used to finance illiquid assets: for example, when deposits are used to finance a small business. Similarly, negative weights are applied to liquid assets, illiquid liabilities, and equity. In this situation, liquidity decreases when illiquid liabilities or equity is used to finance liquid assets. In the third step, liquidity creation measures are created by combining the activities with their respective weights. Thus, CATFAT and CATNONFAT measure liquidity created on the balance sheet including or excluding off-balance sheet activities, respectively. As Berger and Bouwman (2012) reasoned, the CATFAT is the preferred measures for liquidity creation due to the fact that off-balance sheet activities provide liquidity in functionality similar to on-balance sheet items. Berger and Bouwman (2012) show that on average only 50 percent of liquidity is created from on-balance activities, while the other half is created from off-balance sheet activities. Furthermore, off-balance activities shape the 2 See Berger and Bouwman (2012) for details. 9

future balance sheet because they involve the creation of contingent assets and liabilities that give rise to their future potential placement on the balance sheet (Saunders and Cornett, 2013). Our initial sample comprises firms in the firm-level Corporate Governance Quotient (CGQ) dataset compiled by Institutional Shareholders Services (ISS). The firm level minority shareholder protection data are available for a total of 495 BHCs for the years 2003 through 2009. Starting in 2003, every two years, ISS has added additional governance attributes to the CGQ matrix and more firms to the dataset. For consistency, we construct governance measures using only the attributes that are consistently available across our study period. As a result, we are left with 309 total BHCs with ISS governance measure. Following Aggarwal, Erel, Stulz, and Williamson (2008) and Jiraporn, Kim, and Kim (2011), we award the attributes to form the index for BOARD, COMPENSATION, and GOV. The GOV index is constructed by attributing one point for each of the 50 governance standards listed in the Appendix B that apply to the firm. 3 The 50 items cover eight categories in governance such as audit, board of directors, charter/bylaws, director education, executive and director compensation, ownership, progressive practices, and state of incorporation. Each of these eight categories measures a different dimension of minority shareholder protection. BOARD index includes important board characteristics such as board independence, board size, CEO separation etc. Significant characteristics such as no interlocking among directors, option repricing, approval of incentive plans, etc are included in the COMPENSATION measure. We include the comprehensive measures of minority shareholder protection: GOV index, BOARD, and COMPENSATION in our analyses. 3 The minimum standard is provided and described in ISS Corporate Governance Quotient- CGQ Best Practices Manual published December, 2008. 10

The advantage of using ISS governance matrix measures, as concluded by Brown and Caylor (2006), is that it is broader in scope, incorporates more firms and is more dynamic than other governance indices commonly used by other studies. According to Aggarwal, Erel, Ferreira, and Matos (2010), the GOV index represents the empowered minority protection. According to Beltratti and Stulz (2012), the BOARD index is higher for more independent boards and lower for staggered boards. Similarly, the COMPENSATION measure is higher for those with more stringent rules and practices in executive and director compensations. Our final sample is formed by merging the CGQ, governance matrix data with three other data sources: Compustat quarterly, CRSP, and Bank Regulatory for Bank Holding Companies. We obtain the accounting and financial ratios from Compustat database and U.S. annual CPI index from CRSP for firms that have GOV index. Following previous research, we zero the records with missing value in research and development (R&D) expenses, taxes and interest expense, and dividend and repurchases to fully utilize the data. We gather loans information, bank assets, equity, etc. in the financial statements from Bank Holding Company Performance Reports. Finally, we winsorize the sample at the 1 st and 99 th percentile to mitigate the effects of outliers. In our analyses, we control for the distance to insolvency and measures for risk taking behaviors by including z-score, risk weighted total asset ratio, and credit risk ratio. The distance to insolvency, z-score, is computed by following Laeven and Levine (2009). The z-score is a measure of the likelihood of insolvency and it is defined as the return on assets plus the capital asset ratio divided by the standard deviation of asset returns. As suggested by Laevene and Levine (2009), the z-score are highly skewed and thus are log-transformed. We also control for risk-taking behavior by including the risk weighted assets ratio and credit risk ratio in our analyses. The risk weighted assets ratio is computed by dividing risk weighted assets by total 11

assets. The credit risk ratio is obtained by dividing the non-performing loans by the total loans. In addition to risk taking behavior, we also control for the crisis period, bank size using the log of total assets, market to book value ratio, and bank performance measured by Return on Assets (ROA), Return on Equity (ROE), and quarterly stock returns. 3.2 Dynamic Panel GMM Model We employ dynamic adjustment and Generalized Moment of Methods (GMM) to investigate our research hypotheses. We specify an adjustment dynamics that relates the actual (observed) level of liquidity creation to its optimal (unobserved) equilibrium value. This allows us to consider both the short-run and long-run components of liquidity creation. These dynamic relationships are characterized by the presence of a lagged dependent variable among the independent variables of the model. However, dynamic panel data models pose serious estimation challenges, as the conventional panel data estimators are inconsistent in the presence of a lagged-dependent variable (Baltagi, 2005). Also, it is our purpose to gain a robust sense of the role governance and risk-taking on liquidity creation in order to suggest best practices or regulatory guidance. However, the notion of causation is delicate: Does certain characteristics or risk behavior lead banks to make liquidity choices or certain liquidity choices lead a bank to have a culture of risk-taking behavior? Does certain characteristics lead banks to create liquidity or liquidity creation lead to better governed institutions? Due to the endogeneity problem, most of these relationships should often be interpreted as correlation and not causalities. To solve this problem, one would usually use fixed-effects instrumental variables estimation (two-stage least squares) assuming that appropriate instruments are available. Alternatively, one may use a number of GMM estimators (Arellano and Bond, 1991; Arellano and Bover, 1995; Blundell and Bond, 1998). Arellano and Bond (1991) transform the model into first differences to wipeout the individual specific effects. 12

Sequential moment conditions are then used where lagged levels of the variables are instruments for the endogenous differences. Arellano and Bover (1995) and Blundell and Bond (1998) propose the use of extra moment conditions arising from the model in levels when certain stationarity conditions of the initial observation are satisfied. The resulting system GMM estimator combines moment conditions for the model in first differences with moment conditions for the model in levels. In our study, we use the Arellano-Bover/Blundell-Bond system GMM estimator that uses lagged differences of y i,t as instruments for equations in levels, in addition to lagged levels of y i,t as instruments for equations in first differences (Arellano and Bover, 1995) where y i,t is liquidity creation, the dependent variable in our model. It has been well documented (Blundell and Bond, 1998) that the first differences GMM estimator can have very poor finite sample properties in terms of bias and precision when the series are persistent, as the instruments are then weak predictors of the endogenous changes. The system GMM estimator has been shown in Monte Carlo studies by Blundell and Bond (1998) and Blundell, Bond, and Windmeijer (2000) to have much better finite sample properties in terms of bias and root mean squared error than that of the difference GMM estimator. This method assumes that there is no autocorrelation in the idiosyncratic errors and requires the initial condition that the panel-level effects be uncorrelated with the first difference of the first observation of the dependent variable. The following regression equation is formulated to test empirically the main hypotheses. ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (1) ( ) ( ) where LC stands for liquidity creation CATFAT or CATNONFAT; RWA is the risk weighted assets ratio; CRisk is the credit risk; G is the governance index, GOV, BOARD, or 13

COMPENSATION; crisis is a dummy variable defined as 1 during crisis period from 2007 to 2009 and 0 otherwise; Equity is the ratio of market value fiscal to total assets; M/B is the market value fiscal divided by total assets; Performance measures include stock return computed as ln(p t )-ln(p t-1 ) where P t is the quarterly closing price, ROA, and ROE; i=1,,n indexes the BHC, and t=1,.., T indexes the time period, and Ln is the natural logarithmic. represents timespecific effects, firm-specific effects, and is a random disturbance that is assume to be normally distributed with mean 0 and variance. In the system GMM, first-differenced variables are used as instruments for the equations in levels and the estimates are robust to unobserved heterogeneity, simultaneity and dynamic endogeneity, if it exists. To test model specification validity, we calculate the Hanse/Sargan test of overidentification of restrictions. This test examines that the instruments and the error term are not correlated. The S1 and S2 statistics measure first- and second-degree serial correlations. Given the use of first difference transformations, we expect some degree of first-order serial correlation. However, the presence of second-degree serial correlation indicates an omitted variables problem. 4. Empirical Results 4.1 Univariate Results The descriptive statistics for variables are shown in Table 1. The average governance score is close to 12, much lower than the average governance score for non-financial firms reported in previous studies (Jiraporn, Kim, and Kim, 2011; Aggarwal, Erel, Stulz, and Williamson, 2008; Aggarwal, Erel, Ferreira, and Matos, 2011). The average BOARD score is 9.25 and the average COMPENSATION score is 4.49. The compensation score of Bank holding companies is even lower than the other two governance scores. The z-score in our final sample is 12.55 or 2.88 after taking natural logarithm, similar to the value reported in Laeven and Levine (2009). The average 14

market-to-book ratio is 0.14 with a median of 0.12 and maximum of 0.82, smaller compared to the non-financial firms. The quarterly stock return is negative, however nearly zero. In Figure 1, we plot the liquidity creation measures CATFAT and CATNONFAT against the z-score, risk weighted assets ratio, GOV, BOARD, and COMPENSATION. Liquidity creation in terms of CATFAT and CATNONFAT is increasing over time, particularly from the third quarter of 2004 to the third quarter of 2007. However, the liquidity creation leveled off in period after the peak of the crisis at the end of 2008. There is a steadily slow increase in z-score from 2003 to 2006. However, in year 2007 this measure experienced a sudden jump, revealing a significant reduction in distance to insolvency (Imbierowicz and Rauch, 2014), and bank stability at the beginning of the sub-prime mortgage financial crisis. The z-score seems to coincide with liquidity creation, indicating that liquidity is positively associated with stability (Panel A). As revealed in Panel B, C, D, and E, liquidity creation from on balance sheet activities seems to increase with risk taking behavior. However, liquidity creation from both on and off-balance sheet activities increases in the index values of G: GOV, BOARD, and COMPENSATION, while the liquidity creation trajectories are not closely adherent to G index. Panels in Figure 1 indicate a deviation in the trajectories in 2008 and 2009 in all measures. We believe, according to the findings by Berger, Bouwman, Kick, and Schaeck (2010), and our findings, that the reduction in the liquidity creation after the onset of sub-prime mortgage financial crisis can be explained by the regulatory interventions and capital injections by regulators. 4.2 Multivariate Results In this section we present the results of the dynamic model based upon the Arellano and Bover (1995) and Blundell and Bond (1998) system GMM estimation of Eq (1). All GMM estimations assume strict exogeneity of all explanatory variables. Consequently, the instruments 15

include the lagged differences of the dependent variable as instruments for equations in levels, in addition to lagged levels of the dependent variable as instruments for equations in first differences. Unlike static estimation models, in dynamic models the first difference transformation is applied to all variables, which is required by the Arellano and Bond s approach in order to remove the fixed effects. Unfortunately, this comes at the cost of reducing the effective sample size. We present estimates of six alternative models. For each regression, we include first and second order correlation tests (S 1 and S 2 ), the Sargan test of instrument validity, and the Wald test of model statistical significance. The statistical tests confirm the absence of second-order serial correlation, and the validity of the model, and the validity of the instruments we use to avoid the endogeneity problem. The first three models, Model 1 through Model 3, estimate the dynamic version of the model using only the z-score and the ratio of risk weighted assets to total assets as risk measures. In Model 4 through Model 6, we add credit risk, defined as the ratio of non-performing loans to total loans, and three risk measures in our estimation. Models 1 and 4 use return on assets (ROA), Models 2 and 5 use return on equity (ROE), and Models 3 and 6 use the quarterly stock return as performance measure. As a starting point, we present statistical results of the estimations of the dynamic liquidity creation model using both liquidity measures CATFAT and CATNONFAT as defined in Appendix A. Tables 2, and 3 report results for specifications using GOV index for CATNONFAT and CATFAT respectively. The GOV index in Table 2 has consistently and significantly positive coefficients across all the models, indicating that empowered minority shareholder protection improves liquidity creation from on-balance activities after controlling for risk taking behavior, the z-score, credit risk, size, M/B ratio, and performance. However, the GOV index coefficients in Table 3 are not significant across all the models, suggesting offbalance-sheet activities are unrelated to the GOV index. 16

We further explore the relation between empowered minority shareholder protection and liquidity creation by including BOARD and COMPENSATION in the models. Tables 4 and 5 present results on liquidity creation from on- and off-balance activities respectively by using the BOARD index. The results are similar to and consistent with the results reported in Tables 2 and 3 using GOV index.the relationship between liquidity ad BOARD index is stronger than with the GOV index. The relationship is only statistically significant for liquidity created from on balance sheet activities (CATNONFAT). In addition, large boards give excessive control to the CEO, harming efficiency (Eisenberg, Sundgren, and Wells, 1998). The BOARD variable considers, among others, as positive characteristics that the CEO serves on no more than two additional boards, and that the size of the board is between six and fifteen members. Evidence suggests that additional directorships of busy directors affect the performance of public companies (Fich and Shivdasani, 2006). Moreover, the BOARD variable considers the separation of duties of the BHC s CEO from the chairman of the board as a positive characteristic. When these two positions are held by different individuals then the quality of the board increases. Thus, we conclude that better minority shareholder protection boards as indicated by higher scores in GOV and BOARD are associated with higher liquidity creation from on-balance-sheet activities, supporting our first hypothesis. Tables 6 and 7 present the impact of executive and director compensation rules and practices: COMPENSATION on liquidity creation CATFAT and CATNONFAT respectively. The effect of COMPENSATION on CATFAT is negative and statistically significant. However, this relation is not statistically significant for CATNONFAT. These findings indicate that COMPENSATION reduces liquidity created on and off-balance activities, while has no impact on CATNONFAT. Previous studies (Cai, Cherny, and Milbourn, 2010) suggest that financial institutions managers that engage in complex transactions, such as derivatives trading would be well-compensated in the event risky activities pay off. In fact, Ben S. Bernanke, the former chairman of the Federal 17

Reserve, blamed compensation practices as the primary reason for misaligned incentives and excessive risk-taking in banking 4. These results support Hypothesis II. In all the models we examined, we control for the relation between risk and liquidity creation by including z-score, a reverse proxy for risk-taking behavior, measuring the distance from insolvency and risk-weighted assets ratio, a measure for risk-taking activities. According to the risk absorption theory, higher capital ratios, which embody high risk-taking activities, may allow banks to create more liquidity by improving bank s ability to absorb risk (Bhattacharya and Thakor, 1993; Repullo, 2004; Von Thadden, 2004; Coval and Thakor, 2005). Thus, we would expect under this context there would be an increase in liquidity creation as risk-taking behavior increases. Wagner (2007) proves an inversely related relationship between liquidity and stability, and risk-taking. He shows that liquidity increases stability. However, this increased liquidity and stability lead to an increased engagement in taking on risks that more than offsets positive direct impact on stability. Beltratti & Stulz (2011) suggest, following this argument, that banks with better governance will take more risks. Excessive risk-taking must therefore involve a breakdown in control or lead to a desire on the part of the board to encourage such activity. A higher z-score indicates that the bank is more stable and a higher risk-weighted assets ratio indicates a higher level of risk taking behavior. We found that as the stability of the bank increases or the distance from insolvency increases, the bank will be able to create more liquidity as the bank s capital structure becomes less fragile. Coincidently, liquidity creation decreases with risk taking activities. We also remove the notion that risk-taking behavior attributes to the changes in liquidity creation by conducting Granger causality test. In our analysis, we fail to reject the null hypothesis that risk-taking behavior does not cause the changes in liquidity 4 Board of Governors of the Federal Reserve System press release, October 22, 2009 http://www.federalreserve.gov/newsevents/press/bcreg/20091022a.htm accessed on February 28, 2014. 18

creation at 0.001 significance level. In addition, coefficient estimates of additional control variables, credit risk, size, M/B ratio, return, and performance. are consistent in all models. In all the models we examine, we also control for the effect of the 2008 financial crisis. The coefficient of the crisis dummy variable indicates a significant decrease in liquidity creation after the second quarter of 2007, at the start of the financial crisis. This effect is consistent across specifications and governance measures. The evidence suggests that liquidity creation was negatively affected during the financial crisis due to contagion among financial institutions as suggested by Diamond and Rajan (2005). The coefficients of other bank characteristics variables offer some important insights. Market to book value and the size of the bank, have positive and significant effect on bank s liquidity creation. These results confirm previous findings on the cross section effects of value and size on liquidity creation of banks in the U.S. (Berger and Bouwman, 2012). Liquidity creation differs considerably by size. According to Berger and Bouwman (2012), large banks with gross total assets of more than $3 billion are responsible for 81% of the industry liquidity creation. 4.3 Robustness Our previous results suggest that during the crisis period, liquidity creation declined significantly. Motivated by this change, it is our intention in this section to analyze if the effect of governance on liquidity creation changed during this period. For that purpose, we estimate Equation (1) with an additional interaction term that considers governance with the financial crisis dummy variable as follows: ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) ( ) (2) ( ) 19

where G is GOV index, BOARD or COMPENSATION. The other variables are defined in the same way as in Equation (1). The crisis period dummy is defined as 1 starting in 2007 and 0 otherwise. The crisis dummy is a proxy for government intervention during the crisis period. Tables 8 and 9 present results for the estimation of equation (2) with CATNONFAT and CATFAT as dependent variables, respectively. Table 8 presents results for liquidity created on the balance sheet. The interaction term of the GOV index and the crisis dummy is positive and significant, but when the interaction term is added into the model, the governance measure by itself becomes insignificant. This result suggests that GOV index mitigates the negative impact of the crisis on liquidity creation. Columns 3 and 4 in Table 8 show results for BOARD. These results show that BOARD quality is independent of the industry environment. The interaction term is not significant, and the relation between BOARD and liquidity creation, CATNONFAT, is positive and statistically significant and in the same magnitude as in Table 4. In addition, for columns 5 and 6, the interaction term of COMPENSATION and the crisis dummy is positive and significant, and we see the same sign in the COMPENSATION variable when the interaction term is included. The negative effect of compensation practices and rules on liquidity creation is mitigated during the crisis possibly due to government intervention in bank compensation and bonuses consistent with regulatory interventions during 2008 and 2009. Table 9 shows results for liquidity creation CATFAT. The models with GOV index yields similar results as in Table 8, revealing that the empowerment of minority shareholders and stringent compensation rules and practices has a mediating effect on liquidity creation by lessening the reduction in liquidity creation caused by government intervention during the recession. For the model using the BOARD variable, the governance measures still shows a positive effect on liquidity creation but not significantly for CATFAT. The interaction effect between BOARD and the crisis period is also insignificant. However when using the 20

COMPENSATION variable, the relation between compensation practices and rules, and liquidity creation is again statistically significantly negative, consistent with previous results (see Table 7). This result suggests that the negative effect of more stringent compensation rules and practices on liquidity creation is unchanged before and during the crisis but its effect is mitigated by government intervention during this period. Considering both coefficients of the interaction term, and the intervention dummy variable, the reduction in liquidity created from on and off-balance sheet activities due to government intervention is alleviated by the better internal governance, i.e., the more empowerment of minority shareholders and more stringent compensation rules and practices. This result has policy implications and supports awareness on minority shareholder protection and compensation incentives, and the role they played on managers and traders behavior during 2006-2008. The debate on internal governance and compensation has now gained even more popularity among policy makers. For example, the say on pay policies adopted in 2011 by the SEC, and required under the Dodd-Frank Act 5. In addition, we split our sample into two subsamples by size, and repeated our analysis from Tables 2 to 7, obtaining the same results 6. Thus, we conclude that our evidence supports both hypotheses, and our previous results as presented in Tables 2 to 7. 5. Conclusion Banks exert an essential role in economy by performing two fundamental functions: 1) creating liquidity and 2) transforming risk to facilitate economic growth. Prior research has revealed that risk transformation is closely related to bank risk-taking, bank regulation, and bank failures. Bank liquidity, however, remains a relatively unexplored area, especially its relationship 5 Amendment to the Securities Act of 1943: SEC Adopts Rules for Say-on-Pay and Golden Parachute Compensation as Required Under Dodd-Frank Act https://www.sec.gov/news/press/2011/2011-25.htm, accessed on February 28, 2014. 6 The results, due to space, are not reported for all the models in Table 2 to 9; however, the results are provided upon request. 21

to the roles of bank risk-taking, bank regulation, internal governance, and the probability of banks failure. We compute two liquidity measures: one exlcuding off-balance sheet activities and the other including off-balance sheet activities as on- and off-balance sheet liquidity creation are equally important. We investigate the role of internal governance on banks liquidity creation by studying bank holding companies over the period of 2003 to 2009. Our evidence shows that the liquidity creation in banks increased over the period from 2003 to 2007 and leveled off after the onset of the sub-prime mortgage financial crisis in the third quarter of 2007. In contrast, the z- score, which is a proxy for the probability of insolvency of banks, stayed stable at a value around 13 during the period from 2003 to 2006 and experienced a sudden drop in the beginning of 2006 and continued to decrease sharply after the third quarter of 2007. Further, the trajectory of liquidity created from on-balance activities is closely cohesive to that of the BOARD and COMPENSATION, while the liquidity created from off-balance-sheet activities slightly deviate from BOARD and COMPENSATION. We examine the impact of internal governance on liquidity created from on- and off-balance sheet activities respectively. In our models, we control for bank stability, risk-taking behavior, and credit risk. We find that in addition to the inverse effect of regulatory interventions during the crisis on liquidity creation (Berger, Bouwman, Kick, and Schaeck, 2010), bank stability and risk-taking behavior have a significant effect on bank liquidity creation as well. Our empirical results show that the empowered minority shareholders in the internal governance of banks, BOARD and COMPENSATION, have a significant impact on liquidity creation. Specifically, stronger minority shareholder protection and more shareholder-friendly boards improve liquidity creation from on-balance activities, while more stringent executive and director compensation rules and practices reduce liquidity creation created from on and off-balance activities. Further, 22

we show that the effect of regulation overlay on bank liquidity creation depends on comparable empowerment of minority shareholder protection. 23

Acknowledgements We are grateful for the generous support of data by Joseph C. Henzlik and Sean Quinn at the Institutional Shareholder Services. The authors thank for useful comments Allen N. Berger, Christa H.S. Bouwman, Laurent Weill, Teng Yuan Cheng, the conferences participants conferences held by the Southern Finance Association meeting in Fajardo, Puerto Rico, 2014 and SMF Conference in Taiwan meeting, 2014. 24

REFERENCES Acharya, V. and H. Naqvi, 2012, The seeds of a crisis: A theory of bank liquidity and risk taking over the business cycle, Journal of Financial Economics 106, 349-366. Aggarwal, R., I. Erel, R. M. Stulz, R. Williamson, 2008, Differences in governance practices between U.S. and foreign firms: Measurement, causes, and consequences, Review of Financial Studies 23, 3131-3169. Aggarwal, R., I. Erel, M. Ferreira, P. Matos, 2011, Does governance travel around the world? Evidence from institutional investors, Journal of Financial Economics 100, 154-181. Arellano, M., O. Bover, 1995, Another Look at the Instrumental Variable Estimation of Error- Components Models, Journal of Econometrics 68, 29-51. Arellano, M., S. Bond,1991, Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations, Review of Economic Studies 58, 277-298. Baltagi, H. B., 2008, Econometric Analysis of Panel Data, John Wiley and Sons, Chichester. Beltratti, A. and R. M. Stulz, 2012, The credit crisis around the globe: why did some banks perform better, Journal of Financial Economics 105, 1-17. Berger, A. N., C. H. S. Bouwman, T. Kick, and K. Schaeck, 2010, Bank liquidity creation and risk taking during distress, Discussion Paper Series 2: Banking and Financial Studies No 05/2010, Deutsche Bundesbank. Berger, A. N., and C. H. S. Bouwman, 2012, Bank liquidity creation, The Review of Financial Studies 22, 3779-3837. Berger, A. N., C. H.S. Bowman, T. Kick, and K. Schaeck, 2012, Bank risk taking and liquidity creation following regulatory interventions and capital support, Working paper. Bhattacharya, S., A. W. A. Boot, A. V. Thakor, 1998, The Economics of Bank Regulation, Journal of Money, Credit, and Banking 30, 745-770. Bhattacharya, S. and A. V. Thakor, 1993, Contemporary banking theory. Journal of Financial Intermediation 3, 2-50. Blundell, R. and S. Bond, 1998, Initial Conditions and Moment Restrictions in Dynamic Panel Data Models, Journal of Econometrics 87, 115-143. Blundell, R., S. Bond and F. Windmeijer, 2000, Estimation in Dynamic Panel Data Models: Improving on the Performance of the Standard GMM Estimator., in B. Baltagi (ed.), Nonstationary Panels, Panel Cointegration, and Dynamic Panels, Advances in Econometrics 15, JAI Press, Elsevier Science. Bryant, J.,1980. A Model of Reserves, Bank Runs, and Deposit Insurance, Journal of Banking and Finance 4, 335-344. Cai, J., K. Cherny, and T. Milbourn, 2010, Compensation and Risk Incentives in Banking and Finance, Economic Commentary Sepember. 25

Caprio, G., L. Laeven, and R. Levine, 2007, Ownership and bank valuation, Journal of Financial Intermediation 16, 584-617. Caprio, G., Jr., R. Levine., 2002, Corporate governance in finance: Concepts and international observations. In: Litan, R.E.,Pomerleano, M., Sundararajan, V. (Eds.), Financial Sector Governance: The Roles of the Public and Private Sectors. Brookings Institution Press, Washington, DC, 17 50. Contessi, S. and H. El-Ghazaly, 2011, Banking crises around the world: Different governments, different responses, The Regional Economist, April. Core, J. E., R. W. Holthausen, D. F. Larcker, 1999, Corporate governance, chief executive officer compensation, and firm performance, Journal of Financial Economics 51, 371-406. Coval, J. D., A. V. Thakor, 2005, Financial intermediation as a beliefs-bridge between optimists and pessimists, Journal of Financial Economics 75, 535-569. De Andres, P., and E. Vallelado, 2008, Corporate Governance in Banking: The Role of the Board of Directors. Journal of Banking & Finance 32, 2570-580. Diamond, D. W., R. Rajan, 2009, The credit crisis: Conjectures about causes and remedies, American Economic Review 99, 606-610. Diamond, D. W., and P. H. Dybvig. 1983, Bank Runs, Deposit Insurance, and Liquidity, Journal of Political Economy 91, 401-419. Diamond, D. W., and R. G. Rajan, 2005, Liquidity shortages and banking crises, The Journal of Finance 60, 615-647. Diamond, D. W., and R. G. Rajan, 2000, A Theory of bank capital, The Journal of Finance 6, 2431-2465. Eisenberg, T., S. Sundgren, and M.T. Wells, 1998, Larger board size and decreasing firm value in small firms, Journal of Financial Economics 48, 35-54. Esty, B. C., 1998, The impact of contingent liability on commercial bank risk taking, Journal of Financial Economics 47, 189-218. Holmstrӧm, B., and J. Tirole, 1998, Public and Private Supply of Liquidity, Journal of Political Economy 106, 1 40. Hoshi, T., and A. K. Kashyap, 2010, Will the U.S. bank recapitalization succeed? Eight lessons from Japan, Journal of Financial Economics 97, 398-417. Imbierowicz, B. and C. Rauch, 2014, The relationship between liquidity risk and credit risk in banks, Journal of Banking & Finance 40, 242-256. John, K., L. Litov, B. Yeung, 2008, Corporate governance and managerial risk taking: Theory and evidence, The Journal of Finance 63, 1679-1728. Kashyap, A. K., R. Rajan, and J. C. Stein, 2002, Banks as Liquidity Providers: An Explanation for the Coexistence of Lending and Deposit-Taking, The Journal of Finance 57, 33 73. 26

Laeven, L., and R. Levine, 2009, Bank governance, regulation and risk taking, Journal of Financial Economics 93, 259-275. Levine, R., 2004, The corporate governance of banks: A concise discussion of concepts and evidence, World Bank Policy Research Working Paper. Macey, J. R. and M. O Hara, 2003, The corporate governance of banks, Economic Policy Review 9, 91-107. Mehran, H., A. Morrison, and J. Shapiro, 2011, Corporate governance and banks: What have we learned from the financial crisis? Federal Reserve Bank of New York staff report. Oshinsky, R., and V. Olin, 2005, Troubled Banks: Why don t they all fail? FDIC Working Paper 2005-03, March 2005, Federal Deposit Insurance Corporation, Washington, D.C. Philippon, T., and P. Schnabl, 2013, Efficient Recapitalization, Journal of Finance 68, 1-42. Repullo, R., 2004, Capital requirements, market power, and risk-taking in banking, Journal of Financial Intermediation 13, 156-182. Saunders, A., E. Strock, N. G. Travlos, 1990, Ownership structure, deregulation, and bank risk taking, The Journal of Finance 45, 643-654. Saunders, A., and M. Cornett, 2013. Financial Institutions Management: A Risk Management Approach. New York: McGraw-Hill. Von T., Ernst-Ludwig, 2004, Bank capital adequacy regulation under the new Basel Accord, Journal of Financial Intermediation 13, 90-95. Wagner, W., 2007, The liquidity of bank assets and banking stability, Journal of Banking & Finance 31, 121-139. 27