Basel III, Ownership Concentration, Risk-taking, and Capital Stability: Evidence from Asia. Abstract

Basel III, Ownership Concentration, Risk-taking, and Capital Stability: Evidence from Asia Abstract This study investigates the associations among bank risk-taking, ownership concentration, and the recently-proposed standard for capital stability (Basel III). Consistent with theory, the evidence shows that a rise in ownership concentration by one standard deviation increases the extent of risk-taking by as much as 6-8%. Although Basel III does not start taking effect until 2013, we hypothetically apply the capital stability standard on a sample of East Asian banks in the period 2005-2009. Our results suggest that an improvement in capital stability by one standard deviation diminishes the extent of risk-taking by 5.37% (as measured by the bank s Z-score). We also find that the standard for capital stability would have been more effective in countries with better economic development. Our results provide insights into the likely effects of Basel III and should be useful to a wide range of audiences, including policymakers, regulators, bankers, as well as practitioners and researchers. JEL Classification: G01, G18, G21, G28, G34, G38 Keywords: Basel III, Risk-taking, Bank Ownership, Capital Stability, Financial Crisis 1

I. Introduction Motivated by public policy considerations, this study examines the associations among bank risk-taking, ownership structure, and the recently-proposed banking regulations (Basel III). Risk taking by banks has tremendous effects on economic welfare, as demonstrated by the recent financial crisis (Bernanke, 1983; Calomiris and Mason, 1997, 2003a, b; Keeley, 1990). Consequently, it is critical to understand the risk-taking behavior of banks. A few prior studies analyze the impact of ownership and governance on the degree of risk taking for non-financial firms (John, Litov, and Yeung, 2008; Agrawal and Mendelker, 1987). With respect to banks, Laeven and Levine (2009) investigate the impact of ownership concentration on bank risk taking and report a positive association, i.e. higher ownership concentration leads to more risk-taking. The purpose of our study is two-fold. First, we extend Laeven and Levine (2009) by analyzing bank risk-taking in a much more recent period. In particular, Laeven and Levine s (2009) sample came from the late 1990 s and early 2000 s. Since then the banking industry has undergone so much transformation, owing to M&A activities, new regulations, corporate governance reforms, as well as the latest financial crisis. As a result, the conclusion in Laeven and Levine (2009) based on the sample from the last decade cannot be readily extended to the more current time period. Second and more importantly, the Basel Committee on Banking Supervision (BCBS), in response to the recent financial crisis, has developed new capital stability rules aimed at preventing financial crises in the future. These new rules are already informally referred to as Basel III. We attempt to ascertain the impact of these new capital adequacy rules on bank risktaking if they had been implemented in the late 2000 s. Furthermore, we also investigate how 2

ownership structure would have interacted with the new rules in shaping the risk-taking behavior of banks. The results of our study should offer interesting insights to policy makers and regulators, as well as researchers and financial economists. As far as risk-taking, theory suggests two possible effects of ownership concentration. First, shareholders in a limited liability are motivated to take more risk (Galai and Masulis, 1976; Esty, 1988). From this perspective, concentrated ownership should increase risk-taking. On the contrary, Burkart et al. (1997) argue that concentrated ownership can bring about excessive managerial oversight, thereby stifling managers incentives to pursue risky projects. Our empirical results, based on East Asian banks from the period 2005-2009, reveal that banks with more concentrated ownership show a tendency to take more risk. The results are robust even after accounting for bank-specific, country-specific characteristics, and variation across the years. We employ two measures of risk taking, namely the bank s z-score and the volatility of equity returns. In terms of economic significance, a rise in ownership concentration by one standard deviation is associated with an increase in risk-taking by 6.11% as measured by the Z- score, and 8.12% as measured by equity volatility. This finding is consistent with the prediction of agency theory, as well as with the evidence in Laeven and Levine (2009). In spite of all the various changes the banking industry has experienced in the last decade, ownership concentration continues to exert significant influence on the extent of risk taking. Furthermore, we recognize possible endogeneity and show that our results are not likely influenced by unobservable bank characteristics. Specifically, we exploit the insight from Altonji et al. (2005) to demonstrate that selection from unobservables would have to be much stronger than selection on observables to explain away the ownership effect. Moreover, to address 3

possible reverse causality, we relate ownership concentration in the earliest year of the sample to the degree of risk-taking in the subsequent years. Ownership in the earliest year could not have resulted from bank performance in the later years. The evidence suggests that ownership concentration more likely brings about risk-taking than vice versa. Finally, we execute an instrumental variable (IV) analysis and find that the OLS and the 2SLS estimates do not differ significantly, indicating that endogeneity should not be much of a concern. In addition, we explore the new capital requirements widely referred to as Basel III. Motivated by the financial crisis of 2008, the new rules were revealed in 2009 and have galvanized a great deal of debate among policy makers, regulators, bankers, as well as researchers and practitioners. Basel III is set to be implemented in the near future. In particular, the first phase of the implementation will start in 2013. Basel III is expected to be adopted in phases over time with complete implementation in 2018. The full impact of Basel III will not be known for a number of years. However, we attempt to ascertain how the new standards would have influenced bank risk-taking if they had been implemented in the second half of the last decade. This experiment should provide useful insights into the likely effects of Basel III in the future. Currently there is a wide range in the degree of implementation of Basel standards among nations. In particular, variations in the implementations of Basel II standards need to be accounted for to accurately study this issue. Towards this end, we study East Asian countries. Our focus on East Asia stems from two pertinent observations offered by researchers on the state of regulatory convergence towards Basel standards. First, East Asia is one of the most Basel IIconvergent regions, second only to the European Union. Europe (non-eu members) and the US 4

have not adopted Basel II standards to significant levels. 1 Second, the consistency in the pace of the convergence approaches across various East Asian nations is remarkable. Further, because the scope of the paper involves understanding the potential effect of Basel III, the logical consideration is to examine countries that have implemented Basel II. East Asian countries serve this purpose best. East Asian characteristics allow our information set to contain less noise, resulting in a more robust empirical analysis and more reliable conclusions. In sum, we choose to study East Asian nations because they have attained a fairly uniform and reasonably high degree of regulatory convergence by successfully implementing Basel II. Basel III concentrates on capital stability, as indicated by a new liquidity standard called the net stable funding ratio (NSFR). This funding ratio seeks to calculate the proportion of long-term assets which are funded by long-term, stable funding (BIS, 2010a) The detailed calculations are explained in the AppendixAppendix. Based on bank data from 2005 to 2009, we apply this capital stability standard on the extent of risk taking by banks. The empirical results show that the net stable funding ratio is significantly related to the bank s z-score, suggesting that it would have been a relevant factor if Basel III had been implemented. Specifically, banks with higher net stable funding engage in less risk taking. Basel III would have been successful in reducing the degree of risk taking, as indicated by the z-score. The impact of this new capital stability standard is not only statistically significance, but it is also economically meaningful. In particular, an increase in capital stability by one standard deviation diminishes the degree of risktaking by 5.37% (where risk is measured by the bank s Z-score; a lower Z-score indicates more 1 For instance, Cho (2010) suggests, Implementation in the US remains in flux since although the advanced approach rules of Basel II were adopted in 2007 for certain mandatory banks, the standardized approaches still remain as proposals. A similar observation is made by Chabanel (2011). 5

risk). Nevertheless, we do not find that the net stable funding ratio has a significant influence on the volatility of equity returns. Laeven and Levine (2009) show that the impact of banking regulations critically depends on ownership structure. That is the marginal effect of regulation on risk can be either positive or negative, contingent on ownership concentration. As a consequence, we explore how Basel III would have interacted with ownership structure. The evidence shows that the interaction effects are not significant. Nevertheless, we discover that the degree of economic development does matter. The standard for capital stability would have been more effective in reducing risk-taking in countries with a higher level of economic development (as indicated by higher GDP per capita). Perhaps, in developed countries, financial and accounting information is more reliable, due to more stringent disclosure requirements, hence more transparency. Enforcement of regulation based on more accurate information may be more effective in these developed countries. Finally, we attempt to shed light on the impact of the recent financial crisis. We find that the equity volatility of banks skyrockets by as much as 46% in the crisis period in our sample (2008-2009). However, the crisis does not appear to change the fundamental relationships among risk-taking, ownership concentration, and regulation. The results of our study contribute to several areas of the literature. First, we contribute to the literature on corporate governance and ownership structure. We show that ownership concentration is a significant determinant of risk-taking by banks. Although prior studies have investigated this subject, we extend the current literature by concentrating on a more recent 6

period, which also allows us to explore the effects of the latest financial crisis. 2 Second, our results contribute to the banking literature, showing that, although banks differ from nonfinancial firms in several ways, including being heavily regulated, their ownership structure is a significant determinant of their risk-taking. Third, our results offer insights into the probable effects of Basel III, expected to start being implemented in 2013. Although not yet applied, Basel III has generated a great deal of debate and discussions (For instance, Ojo, 2010; Allen, Chan, Milne, and Thomas, 2010; and Went, 2010). Our study aptly contributes to the on-going debate on the costs and benefits and other consequences of Basel III. Finally, the recent financial crisis has spawned a plethora of studies that investigate its consequences. We contribute to this current area of the literature by shedding light on the effects of the financial crisis on banks in East Asia. The results of our study should be of particular interest to several parties, including bank regulators, policy makers, practitioners, as well as financial economists. The remainder of this study is organized as follows. Section II discusses the theoretical background as well as the hypothesis development. Section III describes the sample construction and the data. Section IV presents the empirical results on ownership concentration. Section V shows the results on Basel III and the impact of the financial crisis. Finally, the final remarks are offered in Section VI. II. Theoretical Background and Hypothesis Development A. Ownership concentration and bank risk-taking 2 For related studies in this area, please see Saunders, Strock, and Travlos (1990), Anderson and Fraser (2000), Brewer and Saidenberg (1996), Chen, Steiner, and Whyte (1998), Demsetz, Saidenberg, and Strahan (1996), Demsetz and Strahan (1997), Knopf and Teall (1996), Cebenoyan, Cooperman, and Register (1996). 7

There are two opposing hypotheses regarding the impact of concentrated ownership on risk-taking. First, as in any limited liability firm, diversified owners have incentives to increase bank risk after collecting funds from bondholders and depositors (Galai and Masulis, 1976; Esty, 1998). However, managers with bank-specific human capital skills and private benefits of control will tend to advocate for less risk-taking than stockholders without those skills and benefits (Jensen and Meckling, 1976; Demsetz and Lehn, 1985; Kane, 1985). Concentrated ownership can overcome managers tendency for risk avoidance. From this perspective, banks with a more concentrated ownership structure tend to take more risk. By contrast, as the monitoring efforts exerted by large shareholders increases, managerial initiatives to pursue new risky investment opportunities decreases (Burkart et al, 1997). This argument implies less risktaking by managers when concentration of ownership is high. Investor protection laws and banking regulations also play an important part. In addition to empowering equity holders, effective shareholder protection laws reduce the need for the emergence of a large shareholder to mitigate agency problems (Shleifer and Wolfenzon, 2002; John et al., 2000; Castro et al,. 2004). Accordingly, large shareholders should play a less prominent role in shaping corporate behavior in economies with effective shareholder protection laws. Given that shareholder protection in Asia is generally poorer than it is in the U.S. or Europe, the role of large shareholders in increasing risk-taking should be more pronounced in Asia. We hypothesize that ownership concentration is associated with a higher degree of risktaking in Asia. B. Basel III: Capital adequacy standards 8

Throughout the latest global financial crisis, a number of banks struggled to maintain adequate liquidity. In spite of the enormous support from central banks, a number of banks failed or were forced into mergers and acquisitions. The crisis illustrates how quickly and acutely liquidity risks can crystallize and certain sources of funding can evaporate, compounding concerns related to the valuation of assets and capital adequacy (Bank for International Settlements, 2010a). As a consequence, the G20 countries called for a global framework for promoting stronger liquidity buffers at financial institutions, including cross-border institutions. In response, the Basel Committee for Banking Supervision (BCBS) has developed two internationally consistent regulatory standards for liquidity risk supervision. These standards establish minimum levels of liquidity. National authorities are free to adopt arrangements that require more stringent levels of minimum liquidity. The first standard is represented by the liquidity coverage ratio (LCR), which identifies the amount of unencumbered, high quality liquid assets an institution holds that can be used to offset the net cash outflows it would encounter under an acute short-term stress scenario specified by supervisors. The specified scenario entails both institution-specific and systemic shocks. The scenario entails a significant stress, although not worse-case scenario that includes adverse circumstances such as a significant downgrade of the institution s public credit rating, a partial loss of coverage, and increases in derivative collateral calls etc. This standard requires that: (Stock of high-quality liquid assets/ Total net cash outflows over the next 30 calendar days) 100% The second capital adequacy standard is represented by the net stable funding ratio (NSFR). This standard requires a minimum amount of stable sources of funding at a bank relative to the liquidity profiles of the assets, as well as the potential for contingent liquidity 9

needs arising from off-balance sheet commitments, over a one-year horizon. The NSFR aims at limiting over-reliance on short-term wholesale funding during times of buoyant market liquidity and encourage better assessment of liquidity across all on- and off-balance sheet items. This standard requires that: (Available amount of stable funding/ Required amount of stable funding) > 100% We concentrate on the second standard, the net stable funding ratio (NSFR), because the data required to compute the first standard are not readily available. 3 We explore whether NSFR has an impact on the extent of risk-taking by banks, using a sample of East Asian banks in the latter half of the last decade. This is particularly interesting because the full impact of Basel III will not be known for several years to come. However, the results of our study can give us insights into the probable effects of Basel III. In other words, we attempt to answer a hypothetical, albeit particularly intriguing question: what would have been the impact of Basel III on risk-taking if it had been enforced in our sample period? One hypothesis is that NSFR should not have any significant impact on bank risk-taking because Basel III has not been implemented yet. Thus, NSFR has not been given any particular importance by banks. On the contrary, The Basel Committee for Banking Supervision (BCBS) selected NSFR as one of the new capital adequacy standards because they strongly believed that it should be relevant as a significant determinant of bank risk-taking. If the BCBS is correct, there should be a significant association between NSFR and the extent of risk-taking. In 3 Our annual data do not allow us to compute the liquidity coverage ratio, which requires data on daily cash flows. As a result, we focus on the net stable funding ratio only. 10

particular, a larger amount of NSFR should be associated with lower risk-taking. Our objective is to offer empirical evidence on these hypotheses. 4 C. Recent literature As an attempt to prevent another crisis from occurring in the future, in December 2010, the Basel Committee on Banking Supervision announced a package of reforms known as Basel III. Since then, a number of recent studies seek to shed light on the likely effects of Basel III. For instance, Yan, Hall, and Turner (2012) perform a long-term cost-benefit analysis of Basel III for the United Kingdom. Their results reveal a significant long-term positive effect on the UK economy. King (2013) estimates the NSFR for banks in 15 countries. His analysis reveals that the representative bank in 10 countries appears to have an NSFR below the minimum threshold at year-end 2009. Strategies to increase the NSFR are estimated to reduce net interest margins (NIM) by 70-88 basis points. Chalermchatvichien, Jumreornvong, Jiraporn, and Singh (2013) explore the effect of ownership concentration on the implied NSFR ratio and document a U- shaped association. In particular, at lower levels of ownership, an increase in ownership concentration leads to more stable funding. Beyond a certain threshold, however, any further rise in ownership concentration decreases funding stability. III. Sample and Data Description A. Sample construction 4 In addition to the two liquidity ratios discussed above, Basel III will require banks to hold 4.5% of common equity (up from 2% in Basel II) and 6% of Tier I capital (up from 4% in Basel II) of risk-weighted assets (RWA). Basel III also introduces additional capital buffers, (i) a mandatory capital conservation buffer of 2.5% and (ii) a discretionary countercyclical buffer, which allows national regulators to require up to another 2.5% of capital during periods of high credit growth (BIS, 2010). 11

Our data on bank characteristics are derived from several sources, including annual reports and Bloomberg, and the Stock Exchange of Thailand (for Thai banks). We also obtain data from Bank Scope for items such as equity volatility and the cash flow rights of major shareholders. Our calculation of the net stable funding ratio (NSFR) is based on the quantitative impact study conducted by the Basel Committee on Banking Supervision (BCBS). The total sample consists of 68 banks from 11 East Asian countries from the period 2005-2009. 5 The banks in the sample are selected from the top ten largest active commercial banks in each country. The countries covered are Thailand, Hong Kong, Singapore, Indonesia, Malaysia, South Korea, China, Japan, Sri Lanka, India and the Phillipines. 6 Table 1 shows the sample distribution by country. We concentrate on East Asian banks for a number of reasons. First, because the scope of the paper involves understanding the potential effect of Basel III, it is important to consider countries that have implemented Basel II. The East Asian nations have one of the best records, second only to the European Union, in implementing Basel II, both in terms of the adoption of the standards, as well as, the degree of regulatory convergence across nations. However, unlike European Union members, East Asian nations are free from the effects of the European debt crisis. Second, the recent financial crisis originated in the US, prompting a large number of researchers to examine its impact on the US economy in general and the US banking industry in particular (Chari et al. 2008; Gorton and Metrick 2009; Ivashina and Scharfstein 2009). Far less 5 We finish our sample period in 2009 because it is the last year before the announcement of Basel III. Certain data items are not available electronically and have to be collected manually. To keep the sample size reasonable for manual data collection, we choose the five-year period from 2005-2009. This period is also interesting in that it straddles the financial crisis of 2008. 6 By 2011, banks in all of these countries are in compliance with Basel II. 12

coverage has been given to non-us banks. We hope to fill this gap in the literature. Furthermore, in the past decade, East Asia in general and China in particular have experienced phenomenal economic growth, often outpacing the economic growth in the West. Given their increasingly critical role in the global economy, an analysis of East Asian banking institutions should advance our understanding of the effect of regulatory and market forces in emerging economies on global financial markets. Further, owing to its distinctive governance and ownership structure, East Asia offers an interesting setting to examine the role of ownership structure on the risk-taking behavior of banking institutions. B. Variable description Following Laeven and Levine (2009), Z-score is computed as the return on assets (ROA) plus the capital asset ratio divided by the standard deviation of asset returns. The z-score measures the distance from insolvency (Roy, 1952). Insolvency is defined as a state where losses surmount equity. Thus, the Z-score represents the number of standard deviations that a bank s ROA has to drop below its expected value before equity is depleted. Z = (ROA+CAR)/σ (ROA) where ROA = the return on assets CAR = the capital-asset ratio (equity/assets) σ (ROA) = the standard deviation of ROA 7 A higher Z-score indicates that the bank is more stable. As the Z-score is highly skewed, we use the natural logarithm of the Z-score, which is normally distributed. For brevity, we use 7 We calculate the standard deviation or ROA over the sample period. 13

the label Z-score when referring to the natural logarithm of the Z-score in the rest of the paper. Equity volatility equals the annualized volatility of weekly equity returns, also used by Saunders et al (1990), Esty (1998), and Laeven and Levine (2009). One critical advantage of the volatility of equity returns is that it is based on market, rather than accounting data. We control for a number of bank-specific and country-specific factors. As in Laeven and Levine (2009), ownership concentration is measured as the cash flow rights of the largest shareholder. We also include a number of control variables, similar to Laeven and Levine (2009). Revenue growth is annual growth rate in total revenues (EBIT) in the past one year. Loan loss provision is the ratio of the amount of loan loss provision to net interest revenue. Tobin s Q is the market capitalization plus book value of liabilities divided by the book value of total assets. Deposits represent the percentage of the bank s deposits to the total deposits in each country. Most of the bank s funding structure is in deposit s term. As a consequence, the percentage of the deposits also reflects the size of the bank. We control for the country s economic size by including the natural logarithm of the total GDP of each country. Economic development is accounted for by including the natural logarithm of GDP per capita for each country. Finally, we include year dummies to account for possible variation over time. C. Summary statistics Table 2 displays the descriptive statistics. The average Z-score is 3.5320 (median 3.5433), with a standard deviation of 0.6539. 8 Equity volatility averages 0.3991 (median 0.3735). The average ownership concentration is 0.3168 (median 0.2375), higher than the average in 14

Leaven and Levine (2009), implying that bank ownership in Asia is more concentrated. The average loan loss provision is 0.1670 (median 0.1361), lower than the average in Leaven and Levine (2009). Asian banks seem to have lower loan loss provision on average. Tobin s Q averages 1.0363 (median 1.0148). The average revenue growth is 0.2246 (median 0.1082), higher than the average growth in Leaven and Levine (2009). Asian banks grow at a faster speed than western banks. Deposits average 0.1640 (median 0.1109). Finally, the average net stable funding ratio (NSFR) is 3.5280 (median 1.9668). The correlation matrix is displayed in Table 3. Ownership concentration shows a significantly negative correlation with the Z-score, indicating that banks with more concentrated ownership experience higher risk-taking. The correlation coefficient between ownership concentration and equity volatility is significantly positive, again implying greater risk-taking as ownership concentration rises. The net stable funding ratio (NSFR) is positively correlated with the Z-score, suggesting that more capital stability diminishes risk-taking. The correlation between NSFR and equity volatility, however, is not statistically significant. The results from the correlation matrix have to be interpreted with caution because they do not take into account several factors that influence risk-taking. We control for these factors in the regression analysis. IV. Empirical Results: Ownership Concentration A. Regression analysis The framework for our regression analysis can be expressed as follows: 15

Ln Z = a + b 1 (Ownership Concentration) + b 2 (Loan Loss Provision) + b 3 (Tobin s Q) + b 4 (Revenue Growth) + b 5 (Deposits) + b 6 (Log(GDP)) + b 7 (GDP per capita) + Year Dummies Table 4 shows the regression results. Model 1 has the Z-score as the dependent variable. The standard errors are adjusted to account for clustering at the bank level. Ownership concentration carries a significantly negative coefficient. Banks with more concentrated ownership engage in more risk-taking. To ascertain the economic impact, we calculate the standardized coefficient of ownership concentration, which turns out to be 0.33. Thus, a rise by one standard deviation in ownership concentration results in a decrease in the Z-score by 0.2157 (=0.33*0.6539) or 6.11% of the average Z-score. 9 In Model 2, we replace the Z-score with equity volatility as the dependent variable. The coefficient of ownership concentration is significantly positive. Banks with higher ownership concentration experience more risk-taking, in agreement with the result based on the Z-score. A rise in one standard deviation in ownership concentration is associated with an increase in equity volatility by 0.18 or 8.12%. Our results are consistent, regardless of which variable is used to proxy for the degree of risk-taking. The results are in line with the following view: (1) Owners possess a tendency for more risk-taking than managers and debt holders (Galai and Masulis, 1976, Demsetz and Lehn, 1985) and (2) Large owners with considerable cash flow rights are more motivated and have more power to increase bank risk-taking than small shareholders (Jensen and Meckling, 1976, John et al., 2008). Finally, our results also confirm those in Laeven 7 The standard deviation of the Z-score is 0.6539. This number when multiplied by the standardized coefficient yields 0.2157. 16

and Levine (2009), whose sample is primarily from the late 1990 s. Even in a later period after all the governance reforms and a wave of mergers and acquisitions in the banking industry, the association between ownership concentration and risk-taking continues to hold. Our study thus represents a timely update on Laeven and Levine (2009) B. Possible endogeneity It is conceivable that ownership concentration and bank risk-taking might be endogenously determined. There are two types of endogeneity. First, ownership concentration and risk-taking may be related to a third unobservable bank characteristic. If this is the case, then the association between ownership concentration and risk-taking might be spurious. We address this potential problem by exploiting the insight from Altonji, Elder, and Taber (2005). Their study suggests that selection on observables can be used to estimate the potential bias generated by unobservables, i.e. how much stronger selection on unobservables, relative to selection on observables, would have to be to explain away the full estimated effect. 10 This potential bias can be estimated this way, consider two regressions: one with a restricted set of control variables, and one with a full set of controls. Denote the estimated coefficient for the variable of interest from the first regression β R (where R stands for Restricted) and the estimated coefficient from the second regression β F (where F stands for Full). Then, the ratio can be computed as β F /( β R - β F ). 11 The intuition behind the formula is straightforward. First, consider the denominator ( β R - β F ). If the difference between β R and β F is small, the estimate is only slightly affected by selection of observables. As a result, selection on unobservables would 8 Altonji et al. (2005) consider the situation where the explanatory variable is a binary variable. Bellows and Miguel (2006) develop an analogous test for the case where the variable of interest is continuous. 9 See Miguel and Bellows (2006) for the formal derivation. As well, see Altonji et al. (2005) for details of the underlying assumptions. 17

have to be much stronger (relative to observables) to explain away the entire effect. Then, consider the intuition behind β F in the numerator. If β F is large, selection on unobservables would have to be much stronger to explain away the effect (due to selection on observables). Thus, the higher the ratio, the less likely the estimate is influenced by unobservables. We apply this method to our sample and estimate two regressions: one with no controls and another with a full set of control variables. In Table 4, Model 5 and Model 6 show the regression results with no control variables. We use the coefficients of ownership concentration from Model 1 (full model) and Model 5 (restricted model) to estimate the ratio when the Z-score is the dependent variable. The ratio turns out to be 258.82. Consequently, to attribute the entire OLS estimate to selection effects, selection on unobservables would have to be at least 258.82 times stronger than selection on observables. It appears very unlikely that the estimated effect of ownership concentration on bank risk-taking is primarily driven by unobservables. This provides a certain degree of comfort that our results are not spurious due to possibly omitted variables. Similarly, we use the coefficients of ownership concentration from Model 2 and Model 6 to estimate the ratio when equity volatility is the dependent variable. The calculation yields a ratio of 3.11. Again, it is unlikely that our results are mainly attributed to unobservable bank characteristics. 12 Second, there might be endogeneity due to reverse causality. Risk-taking may lead to a particular ownership structure. If this is the case, then the direction of causality may be reversed. 12 For comparison, Altonji et al (2005) report the ratio to be between 2 and 4. They argue that with those numbers the endogeneity bias due to possibly omitted variables is unlikely. Similarly, Nunn and Wantchekon (2011), with their reported ratio lying between 3 and 11.5 do not consider endogeneity due to omitted variables to be of material concern. Given the test results of these previous studies, our conclusion is unlikely affected by the endogeneity bias due to omitted variables. 18

We first address this possible problem by replacing ownership concentration in each given year for each given bank with ownership concentration in the earliest year in the sample. The logic is that ownership concentration in the earliest year could not have resulted from risk-taking in the subsequent years. Thus, if we continue to find a significant association, the direction of causality is much more likely to run from ownership concentration to risk-taking than vice versa. The results are shown in Model 3 and Model 4 in Table 5. The coefficient of ownership concentration in the earliest years is significantly negative in Model 3 and significantly positive in Model 4, consistent with the earlier results. Therefore, reverse causality is unlikely. Causality seems to run from ownership concentration to risk-taking than the other way around. Furthermore, we follow Laeven and Levine (2009) and utilize an instrumental variable to address endogeneity. Like Laeven and Levine (2009), we use the average cash flow rights of other banks in the country as our instrumental variable and execute a two-stage least squares (2SLS) regression analysis. 13 The results remain consistent. Moreover, our Hausman s tests reveal that the OLS and 2SLS estimates are not statistically significant, implying that endogeneity does not pose a problem in our sample. V. Empirical Results: Basel III, Economic Development, and the Financial Crisis A. The impact of capital stability Table 5 shows the regression results, where we include the net stable funding ratio (NSFR) as an independent variable. A higher value of NSFR signifies a higher degree of capital stability. Essentially, we examine how this capital standard rule would have affected bank risk- 10 The assumption behind the selection of this instrumental is that innovations in the risk of one bank will not influence the cash flow rights of other banks (Leaven and Levine, 2009). 19

taking if it had been enforced in our sample period. In Model 1, the dependent variable is the Z- score. NSFR enters with a significantly positive coefficient, suggesting that more stable funding reduces the extent of risk-taking. This Basel III capital stability requirement, if implemented in our sample period, probably would have promoted more cautious investments and diminished risk-taking by banks. With respect to economic significance, a rise in NSFR by one standard deviation results in an improvement in the Z-score by 0.1896 or 5.37%. In Model 2, we turn our attention to equity volatility as the dependent variable. NSFR does not enter with a significant coefficient. Thus, if implemented in the period 2005-2009, NSFR would have affected the bank s Z-score, but not its equity volatility. In terms of explanatory power, our regression models explain 31.79% and 65.17% of the variation in the Z-score and equity volatility respectively. It is unlikely that endogneity unduly influences our results because NSFR was not officially effective in our sample period and thus did not receive much attention. In any event, as a precaution, we substitute the value of NSFR in the earliest year for the actual value in any given year and re-estimate the regressions. The idea is that NSFR in the earliest year could not have resulted from risk-taking in any subsequent years. If we find that early NSFR influences subsequently risk-taking, then causality is much more likely to run from NSFR to risk-taking. This is precisely what we find in Model 3 and Model 4 of Table 5. The coefficients of NSFR are qualitatively similar to those in Model 1 and Model 2. We also test whether our results are influenced by possibly omitted variables. One method is to examine changes in the variables rather than the levels. Using changes removes the impact of unobservable bank characteristics that remain constant over time. The results are displayed in Table 6. In Model 1, the dependent variable is changes in the Z-score from one year 20

to another. The coefficient of changes in NSFR is positive and highly significant. An increase in capital stability is associated with a reduction in risk, confirming the previous results. In Model 2, the dependent variable is equity volatility. The coefficient of changes in NSFR is not significant. In summary, an improvement in capital stability enhances the bank s Z-score, although it does not appear to affect equity volatility. The effect on the bank s Zscore is both statistically and economically significant. This conclusion seems to be robust to the simultaneity bias as well as the omitted-variable bias. B: Possible interaction effects between regulation and ownership Prior literature argues that the relationship between bank risk and ownership structure should vary with regulations (Shleifer and Vishny, 1986; Buser et al., 1981; and John et al., 2000). Laeven and Levine (2009) offer empirical results on this subject. In particular, they examine the interaction effects on bank risk-taking among ownership structure, deposit insurance, and capital adequacy requirements. The results show that the same bank regulation has dissimilar effects on bank risk-taking, contingent upon the comparative power of shareholders in the governance structure of each bank. Mindful of these possible interaction effects, we explore whether the net stable funding ratio, proposed as part of Basel III, interacts with ownership structure in influencing risk-taking. We construct an interaction variable between NSFR and ownership concentration and include it in the regression analysis. The results are displayed in Table 7. The coefficient of the interaction variable is not significant. Unlike Laeven and Levine (2009), we do not find evidence of an 21

interaction effect. The difference in our results might be attributed to the fact that Leaven and Levine (2009) examine regulations that were already in effect such as deposit insurance and capital regulations. Our study, on the contrary, hypothetically investigates the impact of Basel III, which is not yet effective. Thus, the lack of a significant interaction may be due to the fact that bank shareholders could not have adjusted the risk-taking behavior to a regulation that was not actually effective yet. C. The role of economic development It is conceivable that the impact of the Basel III capital stability standard is not uniform across countries with disparate levels of economic development. Perhaps, in developed countries, financial and accounting information is more reliable, due to more stringent disclosure requirements, hence more transparency. Enforcement of regulation based on more accurate information may be more effective in these developed countries. On the contrary, it is possible that developed countries already have in place stringent domestic regulations that control bank risk-taking. Thus, any additional regulation from outside the country may have only a slight marginal effect. On the contrary, in less developed countries where domestic regulations are not as strict, the impact of Basel III may be more pronounced. We execute additional tests to ascertain whether the impact of capital stability varies with the degree of economic development. GDP per capita is widely used as an indicator of economic development. We multiply the net stable funding ratio (NSFR) with GDP per capita and include the interaction term in the regressions. The results are shown in Table 8. Model 1 uses the Z- score as the dependent variable. The interaction variable carries a positive and significant coefficient. The capital stability standard has a stronger effect on risk-taking where economic 22

development is higher. In Model 2, equity volatility is the dependent variable. The coefficient of the interaction term is negative and significant. The impact of capital stability reduces risk further with higher economic development. It does appear that the capital stability standard would not affect risk-taking the same way across countries. It would be more effective in more developed economies. D. The role of the financial crisis of 2008 The financial crisis that began in the fall of 2008 threw economies around the world into severe recessions. The cause of this crisis was the credit boom that reached its peak in mid-2007, followed by the meltdown of sub-prime mortgages and all types of securitized products. This meltdown, in turn, raised concerns about the solvency and liquidity of financial institutions, resulting in a banking panic (Ivashina and Scharfstein, 2009). Due to its tremendous impact on banks and financial markets around the world, we explore how this banking panic affects the relationships among risk-taking, ownership structure, and regulation. Our sample period spans five years from 2005 to 2009. The banking crisis started in 2008 and continued into 2009 and later. We thus classify 2008 and 2009 as the crisis period. We construct a dichotomous variable equal to one for 2008 and 2009 and zero otherwise. This variable is labeled Financial Crisis. Then, we interact this variable with ownership concentration. The coefficient of this variable would reveal the impact of ownership concentration on risk-taking before versus after the financial crisis. The regression results are shown in Table 9. In Model 1, we use the Z-score as the dependent variable. The interaction term does not carry a significant coefficient. Model 2 has equity volatility as the dependent variable. Again, the interaction variable does not enter with a significant coefficient. Consequently, it does 23

not appear that the financial crisis changes the way in which ownership concentration promotes risk-taking. It is interesting to note that the coefficient of the financial crisis variable in Model 2 is significantly positive, implying that equity volatility rises sharply in the crisis period. The coefficient is 0.1654, meaning that, in the crisis period, equity volatility increases by 0.1654 or 41.44% of the average volatility over the entire sample period. We also investigate how the banking panic would have interacted with regulation. We create an interaction variable between Financial Crisis and NSFR. The coefficient of this variable would show whether capital stability affects risk-taking differently before and after the financial crisis. The results are displayed in Table 9. Model 3 uses the Z-score as the dependent variable, while Model 4 uses equity volatility. In both Model 3 and Model 4, the coefficient of the interaction term is not significant. The financial crisis does not influence the relationship between capital stability and risk-taking. Again, it is noteworthy that the coefficient of the financial crisis variable is positive and significant in Model 4. Equity volatility jumps up in the crisis period by 0.1854 or 46.45%. In summary, the empirical evidence shows that ownership concentration and regulation influence risk-taking the same way before and after the crisis. There is no evidence that the crisis fundamentally changes the relationships in a significant manner. Nevertheless, we document a profound impact of the financial crisis on the volatility of equity returns in the crisis period. Although the crisis originated in the U.S, its effects were felt all over the world, including the Asian banks included in our sample. E. Exploring non-linearity 24

Prior literature shows that the impact of ownership structure may not be linear. We explore this possibility by including the quadratic term of ownership concentration in the regressions. There is no evidence of a non-linear relationship. We also split ownership concentration into ranges and assign a dummy variable for each range but do not find significant results. Likewise, it is conceivable that capital stability may exhibit a non-linear relationship with risk-taking. We thus include the square of NSFR to control for possible non-linearity. The results, however, are not significant. F. Implications of the results Our findings have important implications for regulators and policy makers. The findings of our study strongly reinforce those in Laeven and Levine (2009). Because ownership structure influences bank risk-taking, regulations should take into account changes in policies toward bank ownership, such as allowing private equity groups to invest in banks or changing limits on ownership concentration. Capital requirements and liquidity regulations make banks less prone to crises, which impose substantial losses in terms of forgone economic output. Instability in the banking system could lead to severe disruptions in lending and other financial intermediation services, resulting in significant declines in economic activities and ultimately the GDP. For instance, it is estimated that banking crises, on average, cause reductions in GDP by 9-10% (BIS, 2010b). Other studies in the literature also attempt to estimate the cost of a banking crisis, usually measuring changes in GDP from the peak of the business cycle prior to the crisis to a subsequent trough point for GDP. The end point is when GDP remains on a new sustainable path. The estimated temporary GDP losses (relative to the pre-crisis level) range from 6-14%, whereas the permanent losses in 25

GDP are in the range of 2-9.5% (Bordo et al, 2001; Demiguc-Kunt et al., 2000; Hutchison and Noy, 2002; Laeven and Valencia, 2008; Haugh et al, 2009; Cecchetti et al, 2009; Hoggarth et al., 2002; Cerra and Saxena, 2008; Roger et al., 2010; Furceri and Zdzienicka, 2010; Barrell and Davis Liadze, 2010; Boyd et al., 2005; and Haldane, 2010). The results of our study lend support to the capital stability requirement proposed in Basel III. In particular, we find that more stable capital is associated with less risk-taking, as measured by the bank s Z-score. Less risk-taking by banks in turn reduces the probability of a banking crisis. Thus, our results contribute to the area of the literature that examines the extent to which regulatory requirements help prevent a banking crisis (Barrel et al., 2009; Kato et al., 2010; Wong et al., 2010; Tarashev and Zhu, 2008; Miles et al., 2011; Gauthier et al., 2010). VI. Final Remarks This study explores the relationships among risk-taking, ownership structure, and regulation, using a sample of East Asian banks. The results suggest that higher ownership concentration promotes risk-taking. Banks where ownership is more concentrated exhibit a stronger propensity to take risk. These results are consistent with prior literature and represent a timely update, particularly in light of the recent banking crisis. In addition, we shed light on the likely impact of the recently-proposed capital adequacy requirements or Basel III, scheduled to be implemented in phases from 2013 to 2018. We conduct an experiment, where we explore what the impact of Basel III would have been if it had been implemented in our sample period from 2005 to 2009. The results reveal that, by requiring more capital stability, Basel III would have improved the bank s Z-score. The actual impact of Basel III will not be known for several 26

years. But, the results in this study provide an interesting glimpse into the future as to the probable effects of Basel III. Moreover, we test how ownership concentration and regulation may interact with each other. The results, however, do not suggest any significant interaction effects between ownership and regulation on risk-taking. The lack of a significant interaction may be attributed to the fact that Basel III is actually not yet effective. Hence, bank shareholders could not have modified their risk-taking strategies based on a regulation not yet in effect. Future research might explore how bank shareholders adjust their risk-taking behavior when Basel III is already in effect in 2013 and later. Economic development plays an important role. The impact of capital stability on risk-taking is much more pronounced in countries with better economic development. Finally, we examine the impact of the recent banking panic and find that the equity volatility of banks rises as much as 46% in the crisis period, although the crisis does not change the fundamental relationships among risk-taking, ownership structure, and regulation. The results of this study offer helpful insights into the likely effects of Basel III and the recent financial crisis and therefore should be especially useful to a wide range of audiences, including policy makers, regulators, bank executives, financial practitioners, as well as researchers and economists alike. 27

References Agrawal, A. and G. Mendelker, 1987, Managerial incentives and corporate investment and financing decisions, Journal of Finance 42, 823-838. Allen, W.A., K.K. Chan, A. Milne, and S. Thomas, Basel III: Is the cure worse than the disease, Working paper, City University London. Anderson, R.C. and D.R. Fraser, 2000, Corporate control, bank risk taking, and the health of the banking industry, Journal of Banking and Finance 24, 1383-1398. Altonji, J.G., T.E. Elder, and C.R.Taber, 2005, Selection on observed and unobserved variables: Assessing the effectiveness of Catholic schools, Journal of Political Economy 113, 151-184. Bank for International Settlements (BIS), 2010a, Basel III: International framework for liquidity risk measurement, standards and monitoring, Basel Committee on Banking Supervision. Bank for International Settlements (BIS), 2010b, An assessment of the long-term impact of stronger capital and liquidity requirements, Basel Committee on Banking Supervision. Barrell, R., E.P. Davis, T. Fic, D.H.S. Kirby, and I. Liadze, 2009, Optimal regulation of bank capital and liquidity: How to calibrate new international standards, Occasional Paper Series No. 38: Financial Services Authority. Bernanke, B., 1983, Nonmonetary effects of the financial crisis in the propagation of the Great Depression, American Economic Review 73, 257-276. Bordo, M., B. Eichengreen, D. Klingebiel, and M.S. Martinez-Peria, 2001, Is the crisis problem growing more severe? Economic Policy 16, 51-82. Boyd, J.H., S. Kwak, and B. Smith, 2005, The real output losses associated with modern banking crises, Journal of Money, Credit, and Banking 37, 977-999. Brewer, E. III and M.R. Saidenberg, 1996, Franchise value, ownership structure, and risk at savings institutions, Federal Reserve Bank of New York, Research Paper 9632. Brunnermeier, M., 2009, Deciphering the liquidity and credit crunch 2007-2008, Journal of Economic Perspectives 23, 77-100. Burkart, M., K. Gromb, and F. Panunzi, 1997, Large shareholders, monitoring, and fiduciary duty, Quarterly Journal of Economics 112, 693-728. 28

Buser, S., A. Chen, and E. Kane, 1991, The role of demandable debt in structuring optimal banking arrangements, American Economic Review 81, 497-513. Calomiris, C. and J. Mason, 1997, Contagion and bank failures during the Great Depression: The Chicago Banking Panic of June 1932, American Economic Review 87, 863-884. Calomiris, C. and J. Mason, 2003a, Consequences of U.S. bank distress during the Depression, American Economic Review 93, 937-947. Calomiris, C. and J. Mason, 2003b, Fundamentals, panics, and bank distress during the Depression, American Economic Review 93, 1615-1647. Castro, R., G. Clementi, and G. MacDonald, 2004, Investor protection, optimal incentives, and economic growth, Quarterly Journal of Economics 119, 1131-1175. Cebenoyan, A.S., E.S. Cooperman, and C.A. Register, 1999, Ownership structure, charter value, and risk-taking behavior of thrifts, Financial Magement 28, 43-60. Cecchetti, S.G., M. Kohler, and C. Upper, 2009, Financial crises and economic activity, Working paper No. 15379, National Bureau of Economic Research. Cerra, V. and S.C. Saxena, 2008, Growth dynamics: The myth of economic recovery, American Economic Review 98, 439-457. Charlermchatvichien, P., S. Jumreornvong, and P. Jiraporn, 2013, The effect of bank ownership concentration on capital adequacy, liquidity, and capital stability, Journal of Financial Services Research, forthcoming. Chari, V.V., L.J. Christiano, and P.J. Kehoe, 2008, Facts and myths about the financial crisis of 2008, Working paper, Federal Reserve Bank of Minneapolis. Chen, C.R., T.L. Steiner, and A.M. Whyte, 1998, Risk-taking behavior and management ownership in depository institutions, Journal of Financial Research 21, 1-16. Demsetz, H. and K. Lehn, 1985, The structure of corporate ownership: Causes and consequences, Journal of Political Economy 93, 1155-1177. Demsetz, R.S. and P.E. Strahan, 1997, Diversification, size, and risk at bank holding companies, Journal of Money, Credit, and Banking 29, 300-313. Demsetz, R.S., M.R. Saidenberg and P.E. Strahan, 1996, Banks with something to lose: the disciplinary role of franchise value, Federal Reserve Bank of New York Economic Policy Review (October), 1-14. 29

Domirguc-Kunt, A., E. Detragiache, and P. Gupta, 2000, Inside the crisis: An empirical analysis of banking systems in distress, Policy Research Working Paper Series 2431, The World Bank. Esty, B., 1998, The impact of contingent liability on commercial bank risk taking, Journal of Financial Economics 47, 189-218. Furceli, D., and A. Zdzienicka, 2010, Banking crises and short and medium term output losses in developing countries: The role of structural and policy variables, MPRA Paper No. 22078. University Library of Munich. Galai, D. and R. Masulis, 1976, The option pricing model and the risk factor of stock, Journal of Financial Economics 3, 53-81. Gauthier, C., A. Lehar, and M. Souissi, 2010, Macroprudential regulation and systemic capital requirements, Working paper No. 10-4: Bank of Canada. Gorton, G., 2009, Slapped in the face by the invisible hand: Banking and the panic of 2007, Working paper, Yale University. Gorton, G. and A. Metrick, 2009, Securitized banking and the run on repo, Working paper, Yale University. Haldane, A., 2010, The 100 billion dollar question, Accepted Paper Series No.22, Bank of England. Haugh, D., Ollivaud, P., and D. Turner, 2009, The macroeconomic consequences of banking crises in OECD countries, OECD Economics Department Working Paper No.683, OECD Publishing. Hoggarth, G. R. Reis, and V. Saporta, 2002, Costs of banking system instability: Some empirical evidence, Journal of Banking and Finance 26, 825-855. Hutchison, M., and I. Noy, 2002, How bad are twins? Output costs of currency and banking crises, Pacific Basin Working Paper Series 02-02: Federal Reserve Bank of San Francisco. Ivashina, V. and D. Scharfstein, 2009, Bank lending during the financial crisis of 2008, Working paper, Harvard University. Jensen M. and W. Meckling, 1976, Theory of the firm: Managerial behavior and agency costs, and ownership structure, Journal of Financial Economics 3, 305-360. 30

John, K., A. Saunders, and A. Senbet, 2000, A theory of bank regulation and management compensation, Review of Financial Studies 13, 95-125. John, K., L. Litov, and B. Yeung, 2008, Corporate governance and managerial risk taking: Theory and evidence, Journal of Finance 63, 167-1728. Kane, E., 1985, The Gathering Crisis in Federal Deposit Insurance, Cambridge, Mass: MIT Press. Kato, R., S. Kobayashi, and Y. Saita, 2010, Calibrating the level of capital: The way we see it, Working Paper Series No. 10-E-6, Bank of Japan. Keely, M., 1990, Deposit insurance, risk, and market power in banking, American Economic Review 80, 1183-1200. King, M.R., 2013, The Basel III net stable funding ratio and bank net interest margins, Journal of Banking and Finance 37, 4144-4156. Knoft, J.D. and J.L. Teall, 1996, Risk-taking behavior in the U.S. Thrift industry: Ownership structure and regulatory changes, Journal of Banking and Finance 20, 1329-1350. Laeven, L. and R. Levine, 2009, Bank governance, regulation, and risk taking, Journal of Financial Economics 93, 259-275. Laeven, L., and F. Valencia, 2008, Systemic banking crises: A new database, IMF Working Paper No.08/224, International Monetary Fund. Miguel, E. and J. Bellows, 2006, War and institutions: Evidence from Sierra Leone, American Economic Review 96 (2), 394-399. Miles, D., J. Yang, and G. Marcheggiano, 2011, Optimal bank capital, CEPR Discussion Paper No.8333, London, Center for Economic Policy Research. Nunn N. and Wantchekon L., 2011, The slave trade and the origins of mistrust in Africa. American Economic Review 101:3221 3252. Ojo, M., 2010, Basel III and responding to the recent financial crisis: Progress made by the Basel Committee in relation to the need for increased bank capital and increased quality of loss absorbing capital, Working paper, Oxford Brookes University. Roger, W., I. Szekely, and A.A. Turrini, 2010, Banking crises, output loss and fiscal policy, CEPR Discussion Paper No.7815, London, Center for Economic Policy Research. Roy, A.D., 1952, Safety first and the holding of assets, Econometrica 20, 431-449. 31

Saunders, A., E. Strock, and N. Travlos, 1990, Ownership structure, deregulation, and bank risk taking, Journal of Finance 45, 643-654. Shleifer, A. and R. Vishny, 1986, Large shareholders and corporate control, Journal of Political Economy 94, 461-488. Shleifer, A. and D. Wolfenzon, 2002, Investor protection and equity markets, Journal of Financial Economics 66, 3-27. Tarashev, N. and H. Zhu, 2008, Specification and calibration errors in measures of portfolio credit risk: The case of the ASRF model, International Journal of Central Banking 4, 129-173. Went, P., 2010, Basel III Accord: Where do we go from here? Working paper, GARP Research Center. Wong, T., T. Fong, K. Li, and H. Choi, 2010, An assessment of the long-term economic impact of the new regulatory reform on Hong Kong, Working Paper Series, Bank of Hong Kong. Yan, M., M.J.B. Hall, and P. Turner, 2012, A cost-benefit analysis of Basel III: Some evidence from the UK, International Review of Financial Analysis 25, 73-82. 32

Table 1: Sample Distribution by Country Country No. of Banks China 8 Hong Kong 5 India 5 Indonesia 10 Japan 12 Malaysia 3 Phillipines 9 Singapore 2 South Korea 6 Sri Lanka 2 Thailand 6 Total 68 33

Table 2: Summary Statistics Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio f market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. Variable Mean Median S.D. 25 th 75 th Ln (Z) 3.5320 3.5433 0.6539 3.1093 4.0066 Equity Volatility 0.3991 0.3735 0.1411 0.3050 0.4768 Ownership Concentration 0.3168 0.2375 0.2443 0.1020 0.5500 Loan Loss Provision 0.1670 0.1361 0.1324 0.0765 0.2168 Tobin s Q 1.0363 1.0148 0.0797 0.9860 1.0638 Revenue Growth 0.2246 0.1082 1.6873-0.1333 0.3479 Deposits 0.1640 0.1109 0.1451 0.0589 0.2306 Log (GDP) 8.8464 8.3047 1.3869 7.5197 10.3358 GDP per Capita 15,161 4,043 14,883 1,844 30,815 Net Stable Funding Ratio (NSFR) 3.5280 1.9668 2.9576 1.2228 6.1225 34

Table 3: Correlation Matrix Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. Variable 1 2 3 4 5 6 7 8 9 10 1 Ln (Z) 1.00 2 Equity Volatility 3 Ownership Concentration 4 Loan Loss Provision -0.22*** 1.00-0.33*** 0.24*** 1.00-0.37*** 0.41*** 0.15** 1.00 5 Tobin s Q -0.02-0.10 0.30*** -0.10 1.00 6 Revenue Growth -0.03 0.02-0.04-0.14*** 0.06 1.00 7 Deposits 0.11* -0.25 0.09-0.17*** 0.17*** 0.04 1.00 8 Log (GDP) -0.01-0.09-0.36*** -0.00-0.24*** -0.03-0.00 1.00 9 GDP per Capita 10 Net Stable Funding Ratio (NSFR) 0.03-0.07-0.38*** -0.01-0.24*** -0.04-0.04 0.96*** 1.00 0.24*** 0.04-0.01 0.01-0.15** -0.06-0.18*** 0.11* 0.27*** 1.00 35

Table 4: The Impact of Ownership Concentration on Bank Risk-taking Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 36

(1) (2) (3) (4) (5) (6) OLS OLS OLS OLS OLS OLS (clustered by (clustered by (clustered by (clustered by firm) firm) firm) firm) (clustered by firm) (clustered by firm) Dependent Variable Ln (Z) Equity Volatility Ln (Z) Equity Volatility Ln (Z) Equity Volatility Constant 5.6575*** 0.4857** 6.142*** 0.6561*** 3.8112*** 0.3558*** (3.67) (2.45) (3.86) (3.22) (29.91) (22.88) Ownership -0.8800*** 0.1033*** - - -0.8834*** 0.1365*** Concentration (-2.71) (2.91) (-2.96) (3.46) Ownership - - -0.8554** 0.1114*** - - Concentration (-2.54) (3.30) (Earliest Year) Loan Loss -1.684*** 0.2586*** -1.6813*** 0.2546*** - - Provision (-3.39) (4.76) (-3.40) (4.55) Tobin s Q 0.0204 0.0803-0.0496 0.0804 - - (0.02) (0.88) (-0.06) (0.89) Revenue -0.0417* 0.0062*** -0.0400* 0.0060*** - - Growth (-1.87) (4.70) (-1.83) (4.97) Deposits 0.4493-0.2194*** 0.4370-0.2192*** - - (0.88) (-3.96) (0.82) (-4.15) Log (GDP) -0.2208-0.0401* -0.2514-0.0362* - - (-1.23) (-1.82) (-1.36) (-1.71) GDP per Capita 0.0000 0.0000 0.0000 0.0000 - - (0.89) (1.57) (1.06) (1.46) Year dummies Included Yes Yes Yes Yes No No Adjusted R 2 26.53% 65.16% 26.43% 65.69% 10.89% 5.59% 37

Table 5: The Impact of the Capital Stability on Bank Risk-taking Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 38

(1) (2) (3) (4) OLS OLS OLS OLS (clustered by firm) (clustered by firm) (clustered by firm) (clustered by firm) Dependent Variable Ln (Z) Equity Volatility Ln (z) Equity Volatility Constant 3.5798** 0.7096*** 3.7094* 0.7053*** (2.08) (2.96) (1.92) (3.00) Net Stable 0.0641** -0.0007 - - Funding Ratio (2.52) (-0.24) (NSFR) NSFR - - 0.0527* -0.0004 (Earliest Year) (1.81) (-0.18) Ownership -0.9180*** 0.1037*** -0.9694*** 0.1041*** Concentration (-3.09) (2.88) (-3.23) (2.81) (OC) Loan Loss -1.6726*** 0.2585*** -1.6447*** 0.2582*** Provision (-3.71) (4.74) (-3.67) (4.75) Tobin s Q 0.2620 0.0778 0.2732 0.0780 (0.33) (0.84) (0.33) (0.84) Revenue Growth -0.0394* 0.0062*** -0.0417* 0.0062*** (-1.75) (4.66) (-1.91) (4.69) Deposits 0.5682-0.2206*** 0.5650-0.2204*** (1.14) (-3.80) (1.11) (-3.79) Log (GDP) 0.0262-0.0427 0.0089-0.0422 (0.13) (-1.67) (0.04) (-1.65) GDP per Capita -0.0000 0.0000 0.0000 0.0000 (-0.50) (1.42) (-0.32) (-0.18) Year dummies Included Yes Yes Yes Yes Adjusted R 2 31.79% 65.17% 30.17% 65.17% 39

Table 6: Analysis of Changes in Capital Stability Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 40

(1) (2) OLS OLS Dependent Variable Ln (Z) Equity Volatility Constant -0.0005 0.0540*** (-0.04) (6.62) Net Stable 0.0306*** 0.0016 Funding Ratio (3.36) (0.30) (NSFR) Ownership -0.0253 0.0912 Concentration (-0.12) (0.75) (OC) Loan Loss 0.1354-0.0410 Provision (1.60) (-0.83) Tobin s Q 0.2866* -0.4772*** (1.65) (-4.71) Revenue Growth 0.0138** -0.0007 (2.53) (-0.23) Deposits -1.4433** -0.2291 (-2.44) (-0.66) Log (GDP) 0.2527* -0.0898 (1.95) (-1.19) GDP per Capita -0.0000*** 0.0000*** (-5.06) (3.16) Adjusted R 2 18.65% 11.95% 41

Table 7: The Interaction Effects between Ownership Concentration and Regulation Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 42

(1) (2) OLS OLS (clustered by firm) (clustered by firm) Dependent Variable Ln (z) Equity Volatility Constant 3.5269* 0.7224*** (1.99) (3.05) NSFR OC -0.0525 0.0128 (-0.72) (1.41) Net Stable 0.0816-0.0049 Funding Ratio (1.99)* (-1.15) (NSFR) Ownership 0.7032 0.0515 Concentration (-1.60) (0.83) (OC) Loan Loss -1.6654*** 0.2567*** Provision (-3.77) (4.76) Tobin s Q 0.2040 0.9186 (0.25) (0.94) Revenue Growth -0.0380 0.0058*** (-1.65) (4.12) Deposits 0.5959-0.2273*** (1.15) (-3.90) Log (GDP) 0.0322-0.0442* (0.15) (-1.73) GDP per Capita -0.000 0.000 (-0.56) (1.57) Year dummies Included Yes Yes Adjusted R 2 32.12% 65.60% 43

Table 8: The Impact of Economic Development Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 44

(1) (2) OLS OLS (clustered by firm) (clustered by firm) Dependent Variable Ln (z) Equity Volatility Constant 3.7895** 0.44999* (2.33) (1.97) NSFR GDP per Capita 0.0000* 0.0000* (1.72) (-2.00) Net Stable 0.0218 0.0054 Funding Ratio (0.84) (1.41) (NSFR) Ownership -0.7190** 0.0754* Concentration (-2.23) (1.90) (OC) Loan Loss -1.7530*** 0.2700*** Provision (-4.03) (4.98) Tobin s Q 0.0453 0.1086 (0.06) (1.11) Revenue Growth -0.0352 0.0057*** (-1.51) (4.10) Deposits 0.8041-0.2542*** (1.45) (-4.04) Log (GDP) 0.0080-0.0401 (0.04) (-1.63) GDP per Capita -0.0000 0.0000* (-0.84) (1.87) Year dummies Yes Yes Included Adjusted R 2 33.93% 66.10% 45

Table 9: The Effects of the Financial Crisis Ln (Z) is the natural logarithm of the Z-score. Equity volatility is the annualized volatility of stock returns of the bank. Ownership concentration is the cash flow rights of the largest shareholder of the bank. Loan loss provision is the amount of loan loss provision divided by net interest revenue. Tobin s Q is the ratio of market value of equity plus the book value of liabilities divided by the book value of assets. Revenue growth is the growth in total revenues of the bank over the past year. Deposits are the percentage of the bank s deposits to total deposits in each country. GDP per capita is the natural logarithm of GDP per capita for each country. The Financial Crisis dummy is set to one for 2008 and 2009 and zero otherwise. The net stable funding ratio is calculated as explained in the Appendix. *,**, *** indicate statistical significance at the 10%, 5%, and 1% respectively. 46

(1) (2) (3) (4) OLS OLS OLS OLS (clustered by (clustered by firm) firm) (clustered by firm) (clustered by firm) Dependent Variable Ln (Z) Equity Volatility Ln (z) Equity Volatility Constant 5.6071*** 0.5018*** 3.1667* 0.4940** (3.64) (2.49) (1.92) (2.14) OC Financial Crisis 0.2368 0.0658 - - (1.09) (1.22) NSFR Financial Crisis - - -0.0095 0.0002 (-0.65) (0.05) Financial Crisis -0.0004 0.1654*** 0.1184 0.1854*** (-0.01) (8.07) (1.54) (8.70) NSFR - - 0.0687** -0.0007 (2.63) (-0.33) Ownership -0.9886 0.0775** -0.9380*** 0.1032*** Concentration (-3.03) (2.34) (-3.13) (2.90) (OC) Loan Loss -1.6007 0.2546*** -1.5818*** 0.2586*** Provision (-3.26) (4.83) (-3.53) (4.67) Tobin s Q 0.1770 0.0915 0.4289 0.0900 (0.22) (1.03) (0.55) (0.99) Revenue Growth -0.0347 0.0061*** -0.0340 0.0059*** (-1.67) (4.44) (-1.62) (4.15) Deposits 0.4521-0.2204*** 0.5649-0.2222*** (0.88) (-3.90) (1.12) (-3.75) Log (GDP) -0.2289-0.0401* 0.0310-0.0396 (-1.26) (-1.79) (0.15) (-1.57) GDP per Capita 0.0000 0.000-0.0000 0.0000 (0.93) (1.56) (-0.52) (1.32) Adjusted R 2 25.74% 64.76% 31.04% 64.47% 47

Appendix A Components of the Net Stable Funding Ratio (NSFR) 48