Scholarly article on topic 'Risk, capital and financial crisis: Evidence for GCC banks'

Risk, capital and financial crisis: Evidence for GCC banks Academic research paper on "Economics and business"

Share paper
Academic journal
Borsa Istanbul Review
OECD Field of science
{" Z-score" / Capital / 2SLS / Banking / GCC}

Abstract of research paper on Economics and business, author of scientific article — Saibal Ghosh

Abstract Employing data on over 100 GCC banks for 1996–2011, we test the relation between risk and capital. Given the interlinkage between these two variables, the model employs a 3SLS estimation that takes on board this simultaneity. Consistent with the literature, risk is measured by the Z-score, while capital is computed as the ratio of equity to asset. The findings indicate that banks generally increase capital in response to an increase in risk, and not vice versa. Second, there is an uneven impact of regulatory pressure and market discipline on banks attitude toward risk and capital. Additionally, Islamic banks increased their capital as compared to conventional banks. Besides, the evidence testifies to the fact that banks with higher dependence on wholesale funds and less diversified income profile have higher risk.

Academic research paper on topic "Risk, capital and financial crisis: Evidence for GCC banks"

Accepted Manuscript

Risk, Capital and Financial Crisis: Evidence for GCC Banks Saibal Ghosh, PhD

PII: S2214-8450(14)00028-3

DOI: 10.1016/j.bir.2014.06.003

Reference: BIR 34

To appear in: Borsa istanbul Review

Received Date: 10 March 2014 Revised Date: 25 June 2014 Accepted Date: 28 June 2014

Please cite this article as: GhoshS., Risk, Capital and Financial Crisis: Evidence for GCC Banks, Borsa istanbul Review (2014), doi: 10.1016/j.bir.2014.06.003.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Risk, Capital and Financial Crisis: Evidence for GCC Banks


Employing data on over 100 GCC banks for 1996-2011, we test the relation between risk and capital. Given the interlinkage between these two variables, the model employs a 3SLS estimation that takes on board this simultaneity. Consistent with the literature, risk is measured by the Z-score, while capital is computed as the ratio of equity to asset. The findings indicate that banks generally increase capital in response to an increase in risk, and not vice versa. Second, there is an uneven impact of regulatory pressure and market discipline on banks attitude towards risk and capital. Additionally, Islamic banks increased their capital as compared to conventional banks. Besides, the evidence testifies to the fact that banks with higher dependence on wholesale funds and less diversified income profile have higher risk.

JEL classification: G21, G28

Key words: Z-score, capital, 2SLS, banking, GCC

1. Introduction

The relationship between bank capital and risk-taking is one of the key issues in the banking literature. The minimum capital standards advocated by the Basel Committee which are sought to be implemented are premised on the rationale that increased capital enhances bank safety. However, this premise might often turn out to be less than relevant. By way of example, increased capital might induce a bank to assume greater risks. If this effect outweighs the buffer effect of capital, highly capitalized bank might experience a higher probability of failure. Such risk-taking behavior explains why otherwise well-capitalised banks often experience significant declines in their capital position. An offshoot of this relationship is that capital regulation alone might not be sufficient to ensure the soundness of the banking system.

Given the significance of these results, it is not surprising that several researchers have attempted to dissect the relationship between risk-taking and capitalization of banks. This literature focuses on the well-known moral hazard problems in banking. Merton (1977) shows that, given banks limited liability, they might be inclined to decrease their capital and increase risk. Other studies however, demonstrate that the relationship is not necessarily negative if factors other than the option value are considered (Kim and Santomero, 1988; Besanko and Kantas, 1996; Hellmann et al., 2000).

Akin to theoretical studies, empirical studies on this issue also do not appear to have a consensus view. On the one hand, several researchers (Furlong, 1988; Keeley, 1990; Boyd and Graham, 1996) uncover a negative relationship between risk and capitalization, others (Peek and

Rosengren, 1997; Sheldon, 1995) find this relationship to be the opposite. One possible reason for the lack of consensus among these various studies could be the fact that they employ alternative risk measures as dependent variables. In this process, they do not take on board additional factors that affect the relationship between capitalization and bank risk.

The aim of this paper is to push forward the empirical literature by examining this issue for GCC banks. In particular, we examine the relation between capital and risk for GCC banks employing an extended sample period that encompasses the recent financial crisis. Our study seeks to shed light on the association between these two variables and how it was affected during the financial crisis.

The GCC banking system provides a reasonable laboratory to examine this issue in a holistic fashion. These countries share similar economic and social characteristics and are essentially dependent on a single primary commodity for exports. On average, hydrocarbon accounts for nearly half of the region's GDP, contributes nearly 70% of these countries merchandise exports and over three-fifths of government revenues (IMF, 2013). Following the oil boom, real GDP growth in these countries averaged over 6.5% during 2003-08 as compared to less than 4% during the preceding five year period. Non-oil GDP growth improved markedly, averaging nearly 7.5% during this period (IMF 2010 a,b; 2011). The economic crisis and its after-effects, including the headwinds of the Arab Spring slowed down these economies considerably, with real GDP growth dwindling to 0.3% in 2009, although growth has since turned a corner, averaging close to 6% during 2010-12 (IMF, 2013).1 The fiscal and external positions also witnessed an upturn, providing headroom to the authorities for greater economic diversification, while allowing the surpluses to be invested for productive purposes.

Besides the interest in the research question, the paper also augments the literature in three distinct ways. First, we explore whether the behavior of well-capitalised banks differ from those that do not have adequate capital. Second, we examine how market discipline interacts with regulatory pressure to affect bank capital and risk-taking. Finally, we analyse the differential response of Islamic banks as compared to conventional banks on bank capital and risk.

As is well acknowledged, there are three entities that are primarily involved in the risk determination of a bank: the regulator, the shareholders and the management. Therefore, by categorising banks into different groups, we are able to better understand which set of forces influence their risk-taking behavior. More specifically, for inadequately capitalised banks, it seems

1 Much of the high growth during 2010-12 in the GCC countries can be traced to the high GDP growth of Qatar,which grew at 16.6 and 13% during 2010 and 2011, respectively. Excluding Qatar, the average growth rate of GCC during this period is around 4.7%.

likely that regulatory forces will be paramount in determining their risk-taking, since they are at greater risk of failure. On the other hand, for well-capitalised banks, the managers would have greater discretion in choosing risky assets. Therefore, following the literature (Beaty and Harris, 1998; Ke et al., 1999), we postulate that managers' incentives dominate for publicly traded banks. Finally, in case of Islamic banks, the interests of shareholders (as also depositors) would prevail, given the risk sharing principles embedded in their behavior. Our data set of GCC banks contains a clear distinction along these lines and is therefore well -suited to analyse these research questions.

We contribute to the literature in a few important ways. First, to the best of our knowledge, this is one of the early studies for GCC countries to examine the interlinkage between capital and risk. Most studies that have focused on this aspect are primarily based on US economy (Shrieves and Dahl, 1992; Agrawal and Jacques, 1992) or for other developed (Rime, 2001; Heid et al., 2004; Stolz, 2007) and emerging markets (Ghosh et al., 2003; Godwelski, 2005). Even crosscountry studies on this aspect pertain primarily to developed markets. Judged thus, we focus on a homogenous set of countries in a region which share similar economic, social and financial characteristics. Outside of the US, evidence for Turkish banks appears to suggest that regulatory policies exerted an uneven impact on the efficiency of commercial banks during the 2002-10 period (Ozkan-Gunay et al., 2013)..

Second, our paper extends the literature on ownership and bank risk by focusing on the response of Islamic and commercial banks for an extended period. Several studies (Hasan and Dridi, 2010; Abedifar et al., 2013) suggest that there are no significant differences in the stability of Islamic and conventional banks. Focusing on MENA countries, Srairi (2013) finds that Islamic banks display a lower exposure to credit risk as commercial banks. Whether and to what extent do these findings carry over the periods of crisis remains a moot question, which we empirically explore in the present exercise. Employing an expanded cross-national database, Imam and Kpodar (2013) uncovers that per capita income and greater integration with Middle East economies as major factors that explain the rise of Islamic banking around the globe.

Our paper also contributes to the literature that examines the relevance of funding structure for bank risk. The pre-crisis literature opined in favour of market funding, arguing that the 'market discipline' embedded in such funding coupled with its relatively low cost could enable banks to fund their asset expansion in a swift and cost-effective manner (Calomiris and Kahn, 1991). : The recent financial crisis has however exposed the weaknesses of this argument. Huang and Ratnovski (2008) for example, show that banks that relied less on wholesale funding were able to better withstand the impact of the crisis. Demirguc Kunt and Huizinga (2010) found that banks' resilience on non-deposit funds increased their risk. Other studies show that banks that relied

heavily on wholesale funds were more affected by the liquidity crunch, experienced a large abnormal decline in their share prices (Adrian and Shin, 2009; Raddatz, 2010). A parallel literature focuses on corporate funding costs. For instance, using data for 15 emerging markets over the period 1994-2004, Agca and Celasun (2012) report that banking sector reforms lead to lower corporate borrowing costs. This literature however, does not focus on banks, which is one of the major concerns of the paper.

Our paper also adds to the literature on the importance of income structure for bank risk. The global trend towards diversification of bank income sources has provided banks with additional sources of revenue. While such diversification, in principle, has helped foster stability in overall income, it is not altogether clear whether this overt reliance on non-interest income has helped to limit overall banking risk. In general, studies tend to conclude that the greater reliance on non-interest income does not lower earnings volatility in earnings (Stiroh, 2004) or lower bank systematic risk (De Jonghe, 2010). There is also evidence to suggest that diversification might not necessarily be value enhancing: diversified firms are found to trade at anywhere between 8-15% discount as compared to specialized peers (Lang and Stulz, 1994; DeLong, 2001; Leaven and Levine, 2007).

The paper also adds to the literature that analyses the impact of the recent financial crisis. Several studies have examined the behavior of U.S. banks during the crisis (Ivashina and Scharfstein, 2010; Cornett et al., 2011; Huang, 2010; Santos, 2011) as well as for other countries, including Asia (Ree, 2011), African (Louis et al., 2009), EU (Gardo and Martin, 2010) and GCC (IMF, 2010a) banking systems. Kilinc et al.(2012) explore the factors explaining the resilience of Turkish economy to the global crisis and finds that proactive monetary and fiscal response to be important factors.2 However, the interplay between bank risk taking and capital and its interplay with the crisis is an aspect that has not been investigated in prior empirical research.

The rest of the analysis continues as follows. Section 2 provides an overview of the relevant literature. This is followed by the database and variables employed in the study, followed by the empirical strategy (Section 4), results (Section 5) and concluding remarks (Section 6).

II. Relevant literature

The literature on the relationship between bank capital and risk can be classified into theoretical and empirical components. As regards the former, Kim and Santomero (1988) have

2 Another literature considers the relevance of conventional risk management models (e.g., VaR) across a large sample of developed and developing countries during the crisis. The findings suggest that the performance of these models was much less convincing for advanced economies (Koksal and Orhan, 2013).

observed that less risk-averse banks will prefer low levels of capital. Such a negative relationship has been advocated by other researchers as well (Diamond and Dybvig, 1986). A negative relationship between risk and capital can also occur when all deposits have a flat rate insurance, since in that case, the marginal cost of increasing risk or lowering capital is insignificant.

The other line of thinking that contends a positive association between bank capital and risk has emphasized the unintended effects of implementing minimum regulatory capital standards. Koehn and Santomero (1980) and Kim and Santomero (1988) observe that constraints on a bank's leverage due to minimum regulatory standards may cause banks to view leverage and risk as substitutes. As a result, in response to regulatory requirements, a bank that is forced to lower its leverage might end up raising its risk level. As a result, we will observe a positive relationship between bank capital and risk for those banks that have levels of capital near the minimum regulatory requirements.

Another line of argument derives from the fact that higher capital requirements lower the charter value of banks, in turn, compelling them to assume higher risks (Besanko and Kantas, 1996; Hellmann et al., 2000).3 An additional reason for the positive capital-risk relationship follows from the bankruptcy cost avoidance hypothesis of Orgler and Taggart (1983). According to the authors, banks operating with high levels of portfolio risk tend to hold higher capital levels due to the fact that their probability of bankruptcy is higher.

Finally, the positive relationship between bank capital and risk can also be traced to agency theories. For example, Saunders et al. (1990) hypothesize that managers who do not own their bank equity are prone to take lower risks, because they stand to lose much more when the bank becomes insolvent.

Several empirical studies have tested the above hypotheses on the relationship between changes in bank capital and changes in risk. Most of those studies have employed data from U.S. banks. In an early attempt, Shrieves and Dahl (1992) emphasized the endogenous determination of a bank's capital and risk. Within a simultaneous equations framework, they found that the majority of banks mitigate the effects of increases in capital by increasing exposure to asset risk.

Following from this research, several studies have investigated the capital-risk relationship, with mixed results. In case of US, studies found that banks responded to the new capital standards by increasing risk (Jacques and Nigro, 1997; Aggarwal and Jacques, 2001). Rime (2001) offered similar evidence for Switzerland, suggesting that regulatory pressure led banks to

3 Saunders and Wilson (2001) presents evidence that charter value itself may derive from high-risk activities, indicating that minimizing risk-taking also would limit the value of the charter.

increase their levels of capital. Flannery and Rangan (2004) explain the capital build-up of US banks during the 1990s by increased capital requirements such as the FDIC Improvement Act, high profitability of the banking industry along with higher risk levels, and the withdrawal of implicit government guarantees. Cebenoyan and Strahan (2004) found that banks which used the loan sales market for risk management purposes held less capital and were more profitable but riskier than other banks. This evidence is in line with the Froot and Stein (1998) model that active risk management can allow banks to hold less capital and to invest in riskier assets.

In contrast to this positive relation, several studies have also reported an inverse relationship as well. For example, looking at UK banks over 1998-2003, Alfon et al. (2004) uncovered a negative relationship between capital and risk in U.K. banks and building societies. This contrasts with a previous study, which reports that banks tend to overshoot the target in response to meeting the minimum capital standards. Other studies that report a negative association include Rime (2001) for Swiss banks, Das and Ghosh (2004) for Indian state-owned banks, Stolz (2007) for German banks. Cross-country studies (Van Roy, 2008) also report a negative association between capital and risk.

Another strand of the literature links risk taking to bank ownership. This link is best exemplified by considering the objective function of shareholders and the potential principal-agent problems between shareholders and management. While privately-owned banks tend to focus on profit maximization, government-owned banks might have additional considerations and not just profit maximization alone. Early research on this aspect found that banks controlled by managers exhibited lower risk as compared to those in which shareholders had a controlling majority.

Beginning with the line of research, several authors have investigated the interlinkage between ownership and bank risk. On the one hand, cross-country studies consistently highlight that higher government ownership could jeopardize bank stability (La Porta et al., 2002; Barth, Caprio and Levine, 2004). For a large sample of European banks, lanotta et al. (2007) document that government-owned banks are the least stable. Similarly, Cihak and Hesse (2007) reports that cooperative banks are more stable as compared to commercial banks in OECD countries. Saunders et al. (1990) and Laeven and Levine (2008) confirm predictions of the theory that banks with large dominating shareholders take larger risks. For European banks, Barry et al. (2011) demonstrate that ownership structure is important in explaining risk differences in privately owned banks. In a similar vein, Iannotta et al (2013) report that government-owned banks have lower default risk as compared to commercial banks. Unlike the existing literature, our paper considers bank stability across ownership types within GCC countries, holding constant the institutional and macroeconomic framework.

In contrast to these studies, analyses examining the risk behavior of Islamic banks are quite limited. Employing data for 1993-2004 on OECD economies, Cihak and Hesse (2010) documents that small Islamic banks are more stable as compared to similar-sized conventional banks. Subsequent cross-country research (Beck et al, 2010; Abedifar et al., 2013) appears to suggest little difference in terms stability. Focusing primarily on the MENA countries, Hasan and Dridi (2010) finds that pre-crisis profitability of Islamic banks to be higher than their conventional counterparts, although these differences petered out during the crisis, suggesting that higher pre-crisis profitability of the former was not driven by excessive risk-taking.

Closer to the region, Khamis (2010) provides an overview of the financial crisis on GCC countries. Al Hassan et al (2010) analyses the banking sector in the GCC economies, including trends in ownership, balance sheet exposures and risks. Espinosa and Prasad (2010) analyze the delinquent loans in GCC banks and finds that a decline in non-oil GDP growth by 3 percentage points would raise NPLs by roughly 0.3%. Arvai et al. (2014) discusses the macroprudential policies employed by GCC countries.

III. Database and variables

III.1 Database

The analysis employs a detailed bank-level dataset. The core of the data is the information on bank's balance sheet and income statement details as published by Bankscope, a comprehensive, global database containing information on nearly 30,000 public and private banks globally, maintained by International Credit Analysis Limited (IBCA).

We use a sample comprising of an unbalanced panel of annual report data from 1996-2011 for the GCC banking system, comprising commercial and Islamic banks. The sample initially contained nearly 120 banks, but subsequently we deleted the finance and investment companies, including investment banks, providing us with 112 banks. Several banks also do not report data on the dependent variables employed in the analysis, which we exclude from the sample. After this filtering, we have observations on 103 banks at an average of 10.3 years of observations, yielding a maximum of 1065 bank-years. To moderate the influence of outliers, we winsorized the top and bottom 1% of observations for the dependent variable. In 2011, the final year of the sample, these banks accounted, on average, for roughly 75% of total banking assets in their respective countries. Table 1 provides the sample composition, while table 2 describes the variable definitions, including summary statistics.

[Table 1] [Table 2]

The summary statistics suggest that, on average, banks are well-capitalized with equity-to-asset in excess of 10%, the regulatory minimum stipulated by most countries. Banks also appear to exhibit high stability, as evidenced from their high Z-scores. Contextually, it may be mentioned that in 2011, Z-scores of banks in major developed economies such as US, UK, Germany, France and Sweden ranged from a low of 0.9 in UK to a high of 1.45 as in US (World Bank, 2013).

Among others, 15% of banks funding are short-term in nature, although their cost-to-income are among the lowest. The average bank is fairly small in size with high profitability levels, close to 2% of total assets. These profitability numbers are comparable to, and in some cases, even better than those obtaining for advanced economies (See,for example, BIS 2013).

111.2 Measurement of risk

There is limited consensus in the literature as to the measurement of risk.4 Researchers have employed various risk measures, such as risk-weighted assets to total assets (Shrieves and Dahl, 1992; Jacques and Nigro, 1997; Avery and Berger, 2001; Van Roy, 2008); non-performing loan ratio (Shrieves and Dahl, 1992), loan loss reserves (Altunbas et al., 2007), while more recent research employs the Z-score (Boyd and Graham, 1988; Laeven and Levine, 2009; Barry et al., 2011; Bouwens and Verriest, 2014). In our analysis, consistent with recent research, we prefer to use the Z-score. The Z-score indicates the distance from insolvency and combines accounting measures of leverage, profitability and volatility. As Barry et al. (2011) remark, the Z-score comprises of two elements: the first component - RoA/SD(RoA) - measuring asset risk and the second (K/A)/SD(RoA) measuring leverage risk. A higher Z-score indicates that the bank is more stable. Since the Z-score is positively skewed, we use the natural logarithm of Z-score, which is normally distributed (Laeven and Levine, 2009). As is well known, norms for classification of delinquent loans differs markedly across GCC countries (See, for example, World Bank 2012). As a result, we prefer not to employ the NPL variable or the loan loss provisioning variable.

111.3 Measurement of capital

Capital is measured as the ratio of capital to asset. In effect, total capital comprises of all the capital components permitted under the relevant Acts in each country and is comparable to the definition employed in the Basel accord.

4 Since all banks are not listed and most of their assets being not marketable, it s difficult to compute the volatility for the market price of bank's assets. Likewise, since all banks are not rated and have not received any external support, computing sophisticated measures of risk are not possible.

III.4 Control variables

In the capital equation, the control variables include size and profitability. Size is measured as the natural logarithm of total assets. Bigger banks are likely to be able to access equity markets more easily and therefore, expected to have lower capital ratios. Alternately, given the concentrated banking systems that typify these economies, larger banks might be tempted to expand faster to fund asset growth and therefore, need higher capital levels. In the presence of asymmetric information, profitable banks might prefer to increase capital through retained earnings, suggesting an expected positive sign on this variable.

In addition, we include two sets of interaction terms. The first - CAPREG*CAP and CAPREG*ARISK - ascertains how regulatory pressure responds to capital and changes in risk.

The capital pressure variable (CAP) is defined as the ratio of capital ratio stipulated by the regulatory authorities in respective countries to the bank's actual capital adequacy ratio, a higher ratio implying greater regulatory pressure. The second - MREG*CAP and MREF*ARISK - focuses on how banks with market pressure respond to capital and risk.

The control variables in the risk equation include size, funding, index of income profile, inefficiency and the interaction terms, akin to those earlier. Larger banks carry out a wider set of activities, which should increase their ability to diversify portfolios and lower credit risk. Second, banks with greater dependence on wholesale funds are likely to be perceived as more risky, as recent evidence would testify (Raddatz, 2010). On the other hand, it is not evident whether banks with diversified income stream are more or less risky, since the empirical evidence, both for the U.S. and around the world is mixed (Stiroh, 2010). Authors like Hughes et al (1994) and Kwan and Eisenbis (1997) emphasize the importance of analysing the impact of efficiency on bank risk taking. Using data on US banks, they uncover a negative relationship between inefficiency and bank risk. Besides, we include dummy to distinguish between Islamic and conventional banks, in addition to a dummy for crisis. All the specifications control for country and year shocks by including an interaction term between country and year effects.

IV. Empirical Strategy

Following Shrieves and Dahl (1992) and Van Roy (2007), we employ the simultaneous equation setup wherein observed changes in banks' capital and risk-taking consist of two components, a discretionary adjustment and a change caused by factors exogenous to the bank, as given by Eqs.(1) and (2).

DCAPbt =DCAPbdt + ubJ (1)

ARISKbJ = ARISKbJ + wbJ (2)

where ACAP and ARISK are the observed changes in capital and risk levels, respectively, for bank b at time t. The ACAPdbt and ARISKdbt variables are discretionary adjustments in capital and risk and u and w are exogenous random shocks in capital and risk levels, respectively.

Following Shrieves and Dahl (1992), the discretionary changes in capital and risk is modeled using a partial adjustment framework, such that:

DCAPlt = a (CAP* - CAPbtt_,) (3)

ARISK* = b (RISK*,t - RISKbt-i) (4) Substituting (3) and (4) into (1) and (2) and taking on board the simultaneous capital and risk decisions, the observed adjustments in capital and risk can be written as:

DCAP,^t = a (CAP* - CAPb,-) + ub, (5)

DRISKbJ = a(RISK I,t - RISK b t_i) + wb t (6)

In effect, the observed changes in capital and risk are a function of the target capital and risk levels, the lagged capital and risk levels and any random shocks. The target capital and risk levels are not directly observable, but are assumed to depend on a set of observable variables describing the bank's financial condition and country and year characteristics.

The target capital ratio is assumed to depend on bank size, measure of profitability, changes in risk (ARISK) and crisis, ownership and regulatory dummies, as discussed earlier. The variables used to proxy the target risk ratio are bank size, income diversification index, funding profile and several dummies, as discussed above.

Given the simultaneous equation setup of (5) and (6), the empirical strategy has to account for the endogeneity of the regressors ACAP and ARISK. In contrast to the ordinary least squares, 2SLS and 3SLS estimators take the endogeneity into account, thereby producing consistent estimates. As Zellner and Theil (1962) have observed, 3SLS exploits the information that the disturbance terms in the two structural terms are contemporaneously correlated and thereby ensures consistent estimates.5 Since the estimates under the two models are similar, we present the results using the 3SLS technique.

Before discussing the results, Table 3 sets out the descriptive statistics of the dependent and independent variables, both for the overall sample as well as for sub-samples,

5 The 3SLS methodology is sensitive to misspecification or measurement errors. As a result, a comparison with 2SLS estimates as a specification check becomes relevant. In the present case, estimation of equations (5) and (6) with 2SLS produce results similar to those obtained under 3SLS.

classified according to ownership and listing. The results indicate that, on average, Islamic banks are more stable (and therefore, less risky) as compared to conventional banks, broadly consistent with Cihak and Hesse (2008) who find large Islamic banks to be less stable as compared to their conventional counterparts. Profitability of Islamic banks is found to be half of those obtaining for conventional banks, on average. Both these differences are statistically significant at the 0.01 level. In contrast, the table suggests there are hardly any differences in either profitability or stability for listed versus unlisted banks. [Table 3]

The results also indicate significant differences between Islamic and conventional banks as also between listed versus unlisted banks in terms of major characteristics such as size, profitability, income composition, funding strategy and efficiency levels. In particular, Islamic banks are smaller in size and have less diversified income profiles as compared to conventional banks (See, for example, Kamruddin et al., 2008). On the other hand, listed banks are observed to be bigger, more profitable, more diversified in terms of income, less efficient and exhibit lower wholesale dependence. All these differences are statistically significant at conventional levels.

Coming to regulatory pressure, conventional banks are observed to face higher regulatory pressure as compared to Islamic banks. The difference is statistically significant at the 0.05 level. In contrast, Islamic banks face lower regulatory pressure as do listed banks. This suggests that given their ability to access capital markets, listed banks encounter lower regulatory pressure as compared to unlisted ones.

Stability of Islamic banks is higher during non-crisis periods, although their capital levels appear to be lower during non-crisis times. This would imply that Islamic banks operate on higher leverage, although profitability levels are not significantly different across these two categories.

V. Discussion of the results

V.1 Baseline results

Three sets of results are set out in Table 4. Specification 1 estimates the baseline model wherein adjustments in capital and risk differ according to the extent of regulatory and market pressure. Specification 2 allows for differential speeds of adjustment in capital and risk. Finally, specification 3 additionally allows for differences in the coordination of capital and risk adjustments.

We first discuss the control variables. Table 4 (Model 1) indicates that bank size has a negative and significant effect on capital. The negative effect on capital is in line with the empirical literature and means that larger banks increase capital by a magnitude that is lower as compared to

smaller banks. A possible explanation for this is that larger banks have easier access to capital markets or alternately, they prefer to undertake more monitoring rather than hold higher levels of expensive capital (Titman and Wessels, 1988; Van Roy, 2008). RoA has a highly statistically highly significant and positive impact on capital, consistent with previous research that banks with higher earnings can retain more capital (Aggarwal and Jacques, 2001; Rime, 2001; Van Roy 2008).6 This positive relation is also consistent with the pecking order hypothesis, which supports the bank's preference for internal funding owing to lower costs (Myers and Majluf, 1984). [Table 4]

In the risk equation, the coefficient on both Funding is negative while that on Divers is positive. Both these coefficients are statistically significant. In other words, banks with higher wholesale dependence exhibit lower stability (i.e., higher risk): a 1% increase in wholesale dependence lowers bank stability by 0.8 percentage points. Banks with more diversified income streams are observed to be less risky: a 1% increase in income diversification lowers bank risk by 0.8 percentage points. Therefore, a 50% in income diversification from 0.31 to 0.47 - equal to a move from the 25th to the 75th percentile of the distribution - would lower bank risk by over 40% points (=0.8*51=41.2). The fact that bank funding structures might be relevant in influencing bank risk-taking has been acknowledged in recent empirical research (Adrian and Shin, 2009; Ratnovski and Huang, 2009; Raddatz, 2010).

As regards the impact of regulatory pressure, the results in model 1 suggest that, banks with high regulatory pressure increase capital by less, than banks with high buffers, although their response to risk appears to be limited. On the other hand, banks with higher market pressure lower their capital and raise risk as compared to those with no market pressure.

The parameter estimates of lagged capital and risk are statistically highly significant, consistent with previous research for the US (Jacques and Nigro,1997) and elsewhere (Rime, 2001; Das and Ghosh, 2004). The expected negative sign lie in the [0,1] interval. Hence, these can be interpreted as speed of capital and risk adjustments. In general, the speed of capital adjustment is roughly 1.2-1.5 times higher than the speed of risk adjustment. In Model 1, the coefficient on capital adjustment is the highest (-0.66); the corresponding coefficient on risk adjustment is -0.46.

6 Note that capital satisfies the following identity:

Capital =-^-*-RWA-* TA

Risk weighted assets (RWA) Total assets (TA)

or, K=R*P*TA, where K=capital, R=(Capital/RWA); P=(RWA/TA). Re-arranging and denoting K = DK /K , we obtain: R = K- P- TA, suggesting that the adjustment by banks to a change in risk-weighted capital can be effected in three possible ways: (a) raise capital (increase K); (b) lower risk weighted assets (lower P) or (c) lower total assets.

The estimated speeds of adjustment mean that shocks to capital and risk are halved after 1.1 and 1.5 years, respectively. In essence, this implies that the speeds of adjustment are quite high, mirroring the fact that the average bank built up capital and risk over the adjustment period; the latter at a rate slightly faster as compared to the former. This is not surprising, given the rapid loan growth experienced by these GCC economies prior to the crisis.

As regards the response of endogenous variables, the parameter estimate of ARISK in the capital equation is insignificant, whereas the parameter estimate of ACAP in the risk equation is positive and highly significant. Van Roy (2008) also reports similar evidence for European and Canadian banks in the cross-country study, although their US sample banks did not exhibit any discernible response of either capital or risk. The magnitude of the coefficient on ACAP gradually increases as the specification is buttressed with additional variables (See Models 1-3). In other words, after controlling for differences in the coordination of capital and risk adjustments, increases in capital is associated with decrease in risk, consistent with evidence obtaining for Japanese banks (Van Roy,2008). In Model 3, the standardised beta coefficient suggests that an increase in capital by one standard deviation would lower risk by 0.11 standard deviation.

While the coefficient on the interaction term of RPH with lagged capital and risk variables are insignificant, the coefficient on MREG*Cap(t-1) is negative and statistically significant in Models 2 and 3, indicating that banks facing market discipline adjust capital thrice as fast as compared to banks with no such discipline.

Further, we find that the estimated coefficients on RPH*ARISK is significant and positive, while the estimated coefficient of RPH*ACAP is significant, positive and nearly four times larger than the coefficient on ACAP. This finding indicates that capital and risk adjustments are negatively correlated for banks with high regulatory pressure. These positive coefficients are in line with recent evidence for US market (Shim, 2010).The impact of the regulatory pressure variable is extremely large. To see this, consider a bank with ACAP equal to 0.0003, the average for the sample. Ignoring the impact of RPH, the point estimates in Model 3 indicate that an increase in ACAP by the mean value would increase stability (i.e., lower ARISK) by 0.0002 (=0.683*0.0003) percentage points. Taking RPH into account, the point estimates in Model 3 yield an estimate of 0.0008 (=2.623*0.0003) percentage points, nearly 300% higher as compared to without RPH.

Looking at ownership, when significant, the coefficient on Islamic is positive and significant in the ACAP equation, indicating that, after controlling for bank specific and country-year characteristics, changes in capital are higher for Islamic as compared to conventional banks. The effect is quantitatively important, indicating that capital increase for the average Islamic bank is

roughly 0.04 percentage points higher as compared to an average conventional bank. Considering the average capital change in the sample however, this is not a sizeable difference.

V.2 Robustness checks

We undertake several robustness checks of the baseline results. First, following Van Roy (2008), we analyze the marginal effect that market discipline has on the relationship between regulatory pressure and ACAP and between regulatory pressure and ARISK. As observed by Jacques and Nigro (1997), banks with capital above or below the regulatory capital standards might respond in a different manner to the regulatory standards. Banks with capital below the stipulated standards will be under considerable pressure to improve capital ratios, by either increasing capital or by lowering risk weighted assets. On the other hand, banks with higher-than-stipulated capital levels might reduce their capital or increase portfolio risk. Taking these considerations on board, we separately analyze for each of these two bank categories the impact on capital and risk.

Second, we explore the impact of crisis on bank behavior, classified according to their capital levels. Intuitively, the crisis increased the quantum of delinquent loans, necessitating an increase in provisioning and thereby putting pressure on banks' capital and in particular, more on those banks with low levels of capital. How this interaction played out remains an open question. [Table 5]

Finally, we examine the differential response of Islamic banks to changes in capital and risk, and in particular, how it evolved during the crisis. Observers have contended that the positive features embedded in these banks, including the close link between financial transactions and productive flows and the built-in dimensions of governance and risk management appeared to have cushioned them from the worst of the crisis (Akhtar Aziz, 2009; Yilmaz, 2009). Whether and to what extent it impacted their capital and risk behavior remains to be empirically analyzed.

All regressions include the full set of controls and dummy variables, but these are not reported for purposes of brevity.

In Model 1, the coefficients on all three variables - RPH*(1-MREG), RPH*MREG and (1-RPH)*MREG - are statistically significant at the 0.01 level in the ACAP model. Take for example, RPH*(1-MREG). The coefficient is negative in the ACAP equation with a point estimate -0.08. Ceteris paribus, banks which faced high regulatory pressure but no market pressure lowered their capital ratio by 0.08 percentage points more than banks that did not experience either form of pressure. Similar evidence is obtained for the other two interaction terms. In effect, this suggests that banks which experienced both regulatory and market pressures or for that matter, banks experiencing

only market pressure and not regulatory pressure, lowered their capital ratios. However, the size of the effect in all cases is not very large, ranging from 0.05-0.09 percentage points.

On the other hand, as regards banks facing low regulatory pressure, the evidence indicates that, irrespective of whether they encountered market pressures or not, they increased their capital ratios. By way of example, banks experiencing both low regulatory and market pressure improved their capital by 0.1 percentage points, on average, as compared to those not facing any such pressures.

The evolution of regulatory and market pressures was, however, quite different during the crisis. The evidence appears to suggest that banks facing high regulatory pressure, but no market pressures raised their risk levels during the crisis, although there was no perceptible impact on their capital. On the other hand, banks facing low regulatory pressure, but no market pressure raised capital levels during the crisis. In essence, during the crisis, inadequately capitalized banks raised risks perhaps in order to garner market shares, whereas well-capitalized banks further buttressed their capital positions, indicating an asymmetry in their risk and capital adjustments. This is broadly consistent with the data which suggests that overall capital adequacy of banks in these countries has increased significantly after the crisis (IMF, 2013).

Finally, focusing on ownership, the evidence appears to suggest that Islamic banks increased their risk in response to an increase in capital (Model 4). The effect was also manifest during the crisis as well, wherein these banks further raised their risks. The effect of the crisis variable is quite large. To see this, consider the difference between an average Islamic bank and an average conventional bank when ACAP equals 0.00003, the average for the sample. Ignoring the impact of crisis, the point estimates in Col (6) yield an estimate of 0.002% points (=-6.672*0.0003=-0.002). Taking into account the crisis impact, the point estimates in Col.(8) yield a difference of approximately 0.6% points (=-6.672*0.0003-0.655=-0.6). In contrast, the estimates in Model 4 indicate that there is no perceptible impact on capital of risk-taking by Islamic banks.

Under all cases, changes in capital are observed to exert a significant and negative impact on bank risk, and not vice versa, consistent with our baseline estimates.

VI. Concluding remarks

The role of minimum capital requirements in the context of modern banking regulation has been a widely discussed and debated topic in the literature. Efforts are underway by the Basel Committee to devise capital regulations consistent with banks' risk profile and business strategies (BCBS, 2012). However, whether and to what extent does higher capital level encourage or dissuade risk-taking by banks remains an empirical question. While there are several studies that

examine this aspect for emerging and advanced economies, either from a country-specific context or from a cross-country standpoint, the evidence on this aspect for GCC banks is admittedly limited. Given that these countries are planning to implement the Basel III standards and are primarily bank-based economies, an analysis of the relationship between risk and capital for these economies remains an unaddressed area of research.

In this context, the present paper employs data on an extended sample of GCC banks to examine this issue in detail. Three major findings emerge. First, banks generally lower capital in response to an increase in risk, and not vice versa. Second, there is an uneven impact of regulatory pressure and market discipline on banks attitude towards risk and capital. And third, Islamic banks increased their capital as compared to conventional banks.

These findings have useful implications for policy formulation. First, the positive relationship from capital to risk suggests that capital regulation alone may not be enough to ensure bank soundness and would need to be complemented with additional regulatory instruments. Secondly, the negative relationship between regulatory pressure and market discipline for banks with low-capital suggests that there is a need for closer and continuous monitoring. One possible way could be to give priority to risk-based supervision as a way to prevent those banks from gambling in excessively risky strategies. Third, the analysis suggests that ownership matters for bank capital. Importantly as well, the evidence points to risk taking by Islamic banks, especially during the crisis. Given the quasi-equity of Islamic banks in the form of mudarabah saving and investment (S&I) deposits, such risk-taking does not appear to have made a significant dent on their capital position. However, given the limited opportunities for diversification for Islamic banks in these economies, it becomes important to carefully monitor their capital position, especially in times of stress. Finally, the analysis makes an explicit distinction between changes in capital and risk brought about by not only regulatory, but also market pressures. Not much is known as to how market pressure interacts with regulatory pressure in influencing capital and/or bank risk-taking. This interaction of market pressure with regulatory pressure and its effect on capital is also important in light of the fact that minimum capital requirements and market discipline constitute two of the three pillars of the New Basel Accord.


Abedifar, P., A. Tarazi and P. Molyneux (2013). Risk in Islamic banking. Review of Finance 17, 203596.

Adrian, T. and H.S. Shin (2009). Money, liquidity, and monetary policy. American Economic Review 99, 600-605.

Agca, S., and O.Celasun (2012). Banking sector reforms and corporate borrowing costs in emerging markets. Emerging Markets Finance and Trade 48, 71-95.

Aggarwal, R. and K. Jacques (1997). A simultaneous equations estimation of the impact of prompt corrective action on bank capital and risk. In Financial Services at the Crossroads: Capital Regulation in the 21st Century, Federal Reserve Bank of New York.

Al Hassan, A., M. Khamis and N. Oulidi (2010). The GCC banking sector: Topography and analysis.

IMF Working Paper No. 87. IMF: Washington DC.

Alfon I., I Argimon and P. Bascunana-Ambros (2004), What Determines how Much Capital is Held by UK Banks and Building Societies, Occasional Paper Series 22, FSA, London.

Altunbas, Y., S. Carbo, E.Gardener and P.Molyneux (2007). Examining the Relationships between Capital, Risk and Efficiency in European Banking. European Financial Management 13, 49-70.

Arvai, Z., A. Prasad and K. Katayama (2014). Macroprudential policies in GCC countries. IMF Staff Discussion Note 1,IMF: Washington DC.

Avery, R. B. and A.N. Berger (1991). Loan commitments and bank risk exposure. Journal of Banking and Finance 15, 173-192.

Bank for International Settlements (2013). Annual Report. BIS: Basel.

Barry, T.A.,L.Lepetit and A.Tarazi (2011). Ownership structure and risk in publicly held and privately owned banks. Journal of Banking and Finance 35, 1327-40.

Barth, J., G. Caprio and R. Levine (2004). Bank Supervision and Regulation. What Works Best? Journal of Financial Intermediation 13, 205-48.

Basel Committee for Banking Supervision (2012). International regulatory framework for banks. BIS: Basel.

Beatty, A. and D. Harris. (1999). The effects of taxes, agency costs and information asymmetry on earnings management: A comparison of public and private firms. Review of Accounting Studies 4, 299-326.

Beck, T., A.Demirguc-Kunt and O.Merrouche (2010). Islamic vs. conventional banking. Business model, efficiency and stability. World Bank Working Paper 5446. The World Bank: Washington DC.

Besanko, D.,and G.Kantas (1996). The regulation of bank capital: Do capital standards promote bank safety? Journal of Financial Intermediation 5, 160-83.

Bouwens, J., and A.Verriest (2014). Putting skin in the game: managerial ownership and bank risk taking. Harvard Business School Working Paper 70, Cambridge: MA.

Boyd, J.H. and S.L.Graham (1988). The profitability and risk effects of allowing bank holding companies to merger with other financial firms: A simulation study. Quarterly Review Federal Reserve Bank of Minneapolis2, 3-20.

Calomiris, C. W and C. Kahn (1991). The role of demandable debt in structuring optimal banking arrangements, American Economic Review 81, 497-513.

Cebenoyan, A. and P.E. Strahan P. (2004). Risk management, capital structure and lending at banks. Journal of Banking and Finance 28, 19-43.

Cihak, M. and H. Hesse (2007). Cooperative banks and financial stability. IMF Working Paper 2, IMF: Washington DC.

Cihak, M., & Hesse, H. (2008). Islamic banks and financial stability: An empirical analysis. IMF Working paper 16. IMF: Washington DC.

Cornett, M. and H. Tehranian (1994). An examination of voluntary versus involuntary security issuances by commercial banks: the impact of capital regulations on common stock returns. Journal of Financial Economics 35: 99-122.

Das, A. And S.Ghosh (2004). The relation between risk and capital: Evidence from Indian public sector banks. RBI Occasional Papers (Summer), 1-21.

Das, A., and S.Ghosh (2007). Determinants of credit risk in state-owned banks: An empirical investigation. Economic Issues 12, 48-66.

De Jonghe, O. (2010). Back to the basics in banking? A micro-analysis of banking system stability. Journal of Financial Intermediation 19, 387-417.

De Long, G. (2001). Stockholder gains from focusing versus diversifying bank mergers. Journal of Financial Economics 59, 221-52.

Demirguc-Kunt, A. and H.Huizinga (2010). Bank activity and funding strategies: The impact on risk and return. Journal of Financial Economics 98, 626-50.

Diamond, D.W., and P.H.Dybvig (1986).Banking theory, deposit insurance and bank capital regulation. Journal of Business 59, 55-67.

Dothan,U., and J.Williams (1980). Banks, bankruptcy and public regulation. Journal of Banking and Finance 4, 65-88.

Espinosa, R and A. Prasad (2010). Nonperforming loans in the GCC banking system and their macroeconomic effects. IMF Working Paper No. 224. IMF: Washington DC.

Flannery, M. and K. Rangan (2006). Partial adjustment toward target capital structures. Journal of Financial Economics 79, 469-506.

Froot, K. A. and J.C. Stein (1998). Risk management, capital budgeting, and capital structure policy for financial institutions: an integrated approach, Journal of Financial Economics 47, 55-82.

Furlong, F. (1988). Changes in bank risk-taking. Economic Review Federal Reserve Bank of San Francisco (Spring), 2, 5-19.

Furlong, F. and M.Keeley (1988). Capital regulation and bank risk taking: A note. Journal of Banking and Finance 13, 883-91.

Furlong, F., and M.Keeley (1989). Capital regulation and bank risk taking: a note. Journal of Banking and Finance 13: 883-91.

Gardo, S. and R. Martin (2010). The impact of the global economic and financial crisis on Central, Eastern and South-Eastern Europe: A stock-taking exercise. ECB Occasional Paper 115. ECB: Frankfurt.

Ghosh, S., D.M.Nachane, A.Narain and S.Sahoo (2003). Capital requirements and bank behaviour: An empirical analysis of Indian public sector banks. Journal of international Development 15, 145-156.

Godlewski, C.J. (2005). Bank capital and credit risk taking in emerging market economies. Journal of Banking Regulation 6, 1128-45.

Hasan, M., and J.Dridi (2010). The effect of the global crisis on Islamic and conventional banks: A comparative study. IMF Working paper 201. IMF: Washington DC.

Heid, F., D.Porath and S.Stolz (2004). Does capital regulation matter for bank behaviour? Evidence for German savings bank. Deutsche Bundesbank Discussion Paper 3, Bundesbank: Bonn.

Hellmann, T., K.Murdock and J.E.Stiglitz (2000). Liberalization, moral hazard in banking and prudential regulation: Are capital requirements enough? American Economic Review 90, 147-65.

Huang, R. (2010). How committed are bank lines of credit? Experiences in the subprime mortgage crisis. Federal Reserve Bank of Philadelphia Working Paper No. 25, Philadelphia: PA.

Huang, R. and L. Ratnovski (2011), The dark side of bank wholesale funding, Journal of Financial Intermediation 20, 248-63.

Hughes, J. P., W. Lang, L. J. Mester and C-G Moon (1994). Recovering technologies that account for generalized managerial preferences: An application to non-risk-neutral banks. Working Paper 21, Department of Economics, Rutgers University.

Iannotta, G., G.Nocera and A.Sironi (2013). The impact of government ownership on bank risk. Journal of Financial Intermediation 22, 152-76.

Ianotta, G., G. Nocera and A. Sironi (2007). Ownership Structure, Risk and Performance in the European Banking Industry. Journal of Banking and Finance 31, 2127-2149.

Imam, P.A. and K.Kpodar (2013).Islamic banking: How has it expanded? Emerging Markets Finance and Trade 49, 112-37.

International Monetary Fund (2010a). Impact of the global financial crisis on the Gulf Cooperation Countries and challenges ahead. IMF: Washington DC.

International Monetary Fund (2010b). Regional Economic Outlook - Middle East and Central Asia (April). IMF: Washington DC.

International Monetary Fund (2011). Regional Economic Outlook - Middle East and Central Asia (April). IMF: Washington DC.

International Monetary Fund (2013). Regional Economic Outlook - Middle East and Central Asia (November). IMF: Washington DC.

International Monetary Fund (2014). Macroprudential Policy in GCC Countries. Staff Discussion Note 1. IMF: Washington DC.

Ivashina, V. and D. Scharfstein (2010). Bank lending during the financial crisis of 2008. Journal of Financial Economics 97, 319-38.

Jacques, K. and P. Nigro (1997). Risk-based capital, portfolio risk, and bank capital: A simultaneous equations approach. Journal of Economics and Business 49, 533-547.

Kamaruddin, B. H., M.Safa and R. Mohammed (2008). Assessing production efficiency of Islamic banks and conventional bank Islamic windows in Malaysia. International Business Management Research 1, 31--48.

Ke, B., K. Petroni, and A. Saffieddine. (1999). Ownership concentration and sensitivity of executive pay to accounting performance measures: Evidence from publicly and privately-held insurance companies. Journal of Accounting and Economics 28, 185-209.

Keeley, M. (1990). Deposit insurance, risk and market power in banking. American Economic Review 80, 1183-1200.

Khamis, M. (2010). The impact of the global financial crisis on the GCC region: Lessons and reform priorities. Paper presented at the Roundtable on 'Effective and efficient regulation in the MENA region". December 6-7, Doha: Qatar.

Kilinc, M., Z.Kilinc and I.M.Turhan (2012). Resilience of the Turkish economy during the global financial crisis of 2009. Emerging Markets Finance and Trade 48, 19-34.

Kim, D. and A.M. Santomero (1988), Risk in banking and capital regulation. Journal of Finance 43,1219-1223.

Koehn, M. and A.M. Santomero (1980). Regulation of bank capital and portfolio risk. Journal of Finance 35, 1235-1244.

Koksal, B., and M.Orhan (2012). Market risk of developed and emerging countries during the global financial crisis. Emerging Markets Finance and Trade 49, 20-34.

Kwan, S and R.A.Eisenbis (1997). Bank risk, capitalization and operating efficiency. Journal of Financial Services Research 12, 117-131.

La Porta, R., F. Lopez-de-Silanes and A. Shleifer (2002). Government ownership of commercial banks. Journal of Finance 57, 265-301.

Laeven, L. and R. Levine (2007). Is there a diversification discount in financial conglomerates? Journal of Financial Economics 85, 331-67.

Laeven, L. and R. Levine (2009). Bank governance, regulation and risk-taking. Journal of Financial Economics 93, 259-275.

Lang, L., and R.Stulz (1994). Tobin's Q, corporate diversification and firm performance. Journal of Political Economy 102, 1248-80.

Louis, K., N. Leonce and R. Taoufik (2009). Impact of the global financial and economic crisis on Africa. African Development Bank Working Paper No. 96. African Development Bank: Tunisia.

Merton, R.C. (1977). An analytic derivation of the cost of deposit insurance using option-pricing estimates. Journal of Banking and Finance 1, 3-11.

Milne, A., and E.Whalley (2001). Bank capital regulation and incentives for risk-taking. Available at http: //staff.

Myers, S. C., and N. Majluf (1984). Corporate financing and investment decisions when firms have information investors do not have. Journal of Financial Economics 13, 187-221

Myers, S., and N.Majluf (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13: 187-222.

Orgler, Y.E.,and R.A.Taggart Jr.(1983). Implications of corporate capital structure theory for banking institutions. Journal of Money, Credit, and Banking 15, 212-221.

Ozcan-Gunay, E., Z.N.Gunay and G.Gunay (2013). The impact of regulatory policies on risk taking and scale efficiency of commercial banks in an emerging banking sector. .Emerging Markets Finance and Trade 49, 80-98.

Peek, J. and E. Rosengren (1997). Derivatives activity at troubled banks. Journal of Financial Services Research 12, 287-302.

Peek, J., and E.Rosengren (1995). Bank lending and the transmission of monetary policy. Paper presented at the Conference at Federal Reserve Bank of Boston.

Raddatz, C. (2010). When the rivers run dry: Liquidity and the use of wholesale funds in the transmission of the US subprime crisis. World Bank Working Paper 5203. The World Bank; Washington DC.

Ratnovski, L., and R. Huang (2009). Why are Canadian banks more resilient? IMF Working Paper No. 152. IMF: Washington DC.

Ree, J.J. (2011). Impact of the global crisis on banking sector soundness in Asian low-income countries. IMF Working Paper No. 115. IMF: Washington DC.

Rime, B. (2001). Capital requirements and bank behaviour: Empirical evidence for Switzerland. Journal of Banking and Finance 25, 789-805.

Santos, J.A.C. (2011). Bank corporate loan pricing following the subprime crisis. Review of Financial Studies 24, 116-43.

Saunders, A.,E.Strock and N.Travlos (1990). Ownership structure, deregulation and bank risk taking. Journal of Finance 45, 643-54.

Sheldon, G. (1995). A limit-risk capital adequacy rule: An alternative approach to capital adequacy regulation for banks with an empirical application to Switzerland. Swiss Journal of Economics and Statistics 131,773-805.

Shim, J. (2010). Regulatory capital requirement, portfolio risk and capital determinants: Empirical evidence from U.S. property liability insurers. Journal of Banking and Finance 34, 2450-61.

Shrieves, R.E. and D. Dahl (1992). The relationship between risk and capital in commercial banks. Journal of Banking and Finance 16, 439-457.

Srairi, S. (2013). Ownership structure and risk-taking behaviour in conventional and Islamic banks: Evidence from MENA countries. Borsa Istanbul Review 13,115-27.

Stiroh, K.J. (2004). Diversification in banking: Is non-interest income the answer? Journal of Money, Credit, and Banking 36, 853-882.

Stolz, S. (2007). Bank capital and risk taking. Springer: Heiderberg.

Titman, S., and R. Wessels, (1988). The determinants of capital structure choice, Journal of Finance 43, 1-19.

Van Roy, P. (2008). Capital requirements and bank behavior in the early 1990s: Cross country evidence. International Journal of Central Banking 4, 29-60.

World Bank (2012). Bank regulation and supervision database. The World Bank, Washington DC.

World Bank (2013). Financial structure database. The World Bank, Washington DC.

Zellner, A., and H.Theil (1962). Three-stage least squares: Simultaneous estimation of simultaneous equation. Econometrica 30, 54-78.

Table 1: Composition of banks by country

Country Conventional banks Islamic banks Listed Avg no. of years of observations Total observations

Bahrain 11 20 11 7.8 243

Kuwait 6 9 14 9.3 139

Oman 6 0 5 13.3 80

Qatar 7 3 8 11.5 115

Saudi Arabia 9 4 11 13.2 172

UAE 18 10 21 11.3 316

Total 57 46 70 10.3 1065

Table 2: Variable definition and summary statistics

Variable Definition Mean (SD) p.25 p.75

Dependent variables

CAP Capital/Total asset 0.144 (0.202) 0 0.164

Z-score log(l+Z). where Z=(CAP+RoA)/SD(RoA), where RoA=return on asset and SD(RoA) is the rolling standard deviation of RoA based on observations in year t, t-1 and t-2 2.628 (1.065) 1.938 3.228

Control variables

Size Log (total asset) 6.475 (0.685) 6.027 6.995

RoA Profit/Total asset 0.019 (0.058) 0.013 0.028

Loans Loans/Total asset 0.515 (0.210) 0.403 0.658

Funding Short-term funding/Total asset 0.144 (0.154) 0.031 0.207

Cost/Income Cost-to-income ratio 0.300 (0.481) 0.000 0.432

Divers Index of income diversification, defined following Stiroh(2004) as {SH'net + SHNON) 'where SH NET SH - NON NET NET + NON NON NET + NON NET=net interest income; NON = non interest income 0.364 (0.131) 0.306 0.465

RPH Dummy=l if for a bank the ratio of its regulatory to actual capital belongs to the top 25 percentile of the distribution, else zero 0.255 (0.436) 0 1

RPL Dummy=l if for a bank the ratio of its regulatory to actual capital belongs to the bottom 25 percentile of the distribution, else zero 0.253 (0.435) 0 1

MREG Dummy=l if a bank is listed, else zero 0.679 (0.467) 0 1

Islamic Dummy=l if a bank is Islamic, else zero 0.447 (0.497) 0 1

Crisis Dummy=l for 2009, else zero 0.061 (0.239) 0 1

Table 3: Univariate tests of differences across bank characteristics

Full sample Conventional banks Islamic banks Differences In mean (t-test) Listed banks Unlisted banks Differences In mean (t-test)

Panel A: Dependent variable Mean SD Mean SD Mean SD Mean SD Mean SD

Ln (1+Z) 2.628 1.065 2.796 0.941 2.272 1.216 -6.831*** 2.652 0.979 2.552 1.305 -1.109

CAR 0.101 0.163 0.127 0.112 0.069 0.206 -6.909*** 0.106 0.127 0.092 0.221 -1.307

Panel B: Independent variables

Size 6.476 0.685 6.631 0.608 6.134 0.722 10.995*** 6.563 0.617 6.188 0.808 -6.775***

RoA 0.019 0.058 0.018 0.035 0.019 0.089 -0.060 0.022 0.039 0.009 0.096 -2.046**

Funding 0.144 0.154 0.148 0.135 0.136 0.189 1.002 0.125 0.123 0.207 0.215 5.752***

Divers 0.364 0.131 0.397 0.092 0.289 0.169 10.702*** 0.379 0.115 0.314 0.164 -5.801***

Cost/ Income 0.300 0.481 0.315 0.205 0.281 0.683 1.307 0.320 0.424 0.258 0.582 -2.192**

RPH 0.255 0.436 0.270 0.445 0.197 0.399 1.964** 0.262 0.439 0.232 0.424 -0.759

RPL 0.253 0.435 0.201 0.401 0.454 0.499 -5.789*** 0.209 0.407 0.419 0.495 4.881***

Panel C: Non-crisis

Ln (1+Z) 2.656 1.067 2.281 1.221 2.823 0.946 -6.669*** 2.679 0.987 2.577 1.300 -1.056

CAR 0.096 0.155 0.125 0.113 0.061 0.189 7 849*** 0.103 0.129 0.081 0.199 -2.243**


Ln (1+Z) 2.362 1.017 2.479 0.829 2.221 1.198 1.223 2.355 0.839 2.377 1.342 0.084

CAR 0.177 0.245 0.158 0.083 0.200 0.355 0.782 0.139 0.086 0.262 0.414 1.636

Table 4: 3SLS estimation of capital and risk

Model 1 Model 2 Model 3

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


Control variables

Size -0.102 (0.012)*** -0.089 (0.067) -0.092 (0.012)*** -0.075 (0.068) -0.106 (0.012)*** -0.099 (0.069)

RoA 0.939 (0.184)*** 0.953 (0.178)*** 0.963 (0.182)***

Funding -0.852 (0.257)*** -0.817 (0.256)*** -0.831 (0.259)***

Divers 0.847 (0.297)*** 0.763 (0.298)*** 0.846 (0.299)***

Cost/Income 0.099 (0.124) 0.099 (0.123) 0.092 (0.131)

Regulatory pressure

RPH -0.040 (0.014)*** 0.075 (0.078) -0.013 (0.036) -0.123 (0.215) -0.007 (0.038) -0.069 (0.216)

MREG -0.035 (0.014)*** -0.192 (0.081)*** -0.049 (0.019)*** -0.023 (0.198) -0.027 (0.021) -0.114 (0.201)

Adjustment coefficients

CAP (t-1) -0.662 (0.035)*** -0.491 (0.042)*** -0.539 (0.044)***

Risk (t-1) -0.455 (0.034)*** -0.430 (0.059)*** -0.456 (0.059)***

Response of endogenous variables

ACAP 0.651 (0.282)*** 0.653 (0.266)*** 0.683 (0.213)***

ARISK -0.006 (0.013) -0.005 (0.013) -0.084 (0.023)***

Interaction terms

RPH*CAP(t-l) -0.251 (0.229) -0.299 (0.238)

RPH*RISK(t-l) 0.075 (0.076) 0.057 (0.076)

MREG*CAP (t-1) -0.381 (0.064)*** -0.346 (0.066)***

MREG*RISK (t-1) -0.063 (0.071) -0.026 (0.072)

RPH*ARISK 0.027 (0.016)*

MREG*ARISK 0.076 (0.021)***

RPH*ACAP 2.623 (1.371)**

MREG*ACAP -0.577 (0.562)

Dummy variables

Islamic 0.040 -0.134 0.037 -0.141 0.038 -0.125

(0.015)*** (0.089) (0.015)*** (0.091) (0.015)*** (0.089)

Crisis -0.003 0.342 -0.057 0.363 -0.002 0.245

(0.161) (0.937) (0.157) (0.939) (0.162) (0.733)


Period 1996-2011 1996-2011 1996-2011 1996-2011 1996-2011 1996-2011

N.Obs 683 683 683 683 683 683

R-squared 0.451 0.328 0.479 0.326 0.461 0.334

Standard errors in brackets

***, *** and * denote statistical significance at 1, 5 and 10%, respectively

Table 5: 3SLS estimation of capital and risk - Robustness

Model 1 Model 2 Model 3 Model 4

(1) (2) (3) (4) (5) (6) (7) (8)


RPH*(1-MREG) -0.083 (0.032)*** 0.154 (0.181)

RPH*MREG -0.085 (0.021)*** -0.128 (0.126)

(1-RPH)*MREG -0.052 (0.017)*** -0.081 (0.103)

RPL*(1-MREG) 0.229 (0.027)*** 0.008 (0.154)

RPL*MREG 0.115 (0.022)*** -0.043 (0.134)

(1-RPL)*MREG 0.016 (0.018) -0.155 (0.109)

RPH*(1-MR EG)* Crisis -0.004 (0.094) -1.063 (0.488)**

RPH*MREG*Crlsis 0.025 (0.084) -0.445 (0.505)

(l-RPH)*MREG*Crlsis 0.024 (0.076) -0.338 (0.459)

RPL*(1-MR EG)* Crisis 0.364 (0.081)*** 0.086 (0.489)

RPL*MREG*Crlsis 0.043 (0.035) 0.091 (0.058)

(1-RPL)*M REG* Crisis -0.013 (0.063) -0.113 (0.384)

ACAP*lslamlc -6.672 (1.658)***

ACAP*lslamlc*Crisls -0.655 (0.296)**

ARISK*lslamlc -0.445 (0.698)

ARISK*lslamlc*Crisls 0.035 (0.098)

ACAP 1.128 (0.316)*** 1.136 (0.293)*** 1.221 (0.316)*** 6.533 (1.674)***

ARISK -0.047 (0.055) -0.065 (0.054) -0.035 (0.055) -0.193 (0.715)




Period 1996-2011 1996-2011 1996-2011 1996-2011 1996-2011 1996-2011 19962011 1996-2011

N.Obs 683 683 683 683 683 683 914 914

R-squared 0.405 0.142 0.441 0.141 0.437 0.143 0.452 0.228

Standard errors in brackets

***, *** and * denote statistical significance at 1, 5 and 10%, respectively