Impact of Bank Regulation and Supervision on Bank Soundness: An Analysis of 12 Eurozone countries’ banks

(1)

Impact of Bank Regulation and Supervision on Bank Soundness:

An Analysis of 12 Eurozone countries’ banks

Master Thesis

MSc Business Administration: Finance

Irakli Iluridze

S1546813

Supervisor: Prof. Dr. Kasper Roszbach

Co-supervisor: Dr. Henk von Eije

Groningen

November 2011

faculty of economics and business

economics, econometrics & finance

university of

(2)

Abstract

This thesis examines the impact of bank regulation and supervision on bank soundness using a data set of 334 banks in 12 Eurozone countries (the first members of EMU) during the period of 2005 to 2009. Moody’s bank ratings and Z-scores are applied as proxies for bank stability, while regulation on the banking sector is measured by different indices constructed from the World Bank database. The results suggest, after controlling for a number of bank-specific variables, a negative link between a higher level of regulation and banking stability, as measured by bank ratings. Moreover, this study finds that two individual components of regulation - monitoring and information disclosure requirements (positively) and supervisory power (negatively) - are significantly related with bank soundness. Furthermore, the results imply that government-owned banks are vulnerable, whether bank soundness is measured using Moody’s ratings or using Z-scores. Finally, no evidence has been found that foreign banks are less sound than domestic ones.

Keywords: Bank Soundness, Regulation and Supervision, Bank Ratings, Ownership,

Eurozone

(3)

1. Introduction

Banks are probably the most important financial intermediaries in an economy because of the role they play as a provider of liquidity, a monitoring service and a producer of information. Banks in their daily operations are exposed to a wide array of risks which, if not managed and controlled, can result in disastrous consequences for the economy as a whole. Therefore, the financial sector is subject to much closer government supervision than, say, the manufacturing industry.

A key role of banking regulation and supervision is to prevent financial crises or to mitigate their negative effects. In Europe, banking regulation and supervision deals with how public authorities regulate and control their banking systems. The term “regulation” refers to the rules that govern the behaviour of financial intermediaries, and “supervision” refers to the monitoring and enforcement of these rules. Supervision is currently more of a national prerogative, whereas regulation is mostly legislated at the European Union (EU) level. Up until the 2007-2008 financial crisis, there was a clear trend to separate supervision from central banking, but now many believe that central banking and banking supervision should be more closely connected. Macro-prudential supervision (systemic supervision) and micro-prudential supervision (supervising individual banks) are seen by many as dual functions that need a strong institutional link.

What exactly constitutes good supervision? To answer this question, in 1997 a group of representatives of bank supervisors from advanced countries1 – The Basel Committee on Banking Supervision – issued the Core Principles for Effective Bank Supervision (BCPs), a document summarizing best practices in the field. They revised it in 2004 under the name of Basel II Accords. Unlike the United States, all Eurozone countries have adopted and

completely implemented the Basel II Prudential framework2.

1_{The countries represented in the Committee are Belgium, Canada, France, Germany, Italy, Japan, Luxembourg,}

the Netherlands, Spain, Sweden, Switzerland, the United Kingdom and the United States. The Committee consults widely with supervisors from non-member countries.

2

(5)

The financial crisis following the failure of Lehman Brothers in September of 2008 exposed significant weaknesses in the financial system’s regulatory and supervisory framework. Recently, the Basel Committee on Banking Supervision approved a package of proposals, under the name of “Basel III”, to strengthen capital and liquidity regulations in order to promote a more resilient banking sector3.

As banks operate in one of the most heavily regulated environments, research on the impact of banking regulation and supervision on bank soundness has long attracted both theoretical and empirical interest. Economic theory provides conflicting predictions about this concern (Barth et al., 2004a). Some studies argue that government regulation of the banking sector, such as strict capital standards, limits on bank activities, and deposit insurance schemes, prevent excessive risk-taking behaviour by banks thereby facilitating the stability of the banking system (e.g. Demirguc-Kunt et al., 2004; Naceur and Kandil, 2009). Moreover, Beck et al. (2006) support this view and suggest that powerful supervisors have the capabilities to improve the functioning of the financial system by directly overseeing, regulating, and disciplining banks.

Others, however, do not share this optimistic view on regulation. For example, Djankov et al., (2002) argue that governments implement regulations in a manner that supports their political constituencies. According to their view, countries with powerful official supervisors and restrictive regulations will tend to have higher levels of corruption without any

corresponding improvement in bank performance or stability. Further, Barth et al. (2001) indicate that there is no evidence that regulations on banking activities reduce the likelihood of suffering major banking crises. With respect to the power of supervisors, some authors suggest that it can be harmful to banking development (see also Barth et al., 2003) and that it is also negatively associated with overall bank soundness (see also Pasiouras et al., 2006). Finally, Laeven and Levine (2009) argue that strict banking regulations are associated with an increase in banking risk, instead of acting as a device to reduce it. However, the current debate on this topic has not yet reached a consensus and the results of the empirical studies are still ambiguous.

3

(6)

Existing literature investigates this introduced relationship between banking regulation and bank soundness in a wide variety of countries and regions worldwide (e.g. Barth et al., 2004a; Pasiouras et al., 2006; Cihak and Tieman, 2008) and in individual states (e.g. Naceur and Kandil, 1999; Gaganis and Pasiouras, 2009). Only a few studies address the effects of regulation on the banking industry in European Union member states (e.g. Pasiouras et al., 2011; Uhde and Heimeshoff, 2008). The latter study explores country specific effects among Western and Eastern European states. The authors conclude that the banking sectors in countries with highly regulated environments are less competitive and highly concentrated. Since the introduction of a common currency, the European banking system has undergone many changes, such as privatization and consolidation. The rapid wave of mergers and acquisitions resulted in a large presence of foreign banks in European Union countries (e.g. see Hutchison, 2002). It is therefore interesting to see how these changes affect the overall stability in the banking sector. While the effect of ownership on efficiency has been widely researched (e.g. Lensink, 2008, who finds a negative relationship between foreign ownership and bank efficiency), there is much less literature on what effect the presence of foreign banks has on financial stability. Foreign banks may bring better banking practices that improve the safety of the banking system (Pasiouras and Kosmidou, 2007). However, a high level of competition might impede stability (Berger et al., 2004).

Another interesting fact about this set of countries is that EU member states had substantially liberalized their domestic financial systems by the early 1990s, which caused a wave of privatization in the banking sector (Buch and Heinrich, 2002). However, during the 2007-2008 financial crisis many distressed banks were bailed out or nationalized by their governments (e.g. Northern Rock – in Great Britain; ABN-AMRO – in The Netherlands; Anglo-Irish Bank – in Ireland). It is therefore interesting to investigate how government ownership of banks is associated with financial soundness. Most of the empirical literature supports the view that government-owned banks are inefficient institutions and that they are more fragile than their private counterparts (Shleifer and Vishny, 1998; Barth et al., 2004a).

(7)

as the Z-score (Demirgüç-Kunt and Detragiache, 2010 among others). This thesis applies both indicators of bank stability to explore the above mentioned relationship.

The aim of this thesis is to investigate what is the effect of banking regulation and supervision on banking soundness in the 12 original Eurozone countries4. Next, this study examines which separate elements of the regulatory framework are most closely related to financial stability. In addition, this paper empirically answers the following two questions: how is government ownership related to safer banks and how is the safety net of a country influenced by the presence of foreign banks?

This thesis complements previous studies in several different aspects. To begin with, it provides additional evidence by analyzing the impact of regulation and supervision on bank stability specifically in Eurozone countries. The International Monetary Fund (1999) views the macroeconomic environment facing the European Monetary Union (EMU) member states banks as particularly risky. Due to increasing integration, banks have become more

interconnected with each other across borders and this could imply that future financial crises will tend to extend beyond national banking systems. In the absence of a fully centralized regulatory system (in the EMU regulation is decentralized to the national level), increasing cross-border involvement could make the European banking sector as a whole more

vulnerable to crises (Berger and Hefeker, 2008).

Next, while previous scholars have used data from different countries from different regions of the World (e.g. Demirgüç-Kunt and Detragiache, 2010; Pasiouras et al, 2006), it is the first study to use data from a set of countries operating within the same macroeconomic

environment: the EMU. Demirgüç-Kunt et al. (2008) argue that the financial soundness of a bank is significantly affected by macroeconomic variables, such as inflation rate and

exchange rate. By focusing solely on the Eurozone area, where such variables are the same, this thesis examines the effects of regulation on financial stability independently from the external environment.

Finally, most of the papers examining the impact of bank regulation and supervision on bank stability use country-level data (e.g. Barth et al., 2004a and Beck et al., 2006).In contrast, this

4

(8)

thesis focuses on the soundness of individual banks. This study is certainly not the first to investigate the impact of bank regulation and supervision using bank-level data (see, e.g. Demirgüç-Kunt et al., 2008; Klomp and De Haan, 2010). However, while most of these other empirical studies focus on the different aspects of the regulation framework separately (e.g. Pasiouras et al., 2006; Klomp and De Haan, 2010), this thesis applies to average aggregate

index (among others) to measure the total effect of regulation on bank soundness.

(9)

2. Literature review

It is quite remarkable that a limited number of papers have explored the impact of bank regulation and supervision on bank soundness.This probably reflects the complexity of measuring bank regulation and supervision. Overall, two sources of information have been used to construct proxies for bank regulation and supervision.

Some studies use an index measuring the extent to which countries comply with the Basel Core Principles (BPC). For example, Sundararajan et al. (2001) use a sample of 25 countries to examine the relationship between an index of compliance and bank soundness, measured by non-performing loans. They report that BCP compliance is not a significant determinant of bank fragility. Podpiera (2004) extends the set of countries to 65 and finds that better BCP compliance lowers non-performing loans. More recently, using a sample of rated banks, Demirgüç-Kunt et al. (2008) have found a positive relationship between financial soundness and the overall index of BCP compliance. However, this result is not robust because it is sensitive to controlling for the institutional quality of the country and to the exclusion of outliers. Later, Demirgüç-Kunt and Detragiache (2010) explored whether BCP compliance affects bank soundness (measured by Z-score) using data for 3,000 banks from 86 countries. Their results do not support the hypothesis that better compliance with BCPs is associated with sounder banks.

Using the compliance with the BCPs as a measure of bank regulation and supervision has some shortcomings. First, it inevitably contains an element of subjectivity because the evaluation of compliance with principles reflects the judgment of the assessors (Demirgüç-Kunt and Detragiache 2010). Next, the BCP compliance indicator may be weakly associated with bank soundness, because it proxies for the overall quality of the institutional and macroeconomic environment (Demirgüç-Kunt et al., 2008). Finally, BCP compliance does not show how restrictive rules and regulations are, because it focuses on implementation rather than on the quality or enforcement of the law (Cihak and Tieman, 2008).

(10)

2004b) and measures the presence or absence of a series of regulatory features. Using this survey, Barth et al. (2004a) empirically investigate the impact of different dimensions of bank regulation on bank stability and performance. They show that regulatory and supervisory practices that force accurate information disclosure, and empower private sector monitoring of banks, work best to promote bank performance and stability. However, they also suggest that government restrictions on banking activities and entry into domestic banking negatively affect banking development and bank efficiency. Later, Barth el al. (2008) found that official supervisory oversight, disciplinary powers and tightening capital standards do not improve bank efficiency, nor do they lower banking system fragility. Further, Pasiouras et al. (2006) use this survey to construct indicators of bank regulation and supervision. Employing bank level data from 71 countries and 857 banks, they discover that various dimensions of bank regulation and supervision have a significant impact on bank ratings. Recently, Klomp and De Haan (2010) examine the impact of bank regulation and supervision on banking risk, using data for more than 200 banks from 21 developing countries. Their findings suggest that the effect of regulation and supervision differs across banks: most indicators of bank regulation and supervision do not have a significant effect on low-risk banks, while they do affect high-risk banks.

(11)

3. Data and variable description

3.1 Data sources and description

Data on regulatory and supervisory features come from assessments (surveys) carried out by the IMF and the World Bank in the beginning of 2009. Some of these assessments are public information and can be found on the institutions’ websites. Moody’s financial strength ratings have been compiled from the Bureau Van Dijk’s Bankscope database (henceforth

Bankscope). Bankscope reports the bank’s current rating, the last date in which the rating was revised, and the rating that prevailed before then, however, it does not report any other

historical rating information. Data from bank-level financial statements for the years 2005-2009 were obtained from Bankscope and the information about bank ownership is also collected from Bankscope and other miscellaneous sources.

The sample covers 493 rated banks (all of those with a Moody’s rating) from 12 European Union member states5, which are participants of the European Monetary Union (EMU). The original sample (493 banks) is filtered by excluding 63 banks for which not all data are available for the observation period of 2005-2009. Moreover, banks that are missing values for more than one of the balance sheet variables (total assets, equity, net income, deposits, liabilities, cash) for all observation years are dropped. Overall 54 banks fall in this category. In addition, banks that disappear over time due to a failure or because a merger & acquisition6 are deleted as well. The remaining sample consists of 334 banks, of which 290 have a

domestic owner and 94 a foreigner owner. The government controls 55 banks.

Table 1. Descriptive statistics (all countries, for 2005-2009, five year average) for data used for calculating regression variables (LogAssets, Total equity-to-total assets, ROAA, Deposit-to-Liabilities, Liquid assets-to-deposits) and Z-score.

Financial statement variable

Banks Mean Median St. deviation Maximum Minimum

Total Assets 334 81452706,23 12259580,00 212912529,64 1799964500,00 448720,00 Equity 334 3435144,19 745460,00 8229836,75 63883300,00 20460,00

5

11 countries which were joined to EMU in 1999: Austria, Belgium, Finland, France, Germany, Ireland, Italy, Luxembourg, The Netherlands, Portugal and Spain. Greece joined in 2001.

6

(12)

Cash & Marketable securities 334 34663654,62 7307770,00 77623529,09 625916696,00 72764,00 Net Income 334 243303,76 33660,00 928945,89 9147700,00 -2842400,00 Deposits and Short-term Fundings 334 42324698,22 7740800,00 98816932,92 676804600,00 86880,00 Liabilities 334 78013946,98 11559620,00 205600606,62 1764422750,00 389680,00

Note: all the data presented in thousands of Euros.

The bank financial characteristics are five-year averages over the period [t, t-4], where t is the year of survey evaluation, which corresponds to the year 2009. For several banks, the

variables are averaged over a shorter time period because the data are missing or not available for some years 2005-2009. Table 1 displays descriptive statistics for the accounting variables adopted in the study.

Because data on bank rating as well as on regulation are available for only one point in time, our study cannot rely on time series variation. This limitation forces us to prefer an average data set over a panel data set. Moreover, the panel data is very unbalanced due to missing observations.

Table 2. Number of sample banks by countries and size, 2005-2009 Country Number "Large banks"

(average total assets more than

(13)

Total 334 78 108 148

The sample is very unbalanced because Eurozone countries have banking systems of vastly different sizes. Some states are represented by only a handful of banks, and others by over 50 (Table 2). In particular, Italian banks account for 32 percent of this sample (107 banks), while Greece has only 9 rated banks. Moreover, Table 2 provides information on the distribution of the banks by size. Banks which have more than 50 billion euro in average total assets

(henceforth “large banks”) are represented by 78 banks. Overall 148 banks in the sample have less than 10 billion euro in average total assets (henceforth “small banks”).

The banks that are missing values for only one of the balance sheet variables (total assets, equity, net income, deposits, liabilities, cash & marketable securities) for all observation years are included in study sample of 334 banks. In order to compute the missing observations we apply the Expectation-Maximization algorithm (abbreviated as the EM algorithm) of

Dempster et al. (1977). We will first estimate the parameters on the basis of the data we do have and the missing data based on those parameters. Then we will re-estimate the parameters based on the filled-in data, and so on. Subsequently, we use those estimates to solve for the regression coefficients, and then estimate the missing data based on those regression

coefficients. This is the estimation step of the algorithm. In order to avoid the underestimating error in choosing our estimates, the maximization step adds a bit of error to the variances it estimates, and then uses those new estimates to impute data, and so on until the solution stabilizes. At that point we have maximum likelihood estimates of the parameters, and we can use those to make the final maximum likelihood estimates of the regression coefficients. We provide below a technical summary of this method, while a more detailed description can be found in Schafer and Olsden (1998).

Given a statistical model consisting of a set of observed data (Χ ), a set of unobserved (latent) data or missing values (Ζ ), and a vector of unknown parameters (θ), along with a likelihood function L(

θ

;Χ,Ζ)= p(Χ,Ζ

θ

)the maximum likelihood estimate (MLE) of the unknown parameters is determined by the marginal likelihood of the observed data:

(

Χ

)

=

( )

Χ

=

∑

(

Χ

Ζ

)

z

p

(14)

The EM algorithm is an iterative procedure that finds the MLE of the parameter vector by repeating the following two steps:

I. The expectation E-step: Given a set of parameter estimates the E-step calculates the

expected value of the log likelihood function, with respect to the conditional distribution of

Z

given

X

under the current estimate of the parameter

θ

( )t :

( )

(

)

E

( )

[

L

(

X

Z

)

]

Q

t X Z t

,

;

log

,

θ

=

_θ

(2)

II. The maximization M-step: Given a complete-data log likelihood, the M-step finds the

parameter estimates to maximize the complete-data log likelihood from the E-step.

(t )

(

( )t

)

Q

θ

+1

=

arg

max

_θ

(3)

EM algorithm is applicable if:

1. The missing values (

Z)

are discrete, drawn from a fixed number of values, and there is one

latent (missing) variable per observed data point.

(15)

3.2. Variable definition and description

3.2.1 Dependent variables

Moody’s bank rating

The dependent variable is Moody’s bank rating. Moody’s rates banks’ financial strength on a 15-point scale, ranging from E- (least sound) to A+ (most sound). The ratings are transformed into a numerical index by assigning values to each of these ratings from 1 (E-) to 15 (A+).

Moody’s financial strength rating is a comprehensive measure of the ability of a bank to meet its obligations to its creditors. This measure is constructed using quantitative as well as qualitative information about a bank and its operating environment. The main advantage of Moody’s ratings is that they aim to capture bank solvency independent of the safety net, so that cross-country differences in the banking system, which are difficult to observe and measure, should not affect the results (Demirgüç-Kunt, A. et al., 2008).

Because the assessment reflects the judgment of the analysts, it probably contains an element of subjectivity. This is the main disadvantage of bank ratings. It can be argued that analysts may be better disposed toward banks that provide more accurate and timely information even if those banks are not necessarily more sound (Demirgüç-Kunt and Detragiache, 2010). Furthermore, after the recent crisis, the credibility of credit ratings as indicators of bank risk has diminished.

Z-score

This study uses Z-score as an alternative dependent variable in order to measure the bank’s financial stability. This variable can be interpreted as the bank’s distance to default (Boyd and Runkle, 1993). The Z-score is defined as follows:

(16)

where ROA is the rate of return on assets (return on average assets – ROAA is used instead),

E A

is the ratio of equity to total assets and σ (ROA) is an estimate of the standard deviation of the rate of return on assets. In the regressions, we actually use ln(1+Z-score) as the dependent variable in order to smooth out the higher Z-score values.

The Z-score is a very popular measure of bank risk (Demirgüç-Kunt and Detragiache, 2010 among others). It increases with a higher bank profitability and capitalization ratio and decreases with greater returns’ volatility (standard deviation). Therefore, higher values of Z-score imply greater financial strength. Empirical studies have shown the usefulness of this measure. For example, Čihák (2007) finds that distressed banks are characterized by significantly lower Z-scores than other banks. Furthermore, the Z-score is easy to calculate and it combines a bank’s capital, profitability as well as the risk it faces. In addition, unlike ratings, this measure is constructed from accounting information and does not rely on the subjective judgment of the rating agencies’ analysts. Finally, while the Z-score has its limitations, it is a better cross-country comparability measure of fragility than other balance sheet variables, including NPLs, loan spread and interest margin.7

3.2.2 Explanatory variables: Regulation and supervision variables

In this thesis, we use the survey data of The World Bank to compute the aggregate regulation index. The survey is constructed according to the approach of Barth et al.(2004b) and consists of polar, multiple-choice and open questions. This study use 69 yes/no questions, which are applicable to 25 BCPs. The answer “Yes” means the presence of a regulatory feature and takes the value 1. The negative answer (“no”) means the absence of a regulatory feature and takes the value 0. In constructing this score, these questions are classified into five groups following the grouping by Pasiouaris et al. (2006)8: 1) market entry regulations; 2) capital and

7

Because different countries have different accounting reporting rules and market structure, NPLs, loan spreads and interest margins are notoriously difficult to compare across countries.

8

(17)

liquidity requirements; 3) regulations on activities restriction; 4) monitoring and information disclosure requirements; and 5) disciplinary power of the supervisory agency. Then we compute regulation indices for each subgroup separately by aggregating the numerical values for questions and then normalize the sum to obtain a score that varies between zero and one. To obtain an overall index of regulation, the values of individual indices are averaged.

Since Table A1 in the Appendix provides information on the data, sources, and specific survey questions used to construct these regulatory variables, we only briefly define them here in the text and discuss their importance in relation to the previous studies.

Market entry regulations

This variable measures the degree of specific legal requirements for obtaining a licence to operate as a bank and the limitations placed on the ability of foreign banks to enter the domestic banking industry. The measure is based on 11 yes/no questions. A positive answer takes the value 1, otherwise 0. Then the sum of the values is divided by the number of questions. A higher value indicates greater stringency.

(18)

Capital and liquidity requirements

This measure refers to capital and liquidity regulations and takes various questions into account, such as: can regulatory capital include borrowed funds; are the sources verified by the regulatory authorities; are risk elements and value losses considered in calculating

regulatory capital; and is there a minimum liquidity requirement? The variable is based on 16 yes/no questions and calculated by adding 1 if the answer is “yes” and 0 if not. Then the sum of the scores is divided by the number of questions. The index can therefore vary between 0 and 1, with higher values indicating stricter capital and liquidity requirements.

One strand of the empirical literature on bank regulation emphasizes the positive features of capital and liquidity requirements (e.g. Dewatripont and Tirole, 1994; Keeley and Furlong, 1990). Capital serves as a buffer against losses and hence failure. Fernandez and González (2005) find that stringent capital requirements reduce banking risk. Furthermore, Barth et al. (2004a) report that tighter capital requirements are associated with fewer non-performing loans. However, other studies provide conflicting predictions as to whether the imposition of capital and liquidity requirements has positive effects. For instance, Besanko and Kanatas, (1996) and Blum (1999) argue that capital requirements may increase the risk-taking behaviour of banks. Gorton and Winton (2000) report that raising liquidity requirements forces banks to supply fewer deposits, which limits the liquidity-providing role of banks. Santos (2001) argues that a higher capital requirement ratio could be negatively associated with bank development, adversely affecting credit expansion.

Regulations on activities restriction

(19)

it to obtain an index that varies from zero to one. Higher values of this variable indicate more restrictions on banks’ activities.

Moral hazard is the main theoretical reason for restricting bank activities because it

encourages risky behaviour. Banks will have more opportunities to increase risk if they are allowed to engage in a broad range of activities (Boyd et al., 1998). Moreover, large and complex banks may become so politically and economically powerful that they become “too big to discipline.” According to these arguments, governments can promote bank soundness by imposing restrictions on banks’ activities.

However, the empirical evidence generally indicates that restricting bank activities has negative consequences. For example, Barth et al. (2001) find that greater regulatory restrictions on bank activities are associated with a higher probability of suffering a major banking crisis, and lower banking sector efficiency. Later, Barth et al. (2004a) report that restricting bank activities has adverse effects on bank development and stability and they outline several alternative theoretical reasons for allowing banks to participate in a broad range of activities.

Monitoring and information disclosure requirements

This variable measures the extent to which bank information is released to the public and the requirements concerning external auditing. The index is based on 16 yes/no questions. Positive answers take the value of 1 and otherwise 0. Results are summed and the total score is divided by the number of questions. The measure takes values between 0 and 1 and higher values indicate greater monitoring and information provision requirements.

Regulations concerning private-sector monitoring of banks and information disclosure requirements are designed to improve bank performance and stability. Fernandez and

(20)

banks are associated with better banking-sector outcomes (lower net interest margins and fewer non-performing loans).

Disciplinary power of supervisory agency

This variable represents the ability of supervisors to exercise power and to get involved in banking decisions. This measure is calculated on the basis of 22 questions regarding whether the authorities can take specific actions (e.g. prompt corrective action, restructuring, declaring of insolvency) to prevent and correct problems in the banking industry. Positive answears get a value of 1, otherwise its get a value of 0. Results are summed and divided by the number of questions. The index can range between 0 and 1, with higher values indicating greater power of authorities.

Barth et al. (2004a) underline various theoretical advantages and disadvantages of granting broad powers to supervisors. For example, they mention that strong official supervision can prevent managers from engaging in excessive risk-taking behaviour, thereby contributing to bank development, performance and stability. On the other hand, powerful supervisors may use their powers to gain political influence and attract campaign donations. Under these circumstances, powerful supervision might be positively related to corruption or impede bank operations. However, they find no strong association between bank performance and official supervisory power.

Table A2 in the Appendix shows the value of aggregate as well as separate regulation indices by countries. The aggregate index is an average of 5 separate regulation indices. As expected, due to efforts to harmonize bank regulation and supervision across the European Union, the values of the aggregate regulation index across the Eurozone member states are closer to each other: Greece and Portugal have highest degree of regulation (both 0.742), while Austria (0.600) and The Netherlands (0.557) have the least restrictive regulations.

Taking into consideration the mixed results of the literature that examines the impact of

(21)

3.2.3 Bank specific variables

In this study we use five bank specific variables that might affect the financial strength of a bank: size, profitability, capital adequacy, liquidity and founding sources and two dummy variable for ownership (foreign and government). We briefly define these variables in the text that follows, while more details can be found in the Appendix Table A3.

The first variable is bank size. It is based on the total assets of a bank and given the positive skewness in this measure. We take the natural logarithm of total assets to be a proxy for bank size. The bigger the bank is, the higher its degree of monopoly power. This power enables banks to receive high profits and thereby improve their financial soundness (Demirgüç-Kunt and Detragiache, 2010).

The second variable is related to the profitability of a bank. On the one hand, declining trends in profit indicators may signal problems regarding the sustainability of financial institutions. On the other hand, unusually high profits may signal excessive risk-taking (Klomp and De Haan, 2010). We measure profitability by the return on average assets (ROAA), which is commonly used as a proxy of profitability of a financial institution (e.g. Shehzad et al., 2010).

The third measure, capital adequacy, is captured by the total equity to total assets ratio. Capital serves as the last line of defence against the risk of a bank’s insolvency and protects to some degree depositors, creditors and investors (Le Bras and Andrews, 2004). Therefore, better capitalized banks might tend to be financially sounder.

The forth variable is liquidity. As the case of Northern Rock has shown, insufficient liquidity may threaten the survival of a bank, even to be ex-ante solvent9. We use the liquid assets (cash & marketable securities) to deposits ratio as a proxy for liquidity. This variable measures the risk of not having a sufficient reserve of cash to cope with withdrawal of

9

(22)

deposits (Brunnermeier et al., 2009). Higher values denote higher liquidity and reflect the financial health of the institution.

The fifth variable is funding source. One bank might rely relatively more on deposits as a source of funding than, for instance on the bond market. This difference would be captured by the deposit-to-liability ratio. Banks that have higher deposit-to-liability ratios are rated

significantly lower. Thus, funding sources with a short-term maturity such as deposits tend to be riskier than bonds that have a long-term maturity (Pasiouras et al., 2006; Klomp and De Haan, 2010).

Furthermore, two additional variables are added in order to measure the effect of ownership on financial soundness. We include two dummy variables: state and foreign ownership. Banks are considered state-owned if the government has a controlling share (more than 51 percent) or if it is the largest shareholder (at least 20 percent), and the same criteria are used to determine if a bank is foreign-owned (La Porta et al. 2002).

As Barth et al. (2004a) point out, economists have conflicting views about the impact of government ownership on banks. One view holds that governments can help banks to overcome capital-market failures and promote socially desirable investments. Others, in contrast, support alternative arguments and take a less optimistic view on government ownership. Recent empirical evidence seems to support the later view. Caprio and Martinez (2000) find that government ownership is significantly associated with increases in bank fragility. Barth et al. (2004a) indicate that government ownership is negatively correlated with favourable banking outcomes and positively correlated with corruption.

(23)

4. Methodology

4.1 The econometric model

In order to test the relation between the financial soundness of individual banks and the regulation and supervision of the banking sector in different countries, two types of regressions are run using different measures of financial soundness, as presented earlier. Model 1 is specified as follows:

ij

Rating

=

α

Rating_ij is a dependent variable and is measured by Moody’s bank rating.

j

REGULATION is a variable of interest and measures regulation and supervision.

ij

BankChar is a vector of bank’s accounting characteristics.

Stateij is a dummy variable and takes the value 1 if a bank is owned by the state and 0

otherwise.

Foreignij is a dummy variable and takes the value 1 if a bank has a foreigner owner and 0 oif

not.

ij

ε is a standard error term and is clustered by country.

Model 1 is a cross-sectional regression and because of the ordinal nature of the dependent variable (see section 3.2.1), an ordered probit model has been argued to be appropriate for estimation (5). We discuss very briefly the ordered probit model below while more detailed explanation can be found in Brooks (2008).

(24)

i i i

R

*

=

Χ

β

+

ε

(6)

with











>

≤

<

≤

=

−1 * 2 * 1 1 *

2

1

N i i i i

R

if

N

R

if

R

if

R

µ

β

.

Model 2 is specified very similarly to the previous one:

ij

Ζ

=

α

Z

denotes Z-score for bank i of country j.

The second econometric model uses Z-score instead of bank rating as the dependent variable and estimates parameters with an ordinary least squares (OLS) method. Right-hand variables are the same as in (5) so as to facilitate the comparison of the results. However, capitalization ratio and return on average assets are dropped from the set of control variables since these indicators are now used to compute the left-hand-side variable (see section 3.2.1). In order to smooth out the higher values of the Z-score and avoid falling the dependent variable at zero, ln(1+Z-score) is actually used in the regression.

(25)

5. Empirical results

5.1 Correlation matrixes

Table 3 presents correlations among different regulatory and supervisory variables. The strongest correlation (over 0.50) is between monitoring and information disclosure

requirements, and capital and liquidity regulations. The other correlations are considerably lower, and even insignificant. These correlations suggest that there is enough variation across the variables to use them to investigate the effect of various aspects of regulation and

supervision.

Table 3. Correlations matrix of regulatory variables (334 observations)

Market entry regulations Capital and liquidity requirements Regulations on activity restrictions Monitoring and information disclosure requirements Supervisory power Government- owned Foreign-owned Market entry regulations 1.000 Capital and liquidity requirements 0.377** 1.000 Regulations on activity restrictions 0.489*** -0.005 1.000 Monitoring and information disclosure requirements 0.172* 0.588*** -0.245** 1.000 Supervisory power -0.115 0.077 -0.505*** 0.317** 1.000 Government-owned -0.164* 0.015 -0.136* 0.052 0.158* 1.000 Foreign-owned -0.070 0.012 -0.098 0.208* 0.175* -0.200** 1.000 *** statistically significant at the 1% level **statistically significant at the 5% level * statistically significant at the 10% level

(26)

since the largest correlation is between the return on average assets (ROAA) and market capitalization (0.491), while most of the remaining correlations are below 0.4. Hence, all control variables can be simultaneously included in the model, without any particular concerns about multicollinearity (Judge et al., 1982).

Table 4. Correlations among bank’s financial characteristic variables (334 observations) Aggregate index Log. Total assets Return on aver. assets Equity to total assets Liquid assets to deposits Deposits to total liabilities Governm ent-owned Foreign-owned Aggregate index of regulation 1.000 Log. Total assets -0.085 1.000 Return on aver. assets 0.170* -0.116 1.000 Equity to total assets 0.126* -0.383*** 0.491*** 1.000 Liquid assets to deposits 0.064 0.039 0.061 0.013 1.000 Deposits to total liabilities 0.073 -0.231** 0.060 0.051 -0.037 1.000 Government-owned -0.036 0.101 -0.138* -0.112* -0.014 0.055 1.000 Foreign-owned 0.064 -0.166* 0.040 0.133* -0.038 0.116 -0.200** 1.000 *** statistically significant at the 1% level **statistically significant at the 5% level * statistically significant at the 10% level

5.2. Estimation results

5.2.1. Dependent variable: Moody’s bank financial strength rating

(27)

specifications include 5 separate regulatory variables (market entry, capital and liquidity regulation, activity restrictions, monitoring and disclosure requirement and supervisory power) both with and without the banks’ financial characteristics. Two ownership (dummy) variables (government and foreign ownership of banks) are controlled in all specifications.

Table 5. Impact of aggregate and separate regulation indexes on Moody’s bank ratings (Model [5], ordered probit, 334 observations)

with BFS# control variables (1) without BFS control variables (2) with BFS control variables (3) without BFS control variables (4) Aggregate regulatory Index -2.671 (-2.09)** -3.713 (-2.33)** Market entry regulations -0.865 (-0.98) -0.948 (-0.83) Capital and liquidity

requirements -2.365 (-2.40)** -2.570 (-2.59)*** Regulations on activity restrictions -3.663 (-2.20)** -4.429 (-2.71)*** Monitoring and information disclosure requirements 4.134 (2.35)** 6.011 (3.52)*** Supervisory power -1.740 (-2.72)*** -1.908 (-2.61)*** Log. Total assets 0.172

(2.92)***

0.170 (2.88)*** Return on aver. assets 0.659

(4.44)***

0.611 (4.03)*** Equity to total assets -0.016

(-0.95) -0.012 (-0.74) Liquid assets to deposits 0.035 (1.49) _(1.36)0.033 Deposits to total liabilities -0.964 (-2.56)** _(-2.25)**-0.745 Government-owned -0.775 (-3.05)*** -0.866 (-3.51)*** -0.813 (-3.11)*** -0.909 (-3.56)*** Foreign-owned -0.194 (-1.05) -0.307 (-1.39) -0.358 (-1.08) -0.508 (-1.36) Pseudo R-squared 0.08 0.04 0.10 0.06

# Bank financial statement

*** Statistically significant at 1% level ** statistically significant at 5% level * statistically significant at 1% level; z-values are shown in parentheses.

(28)

We turn next to the separate regulatory variables. It is interesting to know whether the separate indices, which make up the aggregate regulation index, affect bank soundness differently. Capital and liquidity requirement, regulations on activity restrictions, monitoring and information disclosure requirements, and supervisory power are significant in both specifications. These variables have a significant impact on bank ratings at the 1% level when we do not control for the bank level variables. However, after controlling for the bank

characteristics significance drops to the 5% level and the coefficients become a bit smaller. The negative and significant sign of capital and liquidity requirements indicates that banks operating in an environment with lower capital and liquidity requirements are assigned higher ratings. This view is consistent with the argument that excessive capital requirements may impede bank stability (e.g. Blum, 1999; Besanko and Kanatas, 1996; Laeven and Levine, 2009).

Regulations on activity restrictions have a negative and significant impact on bank soundness, indicating that banks operating in markets with higher activity restrictions are assigned lower ratings. This results appear to be consistent with empirical evidence indicating that greater regulatory restrictions on bank activities are associated with lower banking sector efficiency and development (e.g. Barth et al., 2001; 2004a; Pasiouras et al., 2006). Moreover, Barth et al. (2004a) argue that a broad range of activities may enable banks to diversify outside their traditional lines of business and thereby make themselves more stable.

Interestingly, the coefficient of monitoring and information disclosure requirements is

positive in both specifications. This result is in line with the findings of empirical studies that indicate that accounting and auditing requirements are effective devices for reducing bank risk (Fernandez and Gonzalez, 2005; Klomp and De Haan, 2010). Further, our results are consistent with the view that regulations that encourage and facilitate private monitoring are associated with better banking sector stability (Barth et al., 2004a). Thus, high information requirements make banks more transparent, which improves their credibility and results in higher ratings from assessors (Demirgüç-Kunt et al. 2008).

(29)

replacing management, have a negative impact on bank stability measured by bank ratings. This result is in line with the findings of Pasiouras et al.(2006) who explore the negative relationship between bank ratings and the disciplinary power of supervisory agencies. Moreover, the results of Barth et al. (2003) indicate that giving more power to government supervisors is associated with a higher level of non-performing loans, while Barth et al. (2004a) argue that powerful supervision is positively related to corruption and therefore does not improve bank development, performance or stability.

As regards the ownership variables, the government-owned variable is significant in all four specifications, while foreign-owned is not significantly associated with bank ratings.

Government ownership is negatively related to Moody’s financial strength rating, indicating that state-owned banks are considered less stable. This finding confirms the empirical evidence that documents a negative link between government ownership and bank stability and development (e.g. Barth et al., 2001; La Porta et al., 2002; Pasiouras et al., 2006)

However, we do not find any significant relationship between foreign ownership of banks and bank ratings. Our results thus contradict the findings of Pasiouras el al., (2006) which show that foreign-owned banks are positively and significantly related to bank ratings.

Finally, we address the bank financial statement variables. The coefficients of total assets and ROAA are significant and positively associated with bank ratings. These findings indicate that larger and more profitable banks are considered to be more stable. However, the deposit-to-liabilities ratio is negatively and significantly (at 5% level) associated with bank

soundness. Thus, the banks with a lower fraction deposits-to-total liabilities are rated more favourably. This result confirms the view that banks with high deposits-to-liabilities ratios are subject to bank run risk (Brunnermeier et al. 2009).

(30)

5.2.2. Dependent variable: Z-score

As mentioned in section 3 Moody’s ratings are a popular measure of bank financial soundness, because they are prepared by professionals with access to different sources of information. However, ratings are subjective. Analysts may, for example, be better disposed towards banks that provide more detailed and timely information even though those banks are not necessarily more sound (Demirgüç-Kunt et al., 2008). To address this issue, we also measure bank soundness using an alternative and more objective indicator- the Z-score. The results are presented in Table 6.

Table 6. Impact of aggregate and separate regulation indexes on Z-score of individual banks (Model [8], OLS, 334 observations)

with BFS# control variables (1) without BFS control variables (2) with BFS control variables (3) without BFS control variables (4) Aggregate Regulatory Index 1.307 (1.04) 1.626 (1.22) Market entry regulations -1.548 (-0.98) -2.015 (-1.23) Capital and liquidity

requirements 0.548 (0.53) 0.595 (0.58) Regulations on activity restrictions -0.130 (-0.14) -0.227 (-0.27) Monitoring and information disclosure requirements 4.931 (2.79)*** 4.752 (2.76)*** Supervisory power -1.655 (2.17)** -1.533 (2.11)** Log. Total assets -0.141

(-2.55)** -0.091 (-1.99)** Liquid assets to deposits -0.003 (-0.16) -0.016 (-0.66) Deposits to total liabilities 0.128 (0.27) 0.352 (0.77) Government-owned -0.632 (-2.48)** -0.640 (-2.48)** -0.690 (-2.73)*** -0.684 (-2.72)*** Foreign-owned -0.214 (-1.10) -0.088 (-0.46) -0.308 (-1.51) -0.232 (-1.17) Constant 3.342 (2.18)** 0.687 (0.63) 3.414 (2.43)** 0.891 (0.95) R-squared 0.11 0.07 0.13 0.11

# Bank financial statement

(31)

The coefficient of the aggregate regulatory index is positive but insignificant. The same is true when we add to the regression the bank accounting variables. These results are different from what we find in the previous section using Moody’s bank ratings. Similarly, Demirgüç-Kunt and Detragiache (2010) find that the Z-score is not significantly associated with regulatory framework (measured by BCP compliance).

The aggregate regulatory index is the average of 5 separate regulatory variables. Therefore, it could be that, even though the aggregate index does not seem to affect bank soundness, some aspects of the regulation framework might be relevant. In fact, it is possible that the overall (average) index is not significant because of the offsetting effects of its separate components (Demirgüç-Kunt and Detragiache, 2010).

Results are shown in columns 3 and 4 of Table 6. In column 3 the regulatory variables enters only with ownership controls, while in column 4 bank specific variables are incorporated as well (the effect of regulatory variables on Z-score increases). Monitoring and information disclosure requirements are positively and significantly (at the 1% level) related to Z-score, while supervisory power is negatively and significantly associated with financial soundness of the bank. The sign and significance of both indices remain unchanged after controlling for bank specific variables. Therefore, countries which require banks to regularly and accurately report their financial data to the public have sounder banks (Demirgüç-Kunt et al., 2008), and banks in countries where regulators have more power to regulate and intervene in bank

activities tend to be riskier. The latter finding supports the view that bank supervisory systems that empower supervisors do not work well (e.g. Barth et al., 2004a, 2008).

Turning to the ownership (dummy) variables, the results are very similar when examining a relationship between regulation and bank rating in previous section. In all estimations (Table 6), government ownership dummy variable enters significantly with a negative sign, thus implying that state-owned banks are less stable (have lower Z-score) than private ones. However, foreign-ownership of banks does not have significant effect on banking sector’s stability.

(32)

becoming larger) has any direct impact on financial stability. Other bank level variables are not significant.

In conclusion, there is no evidence that overall the regulation index is significantly related to bank’s Z-score. However, their two separate indices – monitoring and information disclosure requirements and supervisory power – are significantly associated with this measure. The signs of coefficients are positive and negative, respectively. Government ownership has a negative relationship with Z-score, while the coefficients of foreign-ownership are

insignificant.

The relationship between bank soundness, measured by bank ratings, and regulation

contradicts the relationship uncovered using Z-score as an indicator of bank stability. In the first case the regulation coefficient is negative and significant, while in the second case the relationship is positive, albeit insignificant. A possible explanation can be following:

Moody’s analysts use quantitative as well as qualitative information in order to attach ratings, while the Z-score is a technical measure, based solely on accounting variables.

5.2.3 Sensitivity analysis

It is possible that regulation and supervision have a different effect on banks that differ in terms of their size. For instance, Klomp and De Haan, (2010) argue that the same regulations may affect banking risk differently depending on the ownership structure and size of a bank. Therefore, we examine the impact of regulation on the soundness of large and small banks.

Table 7. Estimation results - Sensitivity analysis

(33)

requirements Regulations on activity restrictions -2.474 (2.42)** 0.675 (0.58) 0.810 (0.36) -1.548 (-0.76) Monitoring and information disclosure requirements 4.474 (3.04)*** 1.139 (1.86)* 2.322 (2.03)** 0.663 (0.58) Supervisory power -1.877 (-2.33)** -2.043 (-2.26)** 0.154 (0.10) -1.447 (1.68)* Log. Total assets 0.608 (3.96)*** 0.619 (3.94)*** 0.057 (0.42) 0.047 (0.32) -0.100 (-0.74) -0.101 (-0.71) -0.209 (-1.163)* -0.248 (-1.82)* Return on aver. assets 1.843 (4.68)*** 2.074 (4.86)*** 0.189 (1.41) 0.227 (1.62) - - - - Equity to total assets 0.038 (0.61) 0.013 (0.20) -0.007 (-0.43) -0.012 (-0.69) - - - - Liquid assets to deposits 0.013 (0.01) 0.046 (0.24) 0.076 (0.55) 0.048 (0.33) -0.006 (-0.18) -0.002 (-0.06) 0.018 (0.84) 0.012 (0.56) Deposits to total liabilities -1.059 (-2.69)** -1.071 (-2.39)** -1.164 (-2.58)** -0.958 (-2.51)** 0.809 (1.01) 0.939 (1.06) 0.487 (1.09) 0.336 (0.68) Government-owned -0.436 (-2.10)** -0.268 (-2.64)** -0.502 (-2.51)** -0.541 (-2.54)** -0.976 (-2.88)*** -0.929 (-2.50)** -0.544 (-1.02) -0.311 (-1.31) Foreign-owned 0.088 (0.30) 0.150 (0.50) -0.240 (-1.06) -0.256 (-1.05) -0.276 (-1.00) -0.231 (-0.79) -0.334 (-1.52) 0.574 (1.04) Constant - - - - 0.974 (0.28) 1.497 (0.40) 3.623 (2.09)** 4.571 (2.27)** Observations 78 78 148 148 78 78 148 148 (Pseudo) R-squared 0.17 0.18 0.06 0.06 0.17 0.19 0.12 0.19 Method of estimation Ordered probit Ordered probit Ordered probit Ordered probit

OLS OLS OLS OLS

# Banks with 50 billion euro or more average total assets. ## Banks with 10 billion euro or less average total assets.

*** Statistically significant at 1% level ** statistically significant at 5% level * statistically significant at 1% level; z-values and t-values are shown in parentheses.

The first four columns of Table 7 show the results for large vs. small banks when bank soundness is measured by Moody’s bank ratings. We find that the effect of the aggregate regulation index on bank stability is higher for large banks. Looking at the various aspects of the regulatory framework, our results indicate that monitoring and information disclosure requirements have the largest impact on large banks, while supervisory power has the largest impact on small banks. However, only large banks are significantly (and negatively) affected by regulations on activity restrictions, while only small banks are significantly and positively influenced by capital and liquidity requirements.

(34)

banks. Next, stated-owned small banks tend to be more fragile than state-owned large banks. Finally, the results show that foreign ownership of banks is not sensitive to bank size.

In the last four columns of Table 7 we show the results of size sensitivity, using the Z-score as an indicator of financial stability. Focusing only on large banks, the coefficient of the

aggregate regulation index is positive and significant (albeit only at the 5 percent level)10 however, the Z-score is not significantly associated with regulation when using the total sample (see table 6). Thus, regulation seems to have some positive impact on the Z-score of this specific group of banks. The financial soundness of small banks is not significantly affected by the regulatory index.

Turning to the separate variables of regulation, monitoring and information disclosure requirement are positively and significantly associated with the Z-score of large banks while capital and liquidity requirements (with a positive sign), and supervisory power (with a negative sign) have a significant impact on the financial stability of small banks. Regarding the ownership variables, government ownership has a stronger (negative) effect on the stability of large banks than it has on small banks.

In conclusion: our sensitivity results indicate that the effect of bank regulation and

supervision on bank soundness is not only conditional on the financial stability of a bank, but also on the size of a bank.

5.2.4. Robustness of the results

Two main robustness test are performed. First, countries that are represented by more than 40 banks (Italy, Germany, and Spain) are excluded from the sample one by one. In this way, we try to check whether the regression results are overly influenced by big countries. Second, we re-estimate the results for the sample including banks that disappeared over the time period due to failure, merger or acquisition.

10

(35)

Table A4 in the appendix shows the results after a robustness check. The main results from model [5] are confirmed (using bank ratings as dependent variable). The coefficients of the aggregate regulation index are consistently negative and significant. However, after the exclusion of Italian banks, the coefficients of two separate regulation variables- capital and liquidity requirements, and regulations on activity restrictions- became smaller and

insignificant. The significance and sign of the parameter of government-owned variables stay unchanged.

In contrast, results for model [8] have not been found to be robust (using Z-score as the dependent variable). Most estimators provide insignificant results as well as a different sign for the relationship between separate regulatory variables and the Z-score. However, the coefficient of monitoring and information disclosure requirements remains positive and significant in all estimations. The exclusion of the Italian banks from the sample also substantially changes the value and significance of the coefficients of many regressors. The aggregate regulation index is significant and positive, implying a positive relationship between restrictive regulation and bank stability. Moreover, market entry regulation, and capital and liquidity requirements enter in the regression significantly with negative and positive signs respectively.

(36)

6. Conclusion and future directions

An adequate regulation and supervision framework is necessary to ensure a safe and sound banking system. However, previous research has come up with mixed results concerning the effectiveness of bank regulation and supervision in reducing banking fragility.

Using averages of data from 334 banks in 12 European Monetary Union member states (the countries that formed the Eurozone in 2001) over the period of 2005–2009, this thesis

examines the impact of bank regulation and supervision on bank soundness. This paper uses the aggregate regulation index (constructed from the WB and Barth et al., (2001b) database) as a proxy for regulation and two proxies for banking stability are used: the “traditional” (Moody’s) bank rating and the Z-score – a more objective measure. Furthermore, this study investigates the relationship between different aspects of regulation and supervision

framework and bank stability in EMU countries. Finally, this paper also examines whether bank ownership patterns (foreign-owned versus state-owned) are associated with bank soundness.

Using bank ratings as a proxy for bank stability and after controlling for bank specific variables (total assets, return on assets, equity to total assets, liquid assets to deposits and deposits to liabilities), the overall results suggest that higher restrictive regulation tends to impede banking stability in this region. This provides more evidence for the view that a higher degree of regulation adversely affects the risk-taking behaviour of a bank (Laeven and Levine, 2009). Among different aspect of the regulatory framework only monitoring and

information disclosure requirements and supervisory power are significantly and robustly associated with bank soundness. The former aspect has a positive effect on banking stability while the latter is negatively related to soundness.

Next, the results indicate that there is no evidence that the aggregate regulation index significantly affects the Z-score. However, the relationship between bank soundness and separate regulatory indices is remarkably similar to the one uncovered using Moody’s ratings:

(37)

Furthermore, the results also show that government-owned banks are more fragile than their private counterparts. This finding confirms the view that state-ownership of banks can not improve banking sector development or stability (e.g. Caprio and Martinez, 2000; Barth et al., 2004a).

Finally, the results do not suggest that foreign-owned banks are more stable than domestic ones in EMU countries. In fact, overall foreign ownership does not have a significant effect on bank soundness.

Limitations

Having discussed the results in section 5, several limitations of the thesis should be

mentioned. First, as previous studies have shown (see e.g. Demirgüç-Kunt et al., 2008; Klomp and De Haan, 2010) measures of regulation and supervision are very likely to be endogenous regressors. Therefore, results might not be as precise as they would be if endogeneity were accounted for.

Second, the regulation data survey captures the restrictiveness of the rules and regulations that are on the books, which may not reflect reality. The degree of actual implementation might be different across countries.

Next, the regulation survey data as well as data on bank ratings are available for only one point in time and therefore the present study cannot rely on time series variation.

Furthermore, the results are only applicable to the commercial banking sector, because data on only few savings and cooperative banks were available.

(38)

Further research

Considering the mixed results provided by various scholars regarding the link between regulation and soundness in the banking sector, future researchers should further investigate this relationship, applying new proxies for regulation (including an index that measures restrictiveness of regulation law on the books as well as the degree of implementation in real life)11 and financial stability. Although Z-scores and bank ratings have been a very popular measures, they still have some disadvantages. For instance, they focus only on individual banks and do not reflect the effects of contagion in the banking system.12 Also, in order to obtain reliable results it is crucial that future researchers use measures that can capture

changes in regulation and stability as precisely as possible (unfortunately, existing databases provide information for only one point in time).

Also, future studies might incorporate a larger set of countries and compare the effects of regulation in different regions of the European Union. For instance, it would be reasonable to split the European Union into two groups of states such as the Eurozone13 and the non-Eurozone countries, and compare the effects of regulation on bank soundness. Another interesting possibility, considering the recent debt crisis in the Eurozone, would be to separate “healthy” Northern EMU states from the “problematic” Southern region and compare the impact of regulation on bank stability in these regions. Does the degree of restrictiveness of regulation, which is chosen by the countries themselves, have a different effect on bank stability in various regions? Or, do other factors, besides regulation, influence the stability of the banking sector? It might be, for instance, that qualitative factors related to shareholder ownership and corporate governance have an important influence.14

11

Constructing such an index requires building a database that consists of a survey about regulation law as well as its real implementation.

12

Z-scores of individual banks between countries are not correlated (see also Demirgüç-Kunt and Detragiache, 2010).

13

At the moment, the euro area consists of 17 countries: Slovenia joined in 2007, Malta and Cyprus - 2008, Slovakia - 2009 and Estonia - 2011.

14

(39)

References

Barth J.R., Caprio Jr., G., Levine, R., 2001. Banking systems around the globe: Do regulations and ownership affect performance and stability? The World Bank , Policy Research Working Paper Series no. 2325.

Barth, J.R., Caprio Jr., G., Levine, R., 2003. Bank regulation and supervision: Lessons from a new database. In: Macroeconomic Stability, Financial Markets, and Economic Development. Jose Antonio Murillo Garza (Ed), Mexico, City: Banco de Mexico.

Barth, J., Caprio Jr., G., Levine, R., 2004a. Bank regulation and supervision: What works best? Journal of Financial intermediation 13, 205-248.

Barth J. R., Caprio Jr. G., and Levine R., 2004b. Guide to the 2003 World Bank Survey, available at http://www.worldbank.org/research/projects/bank regulation.htm

Barth, J., Caprio, Jr., G., Levine, R., 2008. Bank Regulations are Changing: For Better or Worse? World Bank Policy Research Working Paper 4646.

Beck, T., Demirgüç-Kunt, A., Levine, R., 2006. Bank Concentration, Competition, and Crises: First results. Journal of Banking and Finance 30(5), 1581-1603.

Berger, A. N., Demirgüç-Kunt, A., Levine, R., and Haubrich J. B., 2004. Bank Concentration and Competition: An Evolution in the Making. Journal of Money, Credit and Banking,

Blackwell Publishing 36(3), 433-451.

Berger H., and Hefeker, C., 2008. Does Financial Integration Make Banks more Vulnerable? Regulation, Foreign owned Banks, and the Lender-of-last Resort. International Economics

and Economic Policy 4(4), 371-393.

(40)

Besanko, D., and Thakor, A., 1992. Banking Deregulation: Allocational Consequence of Relaxing Entry Barriers. Journal of Banking and Finance 16, 909–932.

Blum J., 1999. Do Capital Adequacy Requirements Reduce Risks in Banking? Journal of

Banking and Finance 23, 755–771.

Boyd, J. H., and Runkle, D., 1993. Size and Performance of Banking Firms. Journal of

Monetary Economics 31, 47–67.

Boyd, J. H., Chang, C., and Smith, B. D., 1998. Moral Hazard under Commercial and Universal Banking. Journal of Money, Credit and Banking 30(32), 426–468.

Brooks, C., 2008. Introductory Econometrics for Finance. 2nd edition. Cambridge University Press, Cambridge.

Brunnermeier, M., Crockett, A., Goodhart, M., Hellwig, C., Persaud, A., and Shin, H., 2009. The Fundamental Principals of Financial Regulation: 11th Geneva Report on the World Economy, ICMB.

Buch, C. M., and Heinrich R. P., 2002. Financial Integration in Europe and Banking Sector Performance. Kiel Institute of World Economics, Working Paper.

Caprio Jr. G., and Martinez, A.E., 2000. Avoiding Disaster: Policies to Reduce the Risk of Banking Crises. World Bank Mimeo, Working paper no. 47, Egyptian Center for Economic Studies.

Cihák, M., 2007. Systemic Loss: A Measure of Financial Stability. Czech Journal of

Economics and Finance 57, 5-26.

Cihak, M., and Tieman, A., 2008. Quality of Financial Sector Regulation and Supervision Around the World. IMF Working Paper 08/190. International Monetary Fund, Washington, DC.

Impact of Bank Regulation and Supervision on Bank Soundness: An Analysis of 12 Eurozone countries’ banks