Upload
raza-rehman
View
185
Download
0
Embed Size (px)
Citation preview
MSC DISSERTATION
Did the financial crisis demonstrate the relative robustness of Islamic banks over their Conventional counterparts?
DEPARTMENT OF ECONOMICS, UNIVERSITY OF WARWICK
Raza Rehman (ID: 1463577)
Supervisor: Mr Alexander Karalis Isaac
September 2015
Word count: 5934 (excluding appendix, footnotes and references)
Abstract: The consequences of the Great Depression raised concerns regarding the reliability of the Conventional
banking system and coupled with the noteworthy growth of Islamic banking, has sparked interest in this new model.
With the literature varying on the relative rigor of the Islamic banking model, this dissertation aims to provide an
alternative perspective on the debate using a sample of 397 banks from 13 countries over a 23 year period.
Employing the Fixed Effects estimation technique, it is found that Islamic banks do not offer the extra element of
safety implicit in their theoretical model, revealing the possibility of a lack of risk management expertise and the
diversity of Islamic banking practices due to different interpretations of the Islamic guidelines
This dissertation proposal is submitted in partial fulfilment of EC 959 MSc Dissertation for the degree of MSc Economics
1
Acknowledgement:
First and foremost, I thank Allah for showering me with His unconditional blessings and giving
me the opportunity to study at a renowned university alongside some of the most talented
academics. Mr Alexander Karalis Isaac has been a great inspiration and his support gave me the
confidence to compile this paper. I would also like to thank my family and friends, who have
always been my pillar of support throughout my each and every endeavour.
2
Contents
1. Introduction ............................................................................................................................. 3
2. Banking theory ........................................................................................................................ 5
3. Islamic Banking framework .................................................................................................... 7
4. Literature Review .................................................................................................................... 9
5. Data ........................................................................................................................................ 11
6. Methodology .......................................................................................................................... 15
7. Empirical results .................................................................................................................... 19
8. Discussion .............................................................................................................................. 21
9. Conclusion ............................................................................................................................. 23
Appendix: ...................................................................................................................................... 24
References ..................................................................................................................................... 29
3
1. Introduction
The finance-growth nexus has been explored in great detail with varying conclusions to the
direction of causality. Evidence supporting financial development as a factor of economic
growth has been widely documented together with arguments for the reverse and even
bidirectional relationship. Nevertheless, it is universally accepted that stability in the financial
sector is vital to economic growth and this was made evident during the recent financial crisis.
The 2007/08 financial crisis, or Great Recession, was one of the worst to hit the United States
since the thirties. The Bureau of Labor Statistics found that unemployment more than doubled
from 4.7% in December 2007 to a peak of 10% in October 2009, with the real Gross Domestic
Product decreasing by 3.8% in early 2009 (Swann, 2009). An upward trend in house prices,
favourable government policies and optimistic expectations attracted a surge in demand for real
estate from both first-time home buyers and aspiring realtors. Also wanting to capitalise on the
situation, banks loaded their portfolios with mortgages to prime but worryingly sub-prime
borrowers. In 2002, less than 10% of U.S. mortgages were subprime but accounted for almost
25% in 2005 (Goodwin et al., 2013).
Under the Basel accords, banks were required to maintain at least an 8 percent capital buffer
against their assets after adjusting for risk however, the activity in the housing market meant
complying with such requirements was going to be costly. To avoid such losses, banks took part
in Securitization and introduced the capital market into its transactions. A period of unorthodox
lending of mortgages, complex financial instrument engineering and lenient regulation brought
about the rapid expansion of the housing bubble. Once house prices stopped rising and the sub-
prime mortgage borrowers started to default, the value of these AAA-rated financial instruments
(i.e. Collateralized Debt Obligations) completely diminished and panic spurred. The ultimate
bust of the housing bubble led to bank runs and the collapse of giants such as Lehman Brothers
and Freddie Mac, triggering the failure of a series of financial institutions that prompted a global
economic downturn.
The trustworthiness of the central culprit of the crisis - the banking system – has since been
under question. The distinctive principles that shape its model coupled with its unprecedented
growth has sparked interest in Islamic Banking and Finance. Pre-crisis, global Islamic banking
4
assets increased by almost 20% per annum till 2013 (Islamic Financial Services Board, 2015)
and according to the World Islamic Banking Competitiveness Report 2013/14 by EY, continued
such patterns following the crisis, growing 16% since 2011. With an increasing Muslim
population, relatively untapped market and repeated episodes of financial market failures1, this
new mode of Islamic Banking and Finance may be of great benefit to the global economy. The
objective of this dissertation is to shed light to the argument of whether the alternative banking
system based on Islamic teachings is the way forward towards a more stable global community.
By analysing the performance of Islamic Banks and their conventional peers across the financial
crisis, the dissertation aims to answer the following question:
“Did the financial crisis demonstrate the relative robustness of Islamic banks over their
Conventional counterparts?”
The paper is organized as follows. Section 2 outlines the theory behind financial
intermediation, with a description of the Islamic banking model and the principles underlying its
framework in Section 3. Section 4 reviews the literature that has inspired the dissertation. Section
5 discusses the data and the methodology adopted is discussed in Section 6. Section 7 follows
with the results of the analysis. A discussion of the results is given in Section 8 and finally,
Section 9 encompasses a conclusion along with potential areas of future research.
1 “…100 crises in the past 35 years” Stiglitz (2003), Dealing with Debt: How to Reform the Global Financial System, page 54.
5
2. Banking theory
In an Arrow-Debreu setting with a complete set of state-contingent markets and absence of
information and transaction costs, any equilibrium allocation is a Pareto optimum and financial
intermediation only creates a distortion to the economy. However, such conditions are unrealistic
and agents face costs of acquiring information and conducting transactions, which if
considerably large may discourage economic exchange. The ability to reduce such costs and in
turn, facilitate trade allows scope for financial intermediation.
This notion was introduced formally by Diamond (1984). His theory embodies the difficulty
faced by individual lenders to screen feasible projects together with minimising the costs
associated with monitoring the actions of borrowers once contracts are signed. Furthermore,
Diamond pointed out that individual lending can lead to either the duplication of information
when each lender monitors or no monitoring whatsoever as lenders will wish to free-ride, i.e.
each person presumes the other will monitor which results in an absence of monitoring. As
intermediaries, banks face cost advantages of framing and monitoring loan contracts which allow
for Pareto optimal allocations to be achieved, ceteris paribus.
Despite banks and other financial institutions being similar in their totality, banks hold a
certain feature that distinguishes them from other financial institutions. This unique function of a
bank is referred to as ‘maturity transformation’; that is, the conversion of long-term illiquid
investments into liquid deposits (Diamond and Dybvig, 1983). Returns are generally higher for
projects that offer payoffs later in the future however, many investors shy away from such
projects. Banks conversely, due to the unique feature of being able to offer depositors access to
their savings before the returns from the investments are realised, fill this void and finance such
projects. This allows depositors with different consumption preferences to smooth consumption
by withdrawing funds according to their expenditure plans. Despite demonstrating that this
creation of liquidity can boost output, Diamond & Dybvig (1983) find that this function has also
one weakness which is that it is vulnerable to ‘bank runs’, i.e. collective withdrawal of deposits
by customers in response to expectations of the bank failing. These bank runs force the bank to
liquidate the long-term investments at a loss that can lead to bankruptcy and have adverse
implications on economic activity.
6
Market rumours and confidence levels play focal roles in the health of the financial sector
due to their contagious property and are the main cause behind bank runs. Lack of confidence
stemming from speculations associated to a particular bank can spread to other stable banks as
depositors will want to withdraw their savings before a run commences, ultimately resulting in a
bank run. Panics of such nature test banking systems and the dissertation aims to supplement the
literature by applying this notion to the resilience of Islamic and Conventional banks during the
Great Depression. Before doing so, the following section gives a brief description of the Islamic
banking system.
7
3. Islamic Banking framework
Islam echoes modern economics in that trade should be encouraged in order to achieve
economic growth and development. Where rationality plays the central role in traditional
economic theory, the Islamic concept encompasses both moral and spiritual dimensions to the
problem on the premise that personal pursuit of welfare maximization does not necessarily
translate to an improvement in the general welfare of the society (Iqbal and Mirakhor, 2013).
Contradictory to popular belief, Adam Smith also held views mirroring those extrapolated by
Islam in his work Theory of Moral Sentiment (1979), in which Smith voices his belief of a
Supreme Power and emphasises that following the guidelines revealed by the Creator will
internalise distortions in a country and produce fair outcomes.
Despite the role of both Islamic banks and Conventional banks being parallel and the
prominent theories of banking applying to both, their underlying models are distinct. The
adherence to Shari’ah, i.e. Islamic law, distinguishes Islamic banks from their Conventional
peers. The Islamic banking model, although introduced a few decades ago, is characterised by
principles revealed during the infancy of Islam, which are defined under fiqh mualamat (rules on
transactions).
The overarching of these is the prohibition of the receipt and payment of Riba, i.e. interest.
Money serves only as a store of value and medium of exchange in Islam and treating it as a
commodity to make money is regarded as a major sin. The condemnation of interest can be
attributed to the following passage of the Quranic scripture;
“O you who have believed, do not consume usury, doubled and/or multiplied, but fear
Allah that you may be successful.”2
Trade plays an integral part of the Islamic economic system and compensation for investment
is valued however, the risk of a venture must be divided amongst the parties involved. In other
words, risk should be shared rather than transferred. Instead of interest, the returns of an
investment stem from profit and loss sharing, which encourages prudent lending, efficient
2 The Holy Quran, Chapter 3 (Al-Imran), Verses 130-131.
8
entrepreneurship and more importantly, an equitable distribution of funds. The following verses
from the Quran express this;
“…Allah has permitted trade and has forbidden interest.”3
“O you who have believed. Squander not your wealth among yourselves in vanity, except it
be a [lawful] trade by mutual consent, and kill not one another. Indeed, Allah is ever Merciful
unto you.”4
Despite the promotion of profitable trade, illicit activities and forbidden sectors such as
pork, alcohol, drugs etc. cannot be extended finance irrespective of potentially high returns.
Excessive uncertainty, or Gharar, via excessive risk-taking and speculation nullify a contract as
any monetary gains obtained are through sheer luck and not effort. Lastly, financial activity must
have an association with the real sector activity such that lending can be linked to the underlying
asset.
To incorporate the principles mentioned above, two modes of financing have been structured;
these are the Mudarabah (passive partnerships) and Musharakah (active partnerships)
arrangements. The Mudarabah contract involves a capital provider (bank) and an investment
manager (borrower) in which the former party remains isolated from the venture. The returns to
each party are a proportion of the profits, if any, according to a pre-agreed ratio but the financial
losses are borne solely by the capital provider. The losses faced by the investment manager are
the time and effort expended. The Musharakah contract differs to the Mudarabah in that the
bank may not be the exclusive investor and can, but is not required to, participate in the
management decisions. Again the profits are shared in accordance to a ratio agreed at the time of
the contract and losses equate to the respective capital investments. Where profit sharing does
not apply, other modes of financing have been designed and a description of these is given in
Table 6 under the Appendix.
3 The Holy Quran, Chapter 2 (Al-Bakarah), Verse 275. 4 The Holy Quran, Chapter 4 (An-Nisa), Verse 29.
9
4. Literature Review
Established in 1975, Dubai Islamic Bank became the first bank to offer Shari’ah compliant
products and services. Since then, the growth in the size and number of Islamic banks and
financial institutions has been unmatched, reaching major economies in the West5 and some
large Conventional banks opening Islamic windows6. Given its recent beginning, literature on
Islamic banking performance is relatively scarce. The majority of the work is theoretical and the
empirical papers focus on country-specific analysis before the U.S. housing bubble even started
to materialise. Nevertheless, contributions examining cross-country comparisons with its
Conventional counterparts across the financial crisis have been documented.
Hasan and Dridi (2010) evaluate Islamic and Conventional banks from eight countries under
four criteria; profitability, asset growth, credit growth and external ratings. Using non-parametric
analysis, they found that Islamic banks outperformed their counterparts in terms of the latter
three measures and matched them in terms of profitability, attributing the results to the adherence
of Shari’ah principles. However, Islamic banks were found to be adversely affected by the
second-round effects due to lack of credit diversification. Furthermore, they found that larger
Islamic banks were more profitable due to economies of scale and strong reputation; contrary to
the results found by Cihak and Hesse (2010).
Distinguishing by size, Cihak and Hesse (2010) use the Z-score on a sample ranging from
1993-2004 to conclude that Islamic banks fared better than non-Islamic banks when small but
become financially weaker as they grow, reflecting the difficulty of scaling credit risk
management systems. The differing results to Hasan and Dridi (2010) could be due to different
samples and different definition of large banks.
Beck et al. (2013) compare business orientation, efficiency, asset quality and stability in 22
countries and found that Islamic banks were less-cost effective but had higher intermediation
ratios and were better capitalized, which supported Islamic banks across the crisis. The results
obtained by Bourkhis and Nabi (2013) contrast the common finding of relative serenity of
Islamic bank performance. Also using the Z-score, they find that there was no significant
5 E.g. United Kingdom, United Sstates, Germany, etc. 6 Department in a Conventional bank that provides Islamic banking products and services.
10
difference in the effect of the crisis on the soundness of both types of banks, implying a
diversion of Islamic banks from their theoretical models.
In theory, the principles underlying the Islamic banking system should act as a defence
against such financial crises and even mitigate the likelihood of such occurrences. In fact, those
that found Islamic banks to have performed better than their Conventional counterparts have
suggested that the Islamic principles have been the reason for this. That is, the sharing of risk,
extension of credit based on moral criteria and the constraint on uncertainty should armour
Islamic banks from such episodes. However, the inconsistent results have motivated my decision
to explore the argument and evaluate whether Islamic banking can be beneficial to financial, and
ultimately, economic stability. By revising the approach used in existing work, the dissertation
will hope to complement the literature and provide an alternative perspective on the Islamic vs
Non-Islamic banking debate and determine whether or not the Islamic banking framework could
prevent such episodes in the future.
11
5. Data
An initial panel of 2489 banks across 32 countries was constructed however, the balance
between Islamic and Conventional banks was considerably disproportionate. To address this,
countries homing either very small and/or very few Islamic banks in relation to Conventional
banks were dropped. Specifically, countries with an Islamic bank presence in terms of total
assets of less than $5bn were excluded. Furthermore, conflict prone countries including
Palestine, Iraq, etc. where also removed from the sample in order for the financial crisis to be the
central focus. The sample was further trimmed by removing countries with a sum of total Islamic
banking assets of less than $950m along with those countries in which total assets of Islamic
banks were much larger or much smaller than the other countries. Finally, banks with data not
extending till the end of the crisis were dropped, resulting in a final dataset of 397 banks of
similar size from 13 countries across 23 years (1993-2015 inclusive). Banks with branches in
other countries were considered as separate entities. Table 1 presents the number of both types of
banks in the final sample.
Table 1: Number of banks in sample
Country Islamic Conventional Total banks
Bahrain 19 12 31
Bangladesh 7 38 45
Egypt 3 23 26
Indonesia 10 62 72
Jordan 3 11 14
Kuwait 11 5 16
Malaysia 18 33 51
Pakistan 9 22 31
Qatar 6 6 12
Saudi Arabia 5 8 13
Sudan 16 9 25
Turkey 4 30 34
United Arab Emirates 9 18 27
Total 120 277 397
12
The variables used can be divided into two categories; bank-specific and macroeconomic
indicators. The first set of variables were obtained from the Bankscope database, which compiles
information financial institutions across the globe. For the macroeconomic variables, the World
Bank database was relied upon. Table 2 provides a brief description of the variables used in the
study and their relation to the dependent variable, i.e. the log transformation of the Z-score
(discussed in Section 6).
Table 2: Description of provisional variables
Variables
(Regressors) Description
Expected
effect Explanation
Equity ratio
+
Capital stock acts as buffer for banks to repay
depositors when facing losses and allows for
diversification and higher income.
Loan Loss Reserve
-
Non-performing loans cause a reduction in equity
while reducing realizable net income.
Cost ratio
-
Higher ratio implies less efficient management of
costs and so, lowers profits and less stability.
Earnings Return on (Average) Equity + Higher profits strengthen asset side of balance
sheet and are a further cushion against bad loans.
Liquidity Ratio of Net Loans to Total Assets Ambiguous Higher ratio implies higher income but increased
exposure to default risk.
Size Logarithm of Total Assets +
Size of the bank is expected to have a non-linear
effect. Positive relation due to Economies of Scale
but only till a certain threshold. Square of logarithm of Total Assets
-
Inflation Annual percentage change in
Consumer Price Index Ambiguous
Inflation will lead to higher profits and higher
costs so effect depends on the dominating effect.
Growth Real GDP growth rate + Economic upturns are associated with higher
profits and lower non-performing loans.
Real interest rate Interest rate adjusted for inflation Ambiguous Effect depends on dominating channel, i.e.
lending rate or deposit rate.
IB Market share Market share of Islamic banks +
A higher share of Islamic banks, in terms of Total
Assets should increase the stability in the financial
sector.
13
The raison d'être for the explanatory variables holds roots to the commonly applied
CAMEL model which abbreviates to Capital, Asset quality, Management, Earnings and
Liquidity. These five elements are reviewed by credit rating agencies and regulatory bodies in
order to evaluate the performance of a bank and have formed the basis for the Financial
Soundness Indicators of the International Monetary Fund. The dimensions of the model are
applied as determinants of a banks Z-score.
Capital ratios attempt to quantify a bank’s financial position and assesses the ability of a
firm to withstand loss or liquidation. Maintaining sufficient capital stock allows banks to protect
themselves from unexpected losses and the Equity ratio is used to account for this. Asset, or
Loan, quality refers to the credit risk associated with a particular asset, i.e. any balance sheet
item that generates an income. A portfolio of outstanding but not received loan income is an
indication of poor asset quality and Loan Loss Reserve to Total Assets is used to represents this.
Due to its qualitative connotation, there has not been a consistent or straightforward method of
representing Management in empirical work. In attempt to capture the quality of management
decisions, the Cost to Income ratio is used as poor management will lead to unnecessarily high
costs. Earnings establish the capacity of a bank to fund its investment decisions and provide an
additional buffer for bad debts and are represented by Return on Equity. Liquidity ratios attempt
to measure the ability of a firm to pay off its short-term debts. The crisis demonstrated the
importance for banks of maintaining current assets to fulfil their immediate commitments and
Net Loans to Total Assets is used as its measure.
To account for the cost advantages from an increase in operational capacity, Size is also
included in the list of regressors and is represented by the logarithm (henceforth, log) of Total
Assets. The square of the log of Total Assets is used to reflect that these Economies of Scale tend
to reverse after a certain level. The state of the economy impacts the environment of the financial
market and the macroeconomic variables reflect the key economic indicators. These include the
growth rate of Real Gross Domestic Product, Inflation rate and Real interest rate. Table 3.1
below displays the descriptive statistics for the bank-specific variables employed for both the
Islamic banks and Conventional banks used in the sample, with Table 3.2 reporting the overall
summary statistics.
14
Table 3.1: Summary Statistics for Islamic and Conventional samples
Variables
Islamic banks Conventional banks
Mean Std. dev Mean Std. dev
Z-score 3.019 8.598 3.565 8.475
Log Z-score 0.486 1.104 0.734 1.087
Equity ratio 9.374 20.142 6.057 10.372
Loan Loss Reserve 1.827 7.276 2.798 7.661
Cost ratio 22.217 51.758 21.999 37.109
Earnings 3.438 11.202 5.158 23.099
Liquidity 15.248 26.105 20.367 27.188
Ln (Assets) 4.853 6.739 5.999 7.036
Market share 0.008 0.039 0.016 0.051 Note: The mean and standard deviations of the internal characteristics of each type of bank is reported in this table. On average, both types of
banks have a similar Z-score in both log and levels, with Conventional banks having a relative better score. The mean of the Size variable shows
that Conventional banks are much larger than Islamic banks, which is further implied by the average Market share. Islamic banks have higher costs, are less liquid and hold less capital stock than their Conventional peers.
Table 3.2: Summary statistics
Variable Observations Mean Std. Dev. Min Max
Log Z-score 9131 0.659 1.098 -4.653 5.699
Equity ratio 9131 7.059 14.142 -97.27 100.00
Loan Loss Reserve 9131 2.505 7.561 0.00 100.00
Cost Ratio 9131 22.065 42.075 0.00 950.00
Earnings 9131 4.638 20.269 -650.26 741.32
Liquidity 9131 18.820 26.966 0.00 109.14
Ln (Assets) 9131 5.653 7.158 0.00 19.01
(Ln (Assets))2 9131 83.193 109.318 0.00 361.32
Growth 9131 3.237 3.684 -13.127 30.012
Inflation 9131 4.8203 8.9959 -4.8633 132.824
Real Interest rate 9131 1.6301 4.8529 -24.6002 41.254
Market share 9131 0.0138 0.0479 0.00 0.9212 Note: This table reports the first and second moments, the minimum and maximum values for the bank-specific, macroeconomic and market share variables. Data sources and definitions of the variables are mentioned in the main body of the dissertation.
15
6. Methodology
The European Central Bank defines financial stability as “a condition in which the financial
system – intermediaries, markets and market infrastructures – can withstand shocks without
major disruption in financial intermediation and in the effective allocation of savings to
productive investment.” One such shock can be a bank run. Bank runs have the characteristic of
being contagious in that, speculations regarding the health of a particular bank or financial
institution can readily hinder the stability of another. Such was the case across the financial
crisis, whereby news regarding entities such as Lehman Brothers, Freddie Mac and Northern
Rock mushroomed not only across the United States but also abroad. The dissertation will use
this notion in an attempt to achieve its objective and captures this by using the Z-score.
The Z-score has become a popular measure of bank soundness. Traditionally, the statistical
Z-score measures the distance of an observation from the mean in terms of standard deviation,
allowing for a meaningful comparison of different samples of data. Its application has recently
been extended to finance literature, where the Z-score is interpreted as the number of standard
deviations below the mean the Return on Assets has to fall before equity is depleted. A financial
institution is classed as insolvent if its losses exceed the value of equity. The Z-score is
calculated as follows,
where, ROA – Return on (average) Assets
CAR – Capital Asset Ratio
sd (ROA) – Standard deviation of ROA (proxy for return volatility)
Its popularity has stemmed from its simplicity in terms of its calculation, where accounting
metrics are used rather than market data, and in terms of its relation with the probability of
insolvency; a higher Z-score implies lower probability of bank insolvency and in turn, a more
stable bank, vice-versa. Being composed of accounting data expands its application to various
groups of financial institutions including unlisted financial institutions but more importantly
Islamic banks, allowing for an objective analysis of the question at hand.
16
However, Lepetit and Strobel (2015) amongst others extrapolate that the simple Z-score,
amidst its advantages, has an important flaw especially in regards to drawing inferences; it
suffers from a skewed distribution. The relationship between insolvency risk and the Z-score is
only valid based on the assumption that profits, or Return on Assets, are normally distributed.
This may not be the case and as a result, the Z-score has a skewed distribution. To address this,
the dissertation will modify the econometric model used in the literature by applying a
logarithmic transformation to the Z-score based on the practical work of Laeven and Levine
(2009) and findings of Lepetit and Strobel (2015). As robustness checks, the identical model will
be regressed but using the simple Z-score.
The framework advanced by Cihak and Hesse (2010) and Bourkhis and Nabi (2013) is
followed in which bank stability is modelled controlling for bank-specific and macroeconomic
characteristics. But instead of the simple Z-score, the logarithmic transformation is employed in
order to overcome any distributional issues attached to the measure. The variables of interest are
the interaction terms between the Islamic bank dummy and the Crisis period dummy (here, 2007-
2009) along with the market share of Islamic banks, which are both expected to have positive
coefficients.
Given the panel structure of the data, Pooled Ordinary Least Squares (POLS), Fixed Effect
and Random Effect estimation techniques will be considered. Past values of financial metrics
have been found to impact the current position of a bank (Athanasoglou et al., 2008) and to
account for this variation in the dependent variable, the bank-specific and macroeconomic
variables are lagged one period giving the following log-level model,
where, – Z-score for bank in country at time
- vector of bank-specific variables
17
- vector of macroeconomic variables
- Islamic bank dummy (1 if Islamic, 0 otherwise)
- Crisis dummy (1 if 2007-2009, 0 otherwise)
– Market share of bank at time
- vector of year dummies
- vector of country dummies
The POLS estimation technique minimizes the squared residuals of the model in order to
provide the line of best fit and is adopted as a baseline regression in this analysis. However, there
are two key assumption underlying the technique which must be considered when interpreting
the results obtained. These are that the independent variables are uncorrelated with the
disturbance term, i.e. Е ( ' ) = 0, and the observations are homogenous, i.e. = . Given the
fact that different banks specialise in different business areas, concentrating on those that
generate higher profits given the needs of the customers and cost of raising finance, the
assumption of homogeneity is not a sensible one. An attempt to capture these differing
characteristics is made however, not all the individual effects are measurable and this
heterogeneity provides the foundation for exploiting other techniques.
Least Square Dummy Variable (LSDV), Fixed Effects and Random Effects are the most
common approaches when dealing with panel data. The LSDV approach is most effective when
the number of observations are not considerably larger than the period under consideration but
with the dataset comprising of 397 banks across 23 years, the technique is disregarded. Both the
Fixed Effect and Random Effect techniques acknowledge the heterogeneity amongst
observations but differ in its treatment; the former introduces the diversity through dummy
variables for each observation whereas the latter includes it in the innovation term. That is, Fixed
Effects assumes the heterogeneity is correlated with other regressors whereas Random Effects
assumes heterogeneity is uncorrelated with other independent variables. Given both methods are
applicable to most work using panel data, the Hausman test is conducted in order to determine
the most appropriate. The results are given in Table 4.1 and infer that the individual effects do
not meet the exogeneity requirement needed for the Random Effects, so the Fixed Effects
estimation should be adopted. The Fixed Effects technique removes the heterogeneity by
18
demeaning the data, i.e. deducts the mean from each observation. Given the time-invariant nature
of the country dummies, they are automatically dropped in the regressions.
Table 4.1: Hausman test results
Hypothesis Prob>chi2 FE or RE?
: [ ′ ]=0 or no correlation 0.0000 FE
To be able to draw correct inferences from the results of the regressions, it is fundamental
that the variance of the errors is time-invariant. In order to test whether the errors are
heteroskedastic, the Bruesh-Pagan test is run. The results, given in Table 4.2, show the variance
suffers from heteroskedasticity. The presence of this invalidates the inferences drawn from the
hypothesis tests however, by using the command robust in STATA, this can be corrected for.
Table 4.2: Breusch-Pagan test for heteroskedasticity
Hypothesis Prob>chi2 Heteroscedasticity?
: 0.0000 Yes
19
7. Empirical results
Four specifications of the model are considered; the first containing only bank-specific
characteristics, the second including macroeconomic variables and the third introducing the
variables of interest, i.e. the Islamic bank-Crisis interaction term and Islamic bank market share.
In order to dwell deeper into the issue at hand, the interaction term is divided amongst the three
central years of the financial crisis, giving the fourth specification of the model. The results from
the Fixed Effects estimation are displayed in Table 5.2 and those drawn from the final two
specifications are analysed in the following.
From the penultimate model, the coefficient on Cost to Income ratio is expected to have a
negative sign as higher costs relative to the bank earnings puts the bank, Islamic or
Conventional, in a financial predicament and in turn, reduces its robustness. The results for the
variable also suggest such a relation however, are statistically insignificant at the 10%
significance level. Increments to the Loan Loss Reserves, as expected, reduce the log Z-score but
the opposite is indicated for Net Loans to Total Assets. That is, a more liquid position is
beneficial to the stability of a bank as the revenues generated exceed the default risk from the
increase in lending. The coefficients on both variables are insignificant. A higher capital stock
improves the log Z-score of a bank and similarly, so does Earnings, both being statistically
significant. The results on the Size variables demonstrate the existence of Economies of Scale,
i.e. larger banks have higher profits and are more stable but only till a certain threshold. The sign
of the coefficients on log of Total Assets and the Square of the log of Total Assets are positive
and negative respectively, but only the coefficient on the former is statistically significant.
A more prosperous economy is commonly viewed as an important factor for the prosperity of
the financial market and using Real GDP Growth and Inflation as measures of the outlook of a
country, we find this notion holds. Real GDP Growth and the log Z-score have a positive and
significant relation according to the data and conversely, Inflation has an inverse but statistically
insignificant relation. The Real Interest rate impacts the log Z-score via two channels; the
lending rate and the deposit rate. Some countries in the sample7 follow a pegged system and so
interest rates in those countries will follow the rates set by the Federal Reserve. Nevertheless, the
7 Saudi Arabia, UAE, Bahrain and Qatar
20
results suggest the second channel is dominant such that increases in the real interest rate lowers
the stability of a bank, with the coefficient being statistically significant at the 10% level.
Despite the theoretical vigour of the Islamic banking model, the results obtained from the
third specification of the model fail to support this structural advantage across the financial
crisis. The interaction between the Islamic bank dummy and Crisis dummy has a negative
coefficient, implying that Islamic banks were also prone to the effects of the crisis but the result
is statistically insignificant.
Separating the interaction term to account for each of the three years of the crisis in the final
specification, the results suggest the opposite for the variables of interest. Islamic banks are
found to have been adversely impacted in the run up to the crisis but withstood the shocks when
the crisis was in full effect, with the results being statistically significant for 2007 but
insignificant for 2008 and 2009. Considering the implications of the market shares of Islamic
banks on the stability of individual banks, the results suggest that a financial sector with a higher
proportion of Islamic banks is less likely to create vulnerability in the performance of individual
banks. The results however, are insignificant even at the 10% level. In respect to the bank-
specific and macroeconomic variables, the same results and inferences are obtained from the
final specification.
In order to test the robustness of the results, a contemporaneous version of all specifications
of the model are analysed and the results are tabulated in Table 5.3. Other than the coefficient on
Cost-to-Income variable becoming positive but still insignificant, the remaining results hold.
Furthermore, the model is estimated using the simple Z-score, the results of which are reported
in Table 5.4. The R-squared of the model is considerably less than the model considered and
contrary to the results of the final specification, using the simple Z-score generates a positive and
insignificant coefficient on the Islamic bank-Crisis interaction dummy. When distinguishing
between each of the years of the crisis, the results mirror those produced using the log
transformation of the Z-score.
21
8. Discussion
According to the final specification of the model, the results obtained for the bank-specific
and macroeconomic variables are in accordance to the expectations in place and it may be argued
that Islamic banks endured the shockwaves originated in the Western economies.
The hypothesis of whether being an Islamic bank during the financial crisis enhanced the
stability of a bank or not is tested along with whether a larger share of Islamic banks in the
financial market improves stability. The statistically significant negative coefficient on the 2007
Islamic bank interaction term indicates an initial impact of the crisis on the sample group.
However, the insignificant results on the 2008 and 2009 Islamic bank interaction terms compels
one to draw the inference that the stability of Islamic banks was not affected by the financial
crisis.
Being the first such episode encountered by Islamic banks, the lack of experience coupled
with the complexity of the new products could be attributed to the negative coefficient on the
2007 Islamic bank interaction term. The uniqueness of the Islamic banking model opens it up to
risks other than those faced by its Conventional peers, which in a mixed industry, have been a
challenge to identify let alone manage. This inexperience has hindered Islamic banks in
implementing suitable risk management strategies and many have applied either conventional
risk management techniques or variations of these strategies in attempt to manage these unique
risks (Salem, 2013), providing the basis for such results.
A larger presence of Islamic banks in the financial market is expected to create a more
tranquil financial environment but the statistically insignificant results indicate that a financial
market concentrated with Islamic banks does not improve, nor worsen, stability. This could be
due to an absence of relevant and effective regulation or a divergence of Islamic bank practises
from its model.
Chapra and Khan (2000) note that for a continuation of the expansion of the Islamic banking
sector, reforms to the regulatory and supervisory framework are key along with entities
supporting the functions of the bank; in particular, private credit rating agencies. These agencies
will facilitate the management of risks in the more risky modes of financing, i.e. Mudarabah and
Musharaka, improving profitability and curtailing moral hazard. Since their report, multiple
22
rating agencies in Muslim countries have commenced operations8 which could be the root to the
rapid recovery noted in the World Islamic Banking Competitiveness Report 2013/14 by EY.
The debate concerning the harmony between the theoretical model and the practises of
Islamic banks has been an on-going one. Many have critiqued the Islamic banking model and
argued that apart from a change in terminology, they are not any different to their Conventional
counterparts (e.g. Kuran, 1993, 2004) whereas supporters of the model contend that such a
scenario was only possible in the transition away from the conventional system (Ahmed, 1993).
With the verses regarding trade revealed over 4000 years ago, applying them to the current
system has not been a simple task and this has been the main reason for inconsistencies in the
practises among Islamic banks. The rulings on Riba (interest) and gharar (uncertainty) have been
subject to varying interpretations, with some viewing the relevant verses as a prohibition of
interest in its entirety whereas for others they signify a restriction from “excessive” interest. Due
to this, the results may not have reflected the expectations of the variables of interest.
It has been argued that a majority of the Islamic banking contracts were collateralised by real
estate in accordance to the materiality principle and this exposed Islamic banks to the “second-
round effects” of the crisis (Hasan and Dridi, 2010). This concentration of credit exposed Islamic
banks to the economic downturn in the real economy. Issues with data availability, particularly
with Islamic banking income, restricted the calculation of an Income Diversification variable
which would have tested whether a more diversified portfolio improves the stability of a bank or
not.
Even with the efforts to control for any shortcomings, the results must be reviewed with
caution as the study has its limitations. The major weakness, as is the case with the majority of
empirical work, is in regards to the quality and quantity of the data. Despite being a reputable
and frequently sourced database, the data compiled by Bankscope is drawn from the financial
statements published by banks (and other financial institutions) which may suffer from reporting
bias. With differences in auditing and reporting regulations across countries, the accounts may
not reflect the actual stance of the bank. In terms of the quantity of data, Bankscope offers the
general user data for only the most recent 16 years available which varies for every bank. Data
for some banks range from 2015 back but for others start from 2007, resulting in discrepancies in
8 Mainly International Islamic Rating Agency (IIRA).
23
the results. This mis-match in available data was addressed by expanding the timeframe from 16
to 23 years however, this resulted in additional missing values, augmenting the potential omitted
variable bias.
9. Conclusion
This dissertation analyses the premise that Islamic banks play a key role in the stability of the
financial sector. In particular, the extra element of safety drawn from the principles underlying
the Islamic banking model are questioned using the Great Depression as the period of distress.
Additionally, whether or not a larger presence of Islamic banks in the financial market translates
to a more stable environment is also examined.
Employing the Fixed Effects estimation technique, the results suggest that the benefits of the
Islamic banking model are not present in Islamic financial institutions. According to the results,
there was no definite improvement in the log Z-score of Islamic banks across the crisis nor was
there a positive influence from a more concentrated Islamic financial market. Contrary to the
general view, the dissertation concludes that the Islamic banking model is not any more robust
than the Conventional model and attributes this to the deviation of Islamic financial institutions
from their theoretical model.
Given the various interpretations of the word Riba, further study in the area could focus on
classifying groups of Islamic banks that specify their practises according to a particular
definition in order to determine the robustness of the Islamic banking model. Alternatively,
attempts at conducting a meta-analysis between a country with a fully Islamic financial system
and one without an Islamic Banking and Finance presence may also provide a useful perspective
to the argument. All things considered, the transition away from a financial system with a deep-
rooted history may prove to be too large of a step for the international community nevertheless,
the growth of Islamic Banking and Finance cannot be overlooked and embracing some of its key
features may prove to be pivotal for the state of the global financial system.
24
Appendix:
Note: This table presents the coefficient and t-statistics for the fourth specification of the model using POLS, Fixed Effect and Random Effects techniques. A significance level of <1% is
denoted by 3 stars, 1-5% by 2 stars and 5-10% by 1 star. POLS assumes homogeneity and so the results must not be taken at face value. The fixed effects and random effects addresses
the heterogeneity amongst the banks in their respective treatment methods.
Table 5.1: Regressions Results
(Dependent variable: Log Z-score) Pooled OLS Fixed effect Random effect
Variable Coefficient t Coefficient t Coefficient Z
Equity (-1) 0.0076*** 6.09 0.0081*** 4.39 0.0079*** 4.00
LLR (-1) -0.0017 -0.74 -0.0026 -0.89 -0.0023 -0.78
Cost-Income Ratio (-1) -0.0009* -1.91 -0.0004 -1.02 -0.001 -1.20
Earnings (-1) 0.0009 1.46 0.001* 1.72 0.001* 1.71
Liquidity (-1) 0.0014 1.11 0.0004 0.20 0.0007 0.33
Ln (Assets) (-1) 0.0997*** 6.47 0.1053*** 4.27 0.1022*** 4.22
(Ln (Assets))2 (-1) -0.0008 -0.99 -0.0016 -1.07 -0.0013 -0.88
Growth (-1) 0.0172*** 4.43 0.0178*** 4.27 0.0176*** 4.81
Inflation (-1) -0.0049 -0.83 -0.0004 -0.32 -0.0005 -0.41
Real Interest rate (-1) -0.0049** -2.37 -0.0041* -1.89 -0.0043** -2.00
Islamic bank Crisis1 -0.3157** -2.69 -0.2137* -1.94 -0.2466** -2.22
Islamic bank Crisis2 -0.0017 -0.01 0.0916 0.82 0.0617 0.55
Islamic bank Crisis3 -0.1017 -0.88 -0.0094 -0.09 -0.0389 -0.37
Islamic bank share -1.4361** -2.22 0.322 0.87 -0.2626 -0.60
Constant 0.1179 1.80 0.1625 2.62 0.2027 2.33
Observations 8733 8733 8733
R-squared 0.5477 0.5172 0.6973
F 343.74 103.48 6403.12
Prob > F 0.0000 0.0000 0.0000
25
Note: This table presents the results from the Fixed Effects estimation on the four specifications of the model. Given the log-level structure of the model, the coefficients are interpreted as
the percentage change in the dependent variable from a unit change in the regressor, after multiplying the coefficient by 100. Column 1 reports the results from the specification comprising
bank-specific variables only, Column 2 includes macroeconomic variables, Column 3 introduces the Islamic bank-Crisis interaction term along with the Islamic bank market share and
finally, Column 4 divides the Islamic bank-Crisis term across the three years of the crisis. The significance levels are * 0.05 < p-value ≤ 0.10; **: for 0.01 < p-value ≤ 0.05; ***: for p-
value ≤ 0.01.
Table 5.2: Fixed-effect Results
(Dependent variable: Log Z-score) Fixed effect [1] Fixed effect [2] Fixed effect [3] Fixed effect [4]
Variable Coefficient t Coefficient t Coefficient t Coefficient t
Equity (-1) 0.0084*** 4.54 0.0081*** 4.36 0.0081*** 4.41 0.0081*** 4.39
LLR (-1) -0.0025 -0.86 -0.0025 -0.87 -0.0026 -0.88 -0.0026 -0.89
Cost-Income Ratio (-1) -0.0005 -1.02 -0.0005 -1.00 -0.0005 -1.01 -0.0005 -1.02
Earnings (-1) 0.001* 1.84 0.0009* 1.71 0.001* 1.72 0.001* 1.72
Liquidity (-1) 0.0007 0.32 0.0004 0.19 0.0004 0.19 0.0004 0.20
Ln (Assets) (-1) 0.1015*** 4.10 0.1047*** 4.25 0.1055*** 4.28 0.1053*** 4.27
(Ln (Assets))2 (-1) -0.0014 -0.92 -0.0016 -1.04 -0.0016 -1.08 -0.0016 -1.07
Growth (-1) - - 0.0176*** 4.22 0.0177*** 4.25 0.0178*** 4.27
Inflation (-1) - - -0.0003 -0.29 -0.0003 -0.29 -0.0004 -0.32
Real Interest rate (-1) - - -0.0003 -0.29 -0.0041* -1.92 -0.0041* -1.89
Islamic bank Crisis - - - - -0.0439 -0.51 - -
Islamic bank share - - - - 0.334 0.91 0.322 0.87
Islamic bank Crisis1 - - - - - - -0.2137* -1.94
Islamic bank Crisis2 - - - - - - 0.0916 0.82
Islamic bank Crisis3 - - - - - - -0.0094 -0.09
Constant 0.4896 8.13 0.1612 2.60 0.1628 2.62 0.1625 2.62
Observations 8733 8733 8733 8733
R-squared 0.5176 0.5189 0.5167 0.5172
F 113.29 112.82 109.56 103.48
Prob > F 0.0000 0.0000 0.0000 0.0000
26
Note: This table presents the results of the same regression but using the current time period of the independent variables. The sign of the coefficients on the independent variables mirrors those
in the lagged model, apart from that on the Cost to Income ratio. Again, * for 0.05 < p-value ≤ 0.10; ** for 0.01 < p-value ≤ 0.05; *** for p-value ≤ 0.01.
Table 5.3: Robustness Checks – Contemporaneous model
(Dependent variable: Log Z-score) Fixed effect [1] Fixed effect [2] Fixed effect [3] Fixed effect [4]
Variable Coefficient t Coefficient t Coefficient t Coefficient t
Equity 0.0039** 2.90 0.0037** 2.73 0.0037** 2.81 0.0037** 2.82
LLR -0.0104** -2.80 -0.011** -2.85 -0.0107** -2.90 -0.0107** -2.89
Cost-Income Ratio 0.00002 0.04 0.00003 0.05 0.00002 0.03 0.00002 0.03
Earnings 0.00001 0.04 -0.0001 -0.18 -0.0001 -0.32 -0.0001 -0.34
Liquidity -0.0018 -0.73 -0.002 -0.82 -0.0021 -0.85 -0.0021 -0.84
Ln (Assets) 0.155*** 5.36 0.157*** 5.47 0.157*** 5.46 0.157*** 5.46
(Ln (Assets))2 -0.0027 -1.64 -0.0028* -1.73 -0.0037* -1.67 -0.0027* -1.67
Growth - - 0.012** 2.84 0.012** 2.85 0.0121** 2.90
Inflation - - -0.0017 -1.41 -0.0017 -1.39 -0.0017 -1.38
Real Interest rate - - -0.0022 -0.94 -0.0021 -0.93 -0.0019 -0.88
Islamic bank Crisis - - - - -0.0571 -0.67 - -
Islamic bank share - - - - -1.1073 -1.47 -1.115 -1.47
Islamic bank Crisis1 - - - - - - -0.1949* -1.82
Islamic bank Crisis2 - - - - - - 0.1041 1.05
Islamic bank Crisis3 - - - - - - -0.0815 -0.75
Constant -0.0003 -0.02 -0.0003 -0.02 -0.0003 -0.02 -0.0003 -0.02
Observations 9130 9130 9130 9130
R-squared 0.5385 0.5419 0.5537 0.5539
F 148.03 138.09 131.64 125.52
Prob > F 0.0000 0.0000 0.0000 0.0000
Table 5.4: Robustness Checks
(Dependent variable: Simple Z-score)
27
Note: This table presents the results of the same regression but using the simple Z-score as the dependent variable. A considerable difference in the R-squared can be observed. Again, * for 0.05 < p-value ≤
0.10; ** for 0.01 < p-value ≤ 0.05; *** for p-value ≤ 0.01.
Fixed effect [1] Fixed effect [2] Fixed effect [3] Fixed effect [4]
Variable Coefficient t Coefficient t Coefficient t Coefficient t
Equity (-1) 0.113*** 4.98 0.111*** 4.83 0.109*** 4.82 0.109*** 4.82
LLR (-1) 0.0065 0.16 0.0057 0.14 0.0072 0.18 0.0071 0.17
Cost-Income Ratio (-1) -0.0057 -1.17 -0.0057 -1.16 -0.0054 -1.11 -0.0054 -1.11
Earnings (-1) 0.0018 0.60 0.001 0.34 0.0012 0.39 0.0012 0.39
Liquidity (-1) -0.0209 -0.85 -0.0231 -0.93 -0.0228 -0.91 -0.0226 -0.90
Ln (Assets) (-1) 1.358*** 5.03 1.385*** 5.13 1.378*** 5.11 1.376*** 5.11
(Ln (Assets))2 (-1) -0.064*** -3.71 -0.066*** -3.80 -0.066*** -3.80 -0.066*** -3.80
Growth (-1) - - 0.125*** 3.93 0.121*** 3.76 0.121*** 3.79
Inflation (-1) - - -0.0099 -1.29 -0.0098 -1.26 -0.0099 -1.29
Real Interest rate (-1) - - -0.0285 -1.63 -0.0278 -1.58 -0.0274 -1.55
Islamic bank Crisis - - - - 1.2089 1.46 - -
Islamic bank share - - - - 4.426** 2.06 4.334** 2.06
Islamic bank Crisis1 - - - - - - 0.0011 0.00
Islamic bank Crisis2 - - - - - - 1.9746 1.64
Islamic bank Crisis3 - - - - - - 1.6524 1.50
Constant 2.6173 4.98 1.7808 1.80 1.7781 1.79 1.7763 1.79
Observations 8733 8733 8733 8733
R-squared 0.2764 0.2790 0.2694 0.2698
F 50.99 49.76 46.85 44.71
Prob > F 0.0000 0.0000 0.0000 0.0000
28
Table 6: Islamic banking products and services
Term Meaning Explanation
Amana Safekeeping Money deposited for safekeeping which earn no return
Bay’ al-salam Advance cash purchase Buyer makes a payment in advance but the delivery of
the good is made at a later date
Bay’ bithaman ajil Deferred payment sale Sale contract in which buyer makes a payment, either
lump-sum or instalments, after the sale together with a
mark-up
Ijara Leasing One party purchases an asset and another is given the
right to use it but the lessor retains ownership
Istina Deferred payment and delivery Sale contract in which a manufacturer/contractor
agrees to produce/build a certain agreed
product/building at a given price in the future, where
the price can be paid at a later date and in instalments,
depending on preferences of the parties
Ju’ala Service charge Contract to perform a specific task in a given time for
a fee
Muradabah Mark-up financing Agreement in which one party purchases a good
desired by another and sells it to them at a price which
includes a profit agreed to by both parties
Wakalah Agency Contract Contract in which one party (Muwakil) appoints
another (Wakeel) to conduct a certain task on behalf of
the principal, for a fixed commission
Qard Hassana Beneficence loans Loans without interest and profit-sharing
29
References
Acharya, V. V., Richardson, M. (2009). Causes of the financial crisis. Critical Review, 21: 2-3,
Pages 195-210, DOI: 10.1080/08913810902952903.
Ahmad, A. (1993). Contemporary practices of Islamic financing techniques. Islamic
Development Bank, Islamic Research and Training Institute, Research Paper 20. Athanasoglou, P. P., Brissimis, S. N., Delis, M. D. (2008). Bank-specific, industry-specific and
macroeconomic determinants of bank profitability. Journal of International Financial Markets,
Institutions and Money, Volume 18, Issue 2, Pages 121-136. Beck, T., Demirgüç-Kunt, A., Merrouche, O. (2013). Islamic vs Conventional banking: Business
model, efficiency and stability. Journal of Banking and Finance, Volume 37, Issue 2, Pages 433-
447. Bourkhis, K., Nabi, M.S. (2013). Islamic and Conventional banks soundness during the 2007-
2008 financial crisis. Review of Financial Economics, Vol. 22, Issue 2, Pages 68-77. Bureau of Labor Statistics (2012). The Recession of 2007–2009. Washington: U.S. Bureau of
Labor Statistics.
Chapra, M. U., Khan, T. (2000). Regulation and supervision of Islamic banks. Jeddah: Islamic
Development Bank, Islamic Research and Training Institute. Cihak, M., Hesse, H. (2010). Islamic banking and financial stability: An empirical analysis. IMF
Working Paper WP/08/16.
Demirgüç-Kunt, A., Detragiache, E. (1998). The determinants of banking crisis in developing
and developed countries. IMF Working Paper WP/97/106.
Diamond, D.W. (1984). Financial intermediation and delegated monitoring. Review of
Economic Studies, Volume 51, Issue 3, Pages 393-414.
Diamond, D.W., Dybvig, P.H. (1983). Bank runs, deposit insurance and liquidity. The Journal of
Political Economy, Volume 91, Issue 3, Pages 401-419.
European Central Bank (2015). Financial Stability Report. Frankfurt am Main, Germany:
European Central Bank.
EY (2015). World Islamic Banking Competitiveness Report. Middle East and Northern Africa,
EY.
Goodwin, N., Nelson, J., Harris, J., Torras, M. & Roach, B., (2013). Macroeconomics in context
Second Edition, Publisher ME Sharpe.
30
Hasan, M., Dridi, J. (2010). The effect of the global crisis on Islamic and Conventional banks: A
comparative study. IMF Working Paper WP/10/201.
IMF (2006). Financial soundness indicators: Compilation guide. International Monetary Fund.
Iqbal, Z., Mirakhor, A. (2013). Economic Development and Islamic Finance. Directions in
Development. Washington, DC: World Bank. Islamic Financial Services Board, (2015). Islamic Financial Services Industry Stability Report.
Kuala Lumpur, Malaysia: Islamic Financial Services Board.
Khan, F. (2010). How ‘Islamic’ is Islamic Banking? Journal of Economic Behavior &
Organization, Volume 76, Issue 3, Pages 805-820. Kuran, T., (1993). The economic impact of Islamic fundamentalism. Kuran, T. (2004). Islam and Mammon: The economic predicaments of Islamism. Princeton
University Press. Laeven, L., Levine, R. (2009). Bank governance, regulation and risk taking. Journal of Financial
Economics 93, Pages 259-275. Nomani, F. (2006). The dilemma of riba-free banking in Islamic public policy. Islam and the
Everyday World: Public Policy Dilemmas. Routledge, London, Pages 193-223. Reinhart, C. M., Rogoff, K. (2009). This Time is Different: A Panoramic View of Eight
Centuries of Financial Crises. Princeton, Princeton University Press. Roy, A. D. (1952). Safety first and the holding of assets. Econometrica 20, 431.449. Salem, R. A. (2013). Risk management for Islamic banks. Edinburgh, Edinburgh University
Press, Pages 7-12. Smith, A. (2010). The theory of moral sentiments. Penguin.
Stiglitz, J. E. (2003). Dealing with Debt: How to Reform the Global Financial System
(Development and Modernization). Harvard International Review, Volume 25, Issue 1, Pages
54-59. Swann, C. (2009). GDP and the Economy. Washington: Bureau of Economic Analysis.
Wheelock, D. C., Wilson, P. W. (2000). Why do banks disappear? The determinants of US bank
failures and acquisitions. Review of Economics and Statistics, Volume 82, Issue 1, Pages 127-
138.