J Ribaf 2016 03 014

Accepted Manuscript
Title: Credit risk, managerial behaviour and macroeconomic

equilibrium within dual banking systems: Interest-free vs.
Interest-based banking industries
Author: Awatef Louhichi Younes Boujelbene
PII: S0275-5319(16)30059-9
DOI: http://dx.doi.org/doi:10.1016/j.ribaf.2016.03.014
Reference: RIBAF 502
To appear in: Research in International Business and Finance
Received date: 2-4-2015

Revised date: 16-12-2015
Accepted date: 29-3-2016
Please cite this article as: Louhichi, Awatef, Boujelbene, Younes, Credit risk, managerial
behaviour and macroeconomic equilibrium within dual banking systems: Interest-free
vs.Interest-based banking industries.Research in International Business and Finance
http://dx.doi.org/10.1016/j.ribaf.2016.03.014
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
Credit risk, managerial behaviour and macroeconomic
equilibrium within dual banking systems: Interest-free
vs. Interest-based banking industries
List of Authors:
First author: Awatef Louhichi
Affiliation: Faculty of Economics and Management of Sfax, Tunisia
Second Author: Younes Boujelbene
Affiliation: Faculty of Economics and Management of Sfax, Tunisia
1
Credit risk, managerial behaviour and macroeconomic
equilibrium within dual banking systems: Interest-free
vs. Interest-based banking industries
Abstract
In this paper, an attempt has been made to explore the determinants of credit risk in the
banking system with a particular interest toward the Islamic banking industry. We analyze
the link between credit risk and a set of bank-specific and macroeconomic along with
institutional variables using two complementary approaches. First, we investigate the
factors of credit risk using one-step generalized method of moments (GMM) system
estimator. Then, we explore the feedback between credit risk and its determinants in a
panel vector autoregressive (PVAR) model. We have used a sample of Middle Eastern,
North African (MENA) and Asian countries to apply our model. The major purpose of this
paper is to find factors that could explain credit risk within the interest-free banking system
relative to the interest-based one.
Keywords: NPLs, Macro-financial linkages, GMM estimator, Panel VAR analysis, Islamic
Banking system
2
1. Introduction
It is commonly known that bank failure has great adverse effects as it threaten the
whole systemic stability. Reinhart and Rogoff (2010) and Castro (2013) gave evidence that
credit risk, which takes the form of non-performing loans, is one of the main factors
contributing to banking crises. Accordingly, effective risk management is critical deal with
and somewhat avoid this failure. However, an adequate risk management framework
requires principally the knowledge of the main causes or the factors that lead to this risk.
The objective of this empirical work consists in assessing credit risk factors within the
Islamic banking industry compared to the conventional one. The undertaken analysis covers
bank-level, macro-environment along with institutional-environment factors to ascertain
credit risk correlates. The analysis was conducted on a sample of 117 banks operating in the
MENA and South-East Asian countries and observed over 8 years (i.e. from 2005 to 2012).
We have chosen regions where Islamic banks operate alongside and compete with their
conventional counterparts.
Our choice of the Islamic banking industry was not arbitrary. In fact, Islamic banks
commonly synonymous with interest-free banks operate in the interest-free system and this
is one of the important things which differentiate them from the conventional or the
interest-based ones. In particular, and in the recent years, the Islamic banking industry has
been liable to protect itself against abuses reported before and during the crisis thanks to its
undertaken moral values and sets of ethics. In this context, Causse (2012) argued that
“Islamic banks were capable of escaping crises thanks to their very principles”. She further
affirmed that Islamic banks tend to take place in the global system.
To achieve our purpose, we applied two complementary approaches namely one-step

generalized method of moment (GMM) system analysis and panel vector autoregressive
(PVAR) framework. In fact, the first method would allow us to explore the various credit
risk determinants and whether the Interest-free banks behave differently from the interest-
based ones. As for the second approach, it would help test our sample banks‟ fragility to
diverse unexpected exogenous shocks. In particular, and through a PVAR study we will
attempt to examine the causal link between credit risk and the bank‟s managerial behavior
3
(i.e. bank‟s efficiency) in a first stage. In a second stage, we will try to check the macro-
financial links binding credit risk to the different exogenous variables.
Actually, this study is intended to help clarify certain misunderstanding relevant to

some analytical aspects, concerning identification of factors likely to affect credit risk and
improve risk management in Islamic banking. In fact, the present paper‟s modest
contribution lies in providing bankers with some tools whereby credit risk can be more
effectively managed through staff monitoring of the credit risk involved factors. Indeed,
bank managers may take advantage of recognizing the possible defects and trying to re-
conduct the credit risk management strategies in such a way as to face the imbalances,
safeguard or maintain funds or re-invest them. As a matter of fact, the major goal lying
behind such a study consists in further promoting the credit risk management related
policies with regard to Islamic banks as compared to their conventional counterparts.
The remainder of this article is structured as follows. Section (2) involves a comparison
established between Islamic banking and conventional banking industries in terms of
structure and risk profile, notably, in respect of credit risk strategies. As for section (3), it
depicts our undertaken theoretical framework, while section (4) encompasses a description
of the applied data and variables. Section (5) is a description of our empirical strategies.
The empirical results are discussed in section (6) and the final section provides the major
concluding remarks and research potential implications.
2. A comparative overview of credit risk in Islamic and conventional banking

systems
It is worth noting that Islamic banking institutions have experienced an unprecedented

growth over the past few years growing from a niche to a large industry in several
countries. Noteworthy, also, is that in most parts of the world1, Islamic banks operate side
by side with conventional banks. Actually, the remarkable growth of Islamic banking
industry, coupled with a relatively better performance throughout the recent financial crisis
(Hasan and Dridi, 2010), have raised important questions in this respect, particularly “In
what ways does the Islamic banking structure differ from the conventional banking one?”
In effect, Islamic banking has mainly emerged due to negative screening based on Shariah
(Islamic jurisprudence) and religious principles, which exclude any form of interest-based
1
With the exception of IRAN, Sudan and Pakistan.
4
transactions, gambling, short selling, sale of debt and excessive uncertainty in contracts. It
is, therefore, evident that Islamic banking involves a special structure that differs noticeably
the conventional banking system. Basically, the interest-free nature of Islamic banking
requires distinguished products that can be simultaneously considered as part of the asset as
well as the liability aspects of the Islamic banks. Above all, Islamic banking proposes two
major types of contracts: non-participatory or asset-based contracts (Murabahah, Ijarah,
Istisna‟a and Salam) along with the risk-sharing or equity-based contracts (Musharakah,
Mudharabah,) (See Appendix. 2 for more details).
Noteworthy, however, is that a long debate has been taking place over the issue of whether
the Islamic bank‟s operations differ in practice from what is proposed in theory. In this
respect, some elaborated academic studies argue that Islamic banks are operating just like
conventional banks (El-Gamal, 2006). Supporters of this argument highlight that non-
participatory debt-based modes, as used by Islamic banks, appear to be far higher than the
equity-based ones. Consequently, they conclude that the practice of Islamic banking is
relatively quite indistinguishable from conventional banking (Khan, 2010; Chong and Liu,
2009). That is, Islamic banks use some asset-backed debt instruments, such as Murabahah
(sale of merchandise on credit) and Ijarah (operating lease), instead of such joint-venture
financing modes as Musharakah (profit and loss-sharing) and Mudarabah (profit-sharing
contract). Overall, these studies have assumed that Islamic banking turns out to be different
from conventional banking in from rather than substance.
As regard the risk profile, Islamic banks appear to be exposed to traditional banking risks2,
similar to their conventional counterparts. Furthermore, they are exposed to even extra risks
due to their various shariah compliant instruments adopted (Khan and Ahmed, 2001;
Sandarajan and Errico, 2002, Boumedienne, 2011).
Principally, credit risk is defined as the borrower or counterpart‟s inability to fulfill their
obligations in compliance with the agreed terms (Khan and Ahmed, 2001). That is mainly
the case with conventional banks. Still, within the Islamic banks‟ context, credit risks
predominantly take place within almost every applied instrument, even asset-based or risk-
based instruments. In this respect, Khan and Ahmad (2001) have shown that credit risks in
Islamic banks arise when the bank pays money, as it is the case with the Istisna‟ and Salam
2
The traditional risks which occur in the banking industry are principally credit risk, liquidity risk, market
risk and operational risk.
5
contracts, or delivers an asset before receiving its own cash, or provides assets as occurs in
the Murabahah contract. Furthermore, such a risk appears to occur more frequently in the
Mudarabah and Musharakah types of contracts whenever the entrepreneur appears to fail to
pay the bank share. Essentially, such a problem seems to persist mainly in case of dominant
asymmetric information prevailing between the borrower and the lender. In table (1) below,
we further detail the credit risks relevant to each transaction mode or technique used by
Islamic banking, based on the studies conducted mainly by Erico and Sundararajan
(2002), Boumedienne (2011) and Nurul-Kabir et al. (2015).
6
Table .1: Credit risk relevant to different Islamic banking contracts
Contract Associated Credit risk

Mudharabah Credit risk occurs in the case of nonpayment of the bank share (Rabbul-mal) by the
(Profit-sharing contract) entrepreneur (Mudharib) when it is due. In addition, within the Mudharabah
contract the bank does not participate in the decision-making process and is not in
the position to know, decide or monitor the Mudharib‟s activities. Consequently, a
problem may arise to the banks in case of prevailing high information asymmetry
environment, in which they do not dispose of sufficient information on the firm‟s
actual profit.
Musharakah (Profit and The relevant credit risk is dubbed “capital impairment risk”, that is, in case of
loss sharing contract) failure, the bank would lose the capital amount it had provided. Accordingly,
Boumedienne (2011) has argued that there is usually no credit risk associated with
Musharakah, because partners in such a contract type are logically selected
according to integrity, honesty and good reputation. Defaulter clients are rejected.
On the other hand, Erico and Sundararajan (2002) have highlighted that Islamic
banks are exposed to credit risk due to asymmetric information with the investing
partner.
Murabahah In the Murabahah contract, the bank buys and sells a commodity to the client at a
(A mark-up sale of a mark-up (profit). Credit risks take place whenever the buyer (the client) defaults or
merchandise) delays on payment after a due date. Moreover, this risk further increases if the
client cancels his option to buy and abandon the commodity.
Ijarah Credit risk involved in the Ijarah contract arises mainly when the customer (lessee)
(Financial leasing) fails to pay the lease rental when it falls due. Besides, within Ijarah, the lessee can,
in any time, change his mind and cancel the lease. In such a case, the bank will find
itself with an asset purchased with no return.
Salam Within Salam, credit risk takes place if (i) the customer does not honor payment (ii)
(Forward sale) the products are not delivered at all, or, are not delivered on time or else are not
conforming to specification (iii) the price does not cover the whole Salam capital.
Istisna’ The credit risk may consists in a failure to deliver the good, case in which the
(Order to manufacture) Istisna‟ contract can contain a penalty clause to overcome counterpart related risks,
unless the failure is due to factors lying beyond the manufacturer‟s control. In this
case, the bank will have to purchase from an alternative source usually at a
potentially higher price.
Credit risk may emerge once the purchaser in a parallel Istisna‟ defaults to pay
either the due installment or the price in full. In such a situation, the bank will have
the right to retain title of the manufactured good as a security until the last payment
installment has been received.
7
3. Literature review
Exploring the main causes, the consequences along with the compromises of credit risk
within the banking industry has drawn the interest of several authors particularly in
understanding the variables liable to this risk. In fact, to examine the potential factors of
credit risk, studies generally use different proxies of loan quality. However, the ratio of
impaired or non-performing loans to total loans (NPLs) is often used as the evidence that
the quantity or percentage of non-performing loans is frequently associated with bank
failures and financial crises (Khemraj and Pasha, 2009).
A great number of studies investigated the macroeconomic factors that affect credit risk.
In particular Fetsik and Beko (2008),Epinoza and Prasad (2010), Nkusu (2011), Farhan et
al. (2012), Bader and Javid (2013), Sharma Poudel (2013), Castro (2013), Love and Turk-
Ariss (2014), among others, concentrate their research essentially on the impact of
macroeconomic variables on the credit risk growth and conclude that those variables should
be included into the analysis since they have considerable effect on the changes of credit
risk.
Using monthly dataset covering the period between January 1995 and December 2006
to investigate determinants of problem loans of Hungary and Poland, Fetsik and Beko
(2008) reveal that nominal exchange rate, real gross domestic product, real interest rates
and real wages explain variations in NPLs. Epinoza and Prasad (2010) reached the same
conclusion when studying the GCC banking system. Their results highlight the importance
of economic growth and interest rates to the soundness of the banking system.
In a broader study spanning 26 advanced economies and covering period of 1998-2009,

Nkusu (2011) found that raised credit risk depend especially on a deteriorated economic
environment which is characterized by slower growth, higher unemployment rate, higher
interest rates and a fall in equity prices. Similarly, Farhan et al. (2012), when exploring the
Pakistani banking industry, found similar results and added evidence of a significant effect
of the interest and exchange rates. Within the same context, using the granger causality test,
Bader and Javid (2013) reported that macroeconomic indicators are the sizeable
determinants of NPLs. Jakubik and Reininger (2013) and Skarica (2014) found similar
results when analyzing NPLs‟ and factors in Central East and South East Europe (CESEE).
8
Beck et al. (2013a) estimated fixed effects and dynamic panel regressions for 75
advanced and emerging economies during the period 2000-2010. Real GDP growth, share
prices, the nominal effective exchange rates of the local currency and the bank lending
interest rate are found to significantly affect changes in the NPL ratio.
A group of other studies suggested that several bank specific factors are important
determinants of loan problems. For instance, Berger and DeYoung (1997) examined the
causes of banks‟ loan default by exploring the causality link between banking efficiency
and NPLs. They pointed out that managerial inefficiency significantly contributes to
banking troubles. In fact, four hypotheses were in examination: bad luck, bad
management, skimping and moral hazard.
Noteworthy, the studies of Louzis et al. (2012), Thehulu and Olama (2014), Boudriga et
al. (2009), Abbas and Ashraf (2014) and Shingjergi (2013), among others, are along this
line on this research. However, they have since proposed similar and other explanations for
bad or toxic loans.
Other authors, like Messai and Jouini, (2013), Boudriga et al.(2009), Monokroussous
and Thomakos (2014), Klein( 2013), Khmeraj and Pasha (2009), Boudriga et al. (2010),
Zribi and Boujelbene (2011) and Al-Wesabi and Ahmad (2013), among others, combine
the bank-level and country-level variables to explain credit risk.
In parallel, another strand in the literature used the vector autoregressive (VAR)
methodology to account for feedback effects of the deterioration of banks‟ loan quality on
the macro economy. A body of literature (Monokroussous and Thomakos, 2014; Saeed and
Izzeldin, 2014; Klein, 2013; Love and Turk-Ariss, 2014) emphasizes the strict assumption
of exogenous macro fundamentals in relation to problem loans. In a VAR system, all the
variables are endogenously determined, and the method allows for the implementation of
multiple shock scenarios that capture the interactions between the bank and the macro
variables.
4. Data and variables description
In this paper, our main objective is to identify the determinants of credit risk of Islamic
banks relative to their conventional counterparts. In accordance with the prior literature
review, we retain two kinds of variables: bank-specific and macro-economic determinants.
The bank-specific determinants are taken from Bankscope database, and the
9
macroeconomic ones are collected from the World Economic Outlook database of the
International Monetary Fund (IMF). We construct an unbalanced panel data comprising 30
Islamic and 87 conventional banks. The dataset consists of a panel of 10 OIC countries
(Organization of Islamic Cooperation)3 between 2005 and 2012 (See table (2) for more
details).
Table .2: Banks by country and region
Region Country Islamic banks Conventional banks All banks

MENA & Turkey United Arab emirates 6 9 15
Bahrain 5 8 13
Kuwait 3 9 12
Saudi Arabia 2 9 11
Qatar 3 7 10
Jordan 2 9 11
Tunisia 1 4 5
Turkey 3 16 19
Asia Bangladesh 2 10 12
Indonesia 3 6 9
Sum 30 87 117
4.1.Dependent variable
According to Berger and Deyoung (1997), Das and Gosh (2007), Al-Samadi (2010) and
Nurul-kabir et al. (2015), credit risk is measured by the NPL ratio that is the total amount
of nonperforming loans held by the bank devised by the total amount of loans. Simply, a
loan is defined as non-performing if payment of interests and principal are past due by 90
days or more, or if there are doubts payments can be made in full (Mamatzakis et al., 2015).
Higher NPL ratio indicates higher banking credit risk.
4.2.Independent variables
4.2.1. Bank-specific factors:
Among the bank-specific factors that could affect NPLs, we use the credit growth
(Credgr). According to Love and Turk-Ariss (2014), credit growth is positively related to
credit risk. That is rapid loan growth is negatively affected by adverse selection which
reduces the bank‟s asset quality. In addition, Jimenez and Saurina (2006) argue that
increased loans are due to the herd behavior and agency problem which encourage
3
We exclude countries that have adopted banking systems that consist only of Islamic banks such as Sudan,
Iran and Pakistan. The reason for limiting the data in this manner is to provide a better comparative analysis
of dual banking systems.
10
managers of banks to lend excessively. Also, Khemraj and Pasha (2009) admitted that
rapid credit growth is often associated with higher NPLs. However, when investigating the
Guyanese banking sector, they found that NPLs is negatively related to credit growth.
Similarly, Boudriga et al. (2010) found the same outcome in the context of the MENA
region. The authors explain that focusing on credit activities allows banks to better control
their borrowers‟ solvency and assess credit risk. Furthermore, Tehulu and Olana (2014)
support this result for the Ethiopian banks. Noteworthy, several studies found that change
in lending does not affect the level of NPLs (Messai and Juini, 2013, Love and Turk-Ariss,
2014).
We also mention the loan loss provisions as a possible factor than can affect NPLs.
Boudriga et al. (2010), Chaibi and Ftiti (2015), Ahmad and Ariff (2007) and Messai and
Jouini (2013) depict that loan loss provisions are positively related to NPLs. In fact, they
are a controlling mechanism of anticipated losses (Hasan and Wall, 2004). Hence, banks
that anticipate high levels of capital losses should create higher provisions to decrease
earnings volatility and reinforce medium term bank solvency (Boudriga et al., 2010).
Bank Profitability (ROA) is also often associated with bank risk. Indeed, greater
performance reduces the risk taking behavior of managers (Boudriga et al., 2010). In
addition, bank performance can reflect a high quality of management (Louzis et al., 2012).
Accordingly, the ROA is expected to be negatively correlated with NPLs. This result is
albeit affirmed by several papers, such as those of Messai and Jouini (2013), Chaibi and
Ftiti (2015), Shingjerji (2013), Makni et al. (2014), Zribi and Boujelbene (2011) and
Thehulu and Olana (2014).
We also sought to examine the validity of the „too big to fail‟ assumption under our
banking sample banks. In fact, bank size (approximated by the natural logarithm of total
assets) is often linked with lower credit risk (Zribi and Boujelbène, 2011; Boudriga et al.,
2010; Thehulu and Olana, 2014). This was justified by the fact that larger banks are more
diversified, therefore a better risk management process which permits to effectively deal
with doubtful borrowers and thus the bank will be less risky. There are other studies which
provide evidence of a positive association between NPLs and bank size. For instance,
Louzis et al. (2012) argue that large banks take excessive risks by increasing their leverage
under the „too big to fail‟ presumption and therefore have more NPLs. A similar result was
found by Chaibi and Ftiti (2015).
11
Investigating the managerial efficiency (CE) relevance with credit risk is also of our
preoccupations in this study. Berger and DeYoung (1997) underlined three hypotheses
surrounding the relationship risk-efficiency in the banking industry which are the “bad
management” hypothesis, the “bad luck” hypothesis and the “skimping” hypothesis. As for
the “bad management” hypothesis, low cost efficiency would cause higher toxic loans. In
fact, low efficiency is a sign of poor management practice and poor risk monitoring which
particularly leads to higher costs and therefore greater risks. The “bad luck” hypothesis
assumes that increased bank risks lead to a decrease of the bank efficiency. In particular,
unexpected external event can cause bad loans and this is unrelated to managers‟ control.
As a consequence, banks have to spend more resources to undertake these toxic loans
which arise to lower cost efficiency. The “skimping” hypothesis suggests that a decrease in
bank efficiency necessarily causes an increase in bank risk. Effectively, when bank
managers apply risk-averse strategy of management, albeit risk will be reduced and the
operating costs will rise in the short-run. Koutsomanoli-Filippaki and Mamatzakis (2009),
Fiordelisi et al. (2011), Louzis et al. (2012) among others, attempted to investigate these
hypotheses within a different context. Nevertheless, to our knowledge, only two studies
exploited these interactions in the context of the Islamic banking industry which are those
of Saeed and Izzeldin (2014) and Setiawan et al. (2014).
Bank capitalization is approximated by the equity to assets ratio (EQA). According to

Lee and Hsieh (2013), the relationship capital-risk is often discussed. This association can
be either positive or negative. In fact, Altunbas et al. (2007) attribute the positive
correlation between capital and risk to the “regulator hypothesis” that is the regulators
encourage banks to increase their capital proportionally to the amount of risk taken. On the
other hand, Berger and DeYoung (1997) refers the negative association to the “moral
hazard hypothesis” that is under-capitalized banks tend to be excessive risk-takers as a
response to moral hazard incentives on the part of banks‟ managers who increase the
riskiness of their loan portfolio (Louzis et al., 2012) which in turn result in higher NPLs.
This negative relationship between capital and NPLs was confirmed by several studies,
such as those of Makri et al. (2014) in the context of the Euro-zone banks, Klein (2013)
throughout Central, Eastern and South Eastern Europe banking industry and Zribi and
Boujelbene (2011) in the case of Tunisian banks.
12
4.2.2. Macro-economic factors
The first variable is an indicator of the overall economic activity; the growth rate of the
gross domestic product (GDPgr). It is widely accepted that economic expansion is
associated with higher employment and rising income which enable borrowers to service
their debts (Louzis et al., 2012; Makri et al., 2014). However, the recession time is
associated with the deterioration of a personal financial situation caused mainly by rising
unemployment (Chaibi and Ftiti, 2015; Bader and Javid, 2013; Sharma poudal, 2013). A
negative association is hence expected between GDPgr and NPLs.
Inflation ((Inf) is another variable to be considered. Previous studies found that this
factor can negatively and positively affect NPLs. In particular, based on Philips curve,
Castro (2013) highlighted that higher inflation implicates lower unemployment and then
higher income leads to the decrease in bad loans. Thus, the impact of Inflation is negative
in this case. On the other hand, Nkusu (2011) emphasizes that higher inflation reduces the
real income and weakens the borrowers‟ ability to repay their debts in time. In this case,
inflation has a positive correlation with NPLs.
We include an aggregate governance index (GOV) compiled via the six governance
indicators of Kaufman et al. (2008) to deal with the impact of the institutional environment
on the bank‟s credit quality. Indeed, this index includes six dimensions of governance
which are: 1) voice and accountability, 2) political stability and violence, 3) government
effectiveness, 4) regulatory burden, 5) rule of low, and 6) Control of corruption. Boudriga
et al. (2010) argue that well functioning institutions coupled with good governance increase
the performance of the financial system which surely results in lower risk. Similarly, Nurul-
Kabir et al. (2015) supported that good governance reduces banking credit risk.
Accordingly, we expect a negative relationship within NPLs.
Finally, we include a dummy variable to control for the financial crisis period (FinCrisis). It
takes the value of one for the years 2008 and 2009 and 0 otherwise (see. Beck et al.,
2013b).
A complete description of all the variables used in this study can be found in table (3).
13
Table 3: Description of the variables
Variables Notation Description Sources

Dependent
Credit risk NPLs Nonperforming loans to total loan ratio Bankscope database
Bank-specific
Credit growth CredGR Credit growth rate Bankscope database
Provisions Prov Loan loss provisions to total loan ratio Bankscope database
Cost efficiency CE Cost efficiency score estimated using Stochastic Authors‟ calculation using
Frontier Analysis data from Bankscope
Profitability ROA Return on asset ratio Bankscope database
Capitalization EQA Total Equity to Total Assets Bankscope database
Size Size Natural logarithm of total assets Bankscope database
Macroeconomic
GDP growth GDPgr Growth rate of gross domestic product World bank
Inflation INF Inflation rate World bank
Governance GOV An aggregate index of governance which Author‟s calculation using
includes: 1)Voice and accountability; 2) Political World governance
instability and violence; 3)Government Indicators compiles by
effectiveness; 4)regulatory burden; 5)Rule of low Kaufman et al., (2008)
and 6)Control of corruption
Crisis period Crisis Dummy variable taking the value of one in 2008
and 2009 years and zero otherwise
5. Methodology
We employ two complementary methods to assess credit risk determinants. First, we

check for the different factors which can affect loan quality by employing one-step
generalized method of moment (GMM) system estimation method. Then, we implement a
panel vector autoregression model (PVAR) to test the different hypotheses concerning the
causality link of credit risk with banking efficiency and to further assess the extent to which
bank-specific and macroeconomic shocks affect the two business-model banking industries.
5.1.Modeling NPLs
This study follows the empirical specification proposed by Tan (2015), which can be
expressed as follows:
∑ ∑
[1]
Where i refer to year and t refers to bank, represents loan quality indicator (i.e.
credit risk proxy) for a specific bank at a specific year, and is the first lagged
dependent variable which captures the persistence in loan quality over time. refers
14
to bank-specific control variables namely credit growth ( ), loan loss provisions
( ), cost efficiency ( ), equity to assets ( ) and a proxy for the influence of
bank size ( ). Two macroeconomic control variables are also used that are
the annual growth of real gross domestic product ( ) and the annual inflation rate
( ). In addition, we include the governance index ( ) to control for the
institutional environment within which the banking system operates. We include a crisis
dummy ( ) to check for the possible pressure in the 2008 financial crisis period.
Finally, the unobserved bank-specific effect and the idiosyncratic error are represented by
and respectively. and are coefficients to be estimated using one-step dynamic
panel estimation performed by the general method of moments (GMM) system estimator
developed by (Bludell and Bond, 1998).
5.2.Panel vector auto-regression model
We use a panel vector autoregression (PVAR) model developed by Love and Zicchino
(2006) to investigate the linkage between credit risk and various bank-levels and macro-
economic correlates. The framework adapted to carry out our analysis is closely matches
that of Love and Turk-Ariss (2014). The authors argue that the advantage of the PVAR is
that it accounts for individual bank specificity in the level of the variables by introducing
fixed effects (μi) and isolating the response of the bank credit channel to macroeconomic
shocks while allowing for unobserved bank heterogeneity. The model is written as:
Where Θ (L) is the lag operator and yit is a vector of variables. To avoid obtaining biased
coefficients that result from the correlation between the fixed effects and the regressors, the
Helmert procedure is used following Love and Zicchino (2006) to remove only forward
mean. This procedure preserves the orthogonality between the transformed variables and
the lag regressors, making it possible to use lagged regressors as instrument and estimate
equation [2] by system GMM (Arellano and Bover, 1995).
We use the Cholesky decomposition to identify orthogonal shocks in our variables and
examine their effect on the remaining variables in the system maintaining other shock
constant. To analyze the response of one variable to an orthogonal shock in another
variable, we focus on the impulse-response functions (IRFs); the response of one variable
to a shock in another variable. We generate confidence intervals for the orthogonalized
15
IRFs with Monte Carlo simulations and identify the response to one shock at a time while
keeping other shock constant.
6. Empirical results and Interpretations
First, in table (4) we present a summary of the retained variables statistics for all the
sample banks, Conventional and Islamic banks.
Table 4: Descriptive statistics of variables’ model by bank’s business model (average

values)
All sample banks Conventional banks Islamic banks
Obs. Mean S.D. Obs. Mean S.D. Obs. Mean S.D.
CredRisk 846 0.049 0.056 647 0.054 0.063 199 0.053 0.067
Credgr 843 0.229 0.328 639 0.209 0.274 204 0.294 0.452
Prov 860 0.013 0.039 656 0.011 0.016 204 0.022 0.076
ROA 877 0.022 0.042 665 0.019 0.021 212 0.033 0.077
EQA 877 0.144 0.079 665 0.133 0.066 212 0.178 0.104
Size 877 6.779 0.623 665 6.840 0.634 212 6.589 0.545
GDPgr 877 5.508 4.599 665 5.515 4.492 212 5.482 4.626
INF 877 5.554 3.768 665 5.675 3.694 212 5.175 3.979
GOV 877 0.291 0.771 665 0.224 0.782 212 0.449 0.730
With regard to our dependent variable which is measured by NPLs ratio, table (4)
shows that Islamic and conventional banks have nearly the same credit risk levels.
However, if we observe the evolution over time of NPLs ratio (see Fig. 1), we notice that
credit risk of conventional banks increases in 2007 until 2009 presumably due to the global
financial crisis which worsened since the end of 2007. As regards the Islamic banks‟ NPLs,
it is noticed that credit risk generally oscillates between 4% and 6% over all the study
period with a slight decrease during 2007. Consequently, one can conclude that the effect of
the global financial crisis had harmful effects for the conventional banking industry as
compared to the Islamic banking one.
Noteworthy, Islamic banks show a higher provisions ratio which indicate that interest-free
banks adopt more prudent loan loss provisioning relative to the interest-based banks. We
also notice that Islamic banks outweigh their conventional counterparts regarding their
profitability and their capitalization. Concerning their size, Islamic banks are on average
somewhat largest than conventional banks.
16
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
2005 2006 2007 2008 2009 2010 2011 2012
Islamic banks Conventional banks
Fig.1. NPLs ratio by years and business model
Next, we present table (5), which reports the cost efficiency scores of all the sample
banks, both Islamic and conventional banks, respectively. In table A.1, We also depict the
translog cost function as derived under SFA for all the sample banks over the period 2005-
2012. The parameter indicates the proportion of the variance in
disturbance due to inefficiency. The γ value, which takes a value between 0 and 1, shows
the contribution of the inefficiency term u to the dichotomous term v+u. The estimated γ
value for the cost frontier model is high (0.918) which shows that inefficiency variation is
more important than any stochastic variation in the frontier model.
As depicted in table 5, conventional banks are most cost efficient, with a mean cost
efficiency score of 94% (compared to 84% for Islamic banks). According to Olson and
Zoubi (2008), the relative inefficiency of Islamic banks could be attributed to the
predisposition of customers for Islamic products regardless of their cost. Over time, (see
Fig.1), cost efficiency for all the banks in the sample shows a declining trend, however that
of Islamic banks fell slightly after 2007 from 86% to 84% then to 82% in 2010 before
climbing up to the pre-crisis scores (86%) since 2011. As for conventional banks, their cost
efficiency was oscillating between 92% and 94% till the period of 2007 but did not increase
compared to previous periods (97% and 96%).
17
Table 5: Cost efficiency scores by year and business model (average values)
Year All sample banks Islamic banks Conventional banks

2005 0.90 0.78 0.97
2006 0.89 0.82 0.96
2007 0.88 0.86 0.93
2008 0.87 0.84 0.94
2009 0.86 0.84 0.94
2010 0.84 0.82 0.93
2011 0.84 0.85 0.94
2012 0.84 0.86 0.92
Mean 0.86 0.84 0.94
1
0.95
0.9
0.85
0.8
0.75
0.7
2005 2006 2007 2008 2009 2010 2011 2012
Islamic banks Conventional banks
Fig.2. Cost efficiency by region and business model
6.1.Dynamic panel estimation results:
In order to test for the validity of the instruments, we employ the Sargan test for the over-
identifying restrictions in the GMM estimation. Results provided in table (6) suggests that
the model is valid and do not suffer from over-identifying problem. In addition, the non-
significance of the AR (2) statistics indicates the consistency of the GMM estimates. Also,
the lagged dependent variable is positive with significant coefficient across all
specifications which prove the dynamic character of model specification (Daher et al.,
2015; Tan, 2015). At this level, we validate the choice of a dynamic specification for our
model. Results are depicted in table (6) below.
18
Table .6: Estimation results: Estimation method is the Arellano and Bond (1991) one-step GMM difference estimator for panel data with
lagged dependent variable

(1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5)
NPL-1 0.806*** 0.839*** 0.805*** 0.809*** 0.820*** 0.534*** 0.637*** 0.615*** 0.609*** 0.626*** 0.805*** 0.809*** 0.800*** 0.785*** 0.781***
Credgr -0.023*** -0.017*** -0.029*** -0.026*** -0.019*** -0.027*** -0.018*** -0.029*** -0.024*** -0.010 -0.020*** -0.018*** -0.018*** -0.022*** -0.023***
Prov 0.336*** 0.306*** 0.302*** 0.306*** 0.252*** -0.031 0.074 0.053 -0.031 -0.088 1.133*** 1.278*** 1.045*** 1.354*** 1.331***
CE 0.101*** 0.073** 0.057 0.016 0.0008 0.030 -0.018 0.014 -0.008 -0.015 0.068* 0.036 0.062* 0.032 0.029
ROA 0.034 0.085* 0.039 0.027 0.047 -0.723*** -0.490*** -0.541*** -0.624*** -0.511*** -0.009 0.011 0.001 0.005 0.010
EQA -0.009 -0.091 -0.083 -0.002 -0.079 0.245*** 0.177*** 0.152** 0.256*** 0.187*** 0.046 0.007 -0.064 0.009 -0.133
Size 0.00001 -0.011 -0.015 -0.011 -0.027*** -0.0008 0.003 -0.011 -0.006 -0.017** -0.006 -0.00002 -0.021* -0.009 -0.034***
GDPgr - -0.001*** - - -0.002*** - -0.001*** - - -0.002*** -0.00009 0.001
INF - 0.0003 - - 0.0005 - 0.0002 - - 0.0004 0.0002 0.0004
Gov - - 0.020*** - 0.021*** - - 0.005 - 0.007* 0.201** 0.027***
Crisis - - - 0.007*** -0.001 - - - 0.007*** -0.0001 -0.003 -0.002
No. Obs. 731 731 731 731 731 555 555 555 555 555 166 166 166 166 166
Instruments 49 58 57 54 56 49 58 57 54 56 56 58 57 54 56
Sargan test 183.77*** 189.02*** 170.51*** 209.27*** 171.3*** 139.35*** 160.67*** 162.77*** 170.35*** 151.86*** 58.6** 78.71*** 66.58** 73.17*** 60.42*
AR (1) test -9.58*** -9.64*** -8.95*** -9.58*** -9.04*** -4.64*** -5.44*** -5.67*** -5.25*** -5.25*** -5.06*** -5.08*** -5.14*** -4.79*** -4.66***
AR (2) test 0.90 1.22 0.77 0.75 0.89 -0.60 -0.14 -0.45 -0.50 -0.21 0.13 0.07 -0.023 0.16 -0.28
*Denote significance at 10% level

**Denote significance at 5% level
***Denote significance at 1% level
19
Regarding the bank-specific variables; credit growth depicts a negative and strongly
significant relationship with credit risk by the three split simple adopted. This result, which
is supported by several studies (Khemraj and Pasha, 2009; Boudriga et al., 2010; Das and
Ghosh, 2007; Tehulu and Olana, 2014), means that excessive lending strategy leads to the
deterioration of the financial health of banks. In the same vein, Boudriga et al. (2010) argue
that focusing on lending activities allows banks to better control of borrowers‟ solvency
which improves credit risk assessment. It is interesting to note that our finding contradicts
previous ones which report a positive association between credit growth and NPLs (Salas
and Saurina, 2002; Jimenez and Saurina, 2006).
As expected, the coefficient of the loan loss provisions is positive and statistically
significant at 1% level within all sample banks and Islamic banks specifications. This
implies that credit risk increases with high provisioning banks. Therefore, banks need to
make greater provisions when loans tend to be potentially impaired. This result is consistent
with the findings of Ahmad and Ariff (2007), Chaibi and Ftiti (2015) and boudriga et al.
(2010). Noteworthy, in the case of the conventional banks, PROV appear to be negatively
related to NPLs but with no significant coefficient.
Cost efficiency is found to be positively associated with NPLs in the most cases which
appear a significant statistical level (i.e. all banks and Islamic banks). This seems consistent
with the moral hazard and skimping hypothesis which state that an increase in bank
efficiency increases banking risk. This finding is consistent with that of Chaibi and Ftiti
(2015) in the case of the German banking system. Indeed, the authors explained this
positive association by the fact that NPLs increase with short-term cost efficiency instead
of with bad management practices.
Profitability as approximated by ROA reveals a negative and statistically significant

relationship with NPLs of conventional banks. Our empirical evidence is consistent with
the findings of Chaibi and Ftiti (2015), Zribi and Boujelbene (2011), Boudriga et al. (2010)
and Messai and Juini (2013) who argue that high profitability banks are less pressured
regarding the revenue creation and consequently less constrained to engage in risky credit
offering.
Banks capitalization is positively related to credit risk solely within conventional

banks‟ case. This funding underlines that strong capitalization lead these bank to an
excessive risk-taking behavior which bring on higher risky assets and higher bad loans.
20
Noteworthy, our result don‟t support the moral hazard hypothesis of Berger and Deyoung
(1997) who explained that destroyed capital leads banks to increase the riskiness of their
loan portfolio which results in higher NPLs in the future.
With regard to the bank size, results indicate that the impact of bank size is negative
albeit significant only within the model (5) of the three specifications. This result diverges
from the results of Khemraj and Pasha (2009) and Louzis et al. (2012). However, it
converges with Boudriga et al. (2010) highlighting that larger banks possess more
resources and are more experienced in dealing with better borrowers unlike small banks
which are exposed to adverse selection problems caused by insufficient competences to
effectively assess the credit quality of borrowers.
Regarding the macroeconomic factors, GDPgr is found to be highly significant and

negatively related to NPLs, as expected. This result suggests that good economic conditions
improve the borrowers and lenders‟ confidence to repay their loans, which reduces banks‟
credit risk (Sharma poudel, 2013). Our funding is in line with Chaibi and Ftiti (2015),
Khemraj and Pasha (2009), Zribi and Boujelbene, (2011) and Messai and Jouini (2013).
Inflation exhibits a negative albeit not significant relationship with credit risk. Chaibi and
Ftiti (2015) explain this negative sign by the fact that higher inflation weakens borrowers‟
ability to service debt by reducing their real income. Unexpectedly, results report that
government effectiveness positively affects credit risk. Our finding is not consistent with
that of Godlewski (2004) and Boudriga et al. (2010) who report a negative relation between
institutional environment and NPLs. Noteworthy, GDPgr denote no significant association
with Islamic banks‟ credit risk. We notice that the inflation rate (INF) denotes an
insignificant association with credit risk in the whole models studied. Unexpectedly, an
improved institutional environment coupled with privileged governance power (GOV)
depicts a positive and statistically significant correlation with credit risk. In this case, we
can notice that regions undertaken in our sample mainly suffers from poor regulatory
power.
Finally, the crisis dummy variable show a positive significant coefficient in the case of
all sample banks and conventional banks specifications. However, no statistically
significant effect is found in the case of Islamic banks. Accordingly, the subprime crisis
causes surely an increase in toxic loans, which is not the case for Islamic banks. At this
21
level, we can affirm that Islamic banks were immunized from the harmful effect of such
crisis.
6.2.Panel VAR analysis
At this level, we sought to investigate how various macro-economic shocks affect bank
level variables in both banks‟ business model. In other words, is an interest-free banking
system more vulnerable in front of macro-economic forces? Results are reported below.
6.2.1. Risk- efficiency issue
The response of interest-based banks to credit chocks is expected to be different from

that of the interest-free banks due to the nature of these latter which only involve with
interest-free instruments. We conduct a panel VAR model with two variables (NPLs and
cost efficiency) within three different panel specifications. Table (7) below display the
estimation results of the system GMM coefficients of the baseline PVAR, table 8 present
variance decompositions (VDCs) and Fig.3 graphs the corresponding impulse-response
functions (IRFs).
Table 7: Main results of a 2-variables VAR model
Panel A: All sample banks Panel B: Conventional banks Panel C: Islamic banks
CE NPL CE NPL CE NPL
CE-1 0.371 -0.018 0.322 -0.090 0.512 0.105
(3.93)*** (-0.44) (2.382)*** (-0.96) (5.21)*** (2.01)**
NPL-1 0.149 0.714 0.004 0.708 0. 109 0.678
(1.98)** (4.29)*** (0.122) (4.83)*** (0.338) (2.47)***
* Denote significance at 10% level

For all sample banks, the effect of NPLs on cost efficiency is found to be positive
and significant at 5% level indicating that the causality would run from bank risk to
efficiency. However, the impact of cost efficiency on problem loans is found to be negative
but not significant. Regarding the conventional banks case, it appears that cost efficiency
performs a poor and non significant response to a variation in credit risk, whereas credit
risk reacts negatively to a change in cost efficiency, although both reactions present
insignificant coefficients. For Islamic banks, the credit risk impact on cost efficiency is
positive but insignificant, whereas the reverse causation is significantly positive at 5%
level. The positive response of credit risk to a variation in efficiency implies that increased
efficiency leads to increased risk.
22
To a full comprehension of the findings, however, one must analyze the impulse
response functions generated that report the response for each VAR variable to its own
innovation and the innovation of other variable. Results are reported in Fig.3.
Impulse-responses for 1 lag VAR of CE NPLL Impulse-responses for 1 lag VAR of CE NPLL Impulse-responses for 1 lag VAR of CE NPLL
(p 5) CE CE (p 5) NPLL NPLL (p 5) CE CE (p 5) NPLL NPLL (p 5) CE CE (p 5) NPLL NPLL
(p 95) CE (p 95) NPLL (p 95) CE (p 95) NPLL (p 95) CE (p 95) NPLL
0.0307 0.0085 0.0226 0.0014 0.0565 0.0338
-0.0003 0.0000 -0.0001 -0.0013 -0.0024 -0.0154

0 6 0 6 0 6 0 6 0 6 0 6
s s s s s s
response of CE to CE shock response of CE to NPLL shock response of CE to CE shock response of CE to NPLL shock response of CE to CE shock response of CE to NPLL shock
(p 5) CE CE (p 5) NPLL NPLL (p 5) CE CE (p 5) NPLL NPLL (p 5) CE CE (p 5) NPLL NPLL
(p 95) CE (p 95) NPLL (p 95) CE (p 95) NPLL (p 95) CE (p 95) NPLL
0.0029 0.0273 0.0033 0.0250 0.0106 0.0735
-0.0023 0.0000 -0.0051 0.0000 -0.0076 -0.0006

0 6 0 6 0 6 0 6 0 6 0 6
s s s s s s
response of NPLL to CE shock response of NPLL to NPLL shock response of NPLL to CE shock response of NPLL to NPLL shock response of NPLL to CE shock response of NPLL to NPLL shock
Errors are 5% on each side generated by Monte-Carlo with 500 reps Errors are 5% on each side generated by Monte-Carlo with 500 reps Errors are 5% on each side generated by Monte-Carlo with 500 reps
Fig.3. Impulse Response Functions to shocks, PVAR Baseline Model with two variables: NPL-CE
i. All sample banks
It appears that the effect of a one standard deviation shock on credit risk to cost
efficiency is positive and large in magnitude. The peak response of efficiency to a shock in
the credit risk takes place after 2 years while it converges towards the equilibrium
thereafter. Consequently, a shock in the credit risk, that would increase NPLs, enhances
cost efficiency in the short run of the first 2 years. This outcome can be explained in terms
of the moral hazard and skimping hypotheses. For instance, banks become efficient in that
they devote fewer resources to risk monitoring activities. The response of credit risk to a
shock in cost efficiency is positive for the first period, while it turns toward the equilibrium
thereafter. In particular, it implies that this relationship might be positive in a short run of
two years but it turns to the equilibrium thereafter which is in accordance with the bad
management hypothesis. Specifically, the bad management hypothesis states that lower
efficiency leads to higher risk. Here increased efficiency lead to decreased risk implying
that good management practices prevent the bank from default risk.
To shed light into our analysis, we also present VDCs which show the percent of the
variation in one variable that is explained by the shock in another variable. We report the
total effect accumulated over 10, 20 and 30 years in table (8). Results provide further light
to IRFs, insinuating the importance of risk in exploring the variation of efficiency.
Specifically, close to 5% of efficiency forecast error variance after 30 years is explained by
credit risk‟s disturbances. On the other hand, a small part, less than 0.6% of the variation of
23
credit risk is explained by efficiency. This result implies that causality would run from risk
to efficiency which confirm the IRFs results and give support to moral hazard and skimping
hypotheses.
ii. Interest-Based relative to Interest-Free banks
Regarding the interest-based banks, Fig. 3 shows a close to zero effect of NPLs on cost
efficiency for the whole study period. On the other hand, the reaction of credit risk to a one
standard deviation shock in cost efficiency is initially positive and large in magnitude with
a smaller confidence interval, and then it turns to be negative before converging to the
equilibrium afterward. Therefore, since the reaction of NPLs to a variation in the cost
efficiency is much more expressive, the idea of bank efficiency preceding problem loans is
reinforced. This is very much on line with the moral hazard and skimping hypotheses as in
the case of all sample banks. Noteworthy that credit risk turn to be negative after the first
year‟s shock which can be attributed to the bad management hypothesis which essentially
argues that increased inefficiency lead to increased risk. Here, enhanced efficiency causes
lower risk on the long run that is good management practices necessarily provoke lower
credit risk.
Turning to interest-free banks (IFBs), fig.3 shows that the response of cost efficiency to
NPLs‟ innovation is estimated close to zero for the whole period. On the other hand, the
impact of one standard deviation shock of cost efficiency on NPLs is initially negative,
although it turns to be positive after the first year. In other words, increasing efficiency
leads to lower risk in the short run of one period that is good managers are able to better
monitor and control their loans activities. However, in the long run, this relationship trails
back to moral hazard and skimping hypotheses suggesting that increased efficiency
increases risk.
Table (8) further shows that VDCs estimations results demonstrate that variation in cost
efficiency explained by credit risk‟s disturbances are much higher for Islamic banks than
for conventional ones (1.8% in the case of IBs as opposed to less than 0.0001% in the case
of conventional banks). Islamic banks also outperform their counterpart in the case of
forecast variance for credit risk. In particular, it is observed that 5% of the forecast error
variance of credit risk after 30 years is explained by shocks in cost efficiency falling to
0.6% in the case of conventional banks.
24
Table 8: Variance decompositions (VDCs) for cost efficiency and NPLs

CE NPL CE NPL CE NPL
CE 10 0.94299898 0.05700102 0.99994100 0.00005900 0.98176333 0.01823667
NPL 10 0.00081042 0.99918958 0.00649487 0.99350513 0.04980013 0.95019987
CE 20 0.94276100 0.05723900 0.99994078 0.00005922 0.98158472 0.01841528
NPL 20 0.00081011 0.99918989 0.00651184 0.99348816 0.05026703 0.94973297
CE 30 0.94276077 0.05723923 0.99994078 0.00005922 0.98158438 0.01841562
NPL 30 0.00081011 0.99918989 0.00651185 0.99348815 0.05026794 0.94973206
6.2.2. Feedback from banking system to the real economy: a panel analysis of
economic and financial shocks and loan portfolio quality
We investigate the macroeconomic and dynamic consequences of an increase in NPLs

using a panel VAR specification. In particular, we were interested in the effect of an
increase in NPLs on both internal and external factors. To this end, we estimate the
coefficient of the system given in equation (2). We report the model with five variables
(GDPgr, INF, UNEMP, NPLs, EQA and ROA) in table (9). Results include panel A
including all sample banks, panel B comprising IBBs and panel C containing IFBs. We
present graphs of the impulse response function and the 5% error bands generated by
Monte Carlo simulation; Fig 4.a. report graphs of impulse responses for panel A, while Fig
4.b and Fig 4.c. presents similar graphs for panel B and C, respectively.
25
Table 9: Main results of a 5-variables VAR model
Panel A: All sample banks

GDPgr INF NPLs EQA ROA
GDPgr-1 0.336 (7.13)*** -0.022 (-0.59) -0.0007 (-2.58)*** -0.0006 (-2.28)** 0.0003 (1.03)
INF-1 -0.133 (-2.85)*** 0.355 (5.23)*** 0.001 (4.04)*** -0.0007 (-1.79)* -0.0006 (-1.47)
NPL-1 7.791 (1.63) 2.692 (0.48) 0.704 (4.48)*** 0.049 (0.29) 0.032 (0.15)
EQA-1 6.641 (0.87) -7.358 (-0.96) -0.006 (-0.05) 0.409 (3.61)*** 0.078 (0.522)
ROA-1 2.332 (0.33) 1.463 (0.33) -0.099 (-1.41) 0.032 (0.25) 0.413 (3.33)***
Panel B: Conventional banks
GDPgr-1 0.308 (5.85)*** -0.016 (-0.37) -0.0008 (-2.62)*** -0.0007 (-3.89)*** 0.0002 (1.64)*
INF-1 -0.119 (-2.13)** 0.320 (4.11)*** 0.001 (3.94)*** -0.0002 (-0.97) -0.0004 (-2.31)**
NPL-1 11.257 (1.62) 5.367 (0.92) 0.644 (4.06)*** 0.064 (1.67)* 0.003 (0.10)
EQA-1 18.977 (1.21) -18.278 (-1.63) 0.008 (0.05) 0.621 (5.03)*** -0.044 (-0.24)
ROA-1 56.371 (3.42)*** 12.184 (1.19) -0.347 (-1.83)* -0.025 (-0.21) 0.326 (2.94)***
Panel C: Islamic banks
GDPgr-1 0.393 (4.69)*** -0.097 (-0.42) -0.0002 (-0.42) 0.0002 (0.33) 0.0005 (1.11)
INF-1 -0.073 (-0.89) 0.406 (3.11)*** 0.001 (2.08)** -0.001 (-1.005) -0.001 (-0.79)
NPL-1 9.015 (1.41) 3.346 (0.34) 0.677 (3.25)*** 0.104 (0.35) 0.113 (0.31)
EQA-1 1.355 (0.27) -2.772 (-0.40) -0.017 (-0.14) 0.241 (2.04)** 0.078 (0.37)
ROA-1 -1.219 (-0.28) -0.096 (-0.02) -0.015 (-0.28) 0.063 (0.55) 0.482 (3.17)***
*Denote significance at 10% level

***Denote significance at 1% level
We proceed now to the analysis of mutual interactions between individual variables.

We examine the effect of innovation in one of the variables on some or all other variables
included in the model. This is done through impulse response analysis.
i. All sample banks:
First, we quantify the effect of a one standard deviation shock in each of the two
systematic (macro-economic) factors on bank‟s credit risk (see Fig.4a). In particular, a one
standard deviation shock to GDPgr translates to a decline in credit risk. Also, a one
standard deviation shock to inflation rate result on a significant increase in NPLs ratio.
Effectively this outcome would suggest favorable economic conditions (i.e. increased
GDPgr) lead to ameliorate the loan quality of banks and thus minimizing credit risk. In
addition, higher inflation rate increase NPLs ratio which is expected as high inflation rates
26
are generally associated with a high loan interest rate (Sharma Poudal, 2013). Thus, high
interest rates increases cost of borrowing which lead to an increase in the obligation of
borrowers resulting in an increase in the doubtful debt and accordingly higher credit risk.
Second, we analyses various interactions among bank-level variables. Results show

that credit risk responds negatively to profitability and capital innovations. This result is
expected and shows that improved profitability coupled by strongest capitalization of bank
lead to lower credit risk.
Additionally, in table (10) (panel A) we present variance decompositions results to

further support our analysis. We observe that, after thirty years, GDPgr and INF explain
about 9% and 5% of the total forecast error variance in NPLs ratio, respectively. While,
both ROA and EQA explain about, only 2% of the total variance in credit quality.
Effectively, theses outcomes demonstrate that macro variables have a somewhat large
explanatory power of credit risk. This result is consistent with Love and Turk-Ariss (2014)
when exploring the Egyptian banking industry.
Impulse-responses for 1 lag VAR of GDPGR INF NPLL EQA ROA

(p 5) GDPGR GDPGR (p 5) INF INF (p 5) NPLL NPLL (p 5) EQA EQA (p 5) ROA ROA
(p 95) GDPGR (p 95) INF (p 95) NPLL (p 95) EQA (p 95) ROA
3.9549 0.1072 0.4629 0.5246 0.3904
-0.1473 -0.7202 -0.0490 -0.1304 -0.2358

0 6 0 6 0 6 0 6 0 6
s s s s s
response of GDPGR to GDPGR shock response of GDPGR to INF shock response of GDPGR to NPLL shock response of GDPGR to EQA shock response of GDPGR to ROA shock
0.5208 3.4540 0.3220 0.1196 0.2389
-0.2451 -0.0387 -0.1742 -0.4835 -0.1503

0 6 0 6 0 6 0 6 0 6
s s s s s
response of INF to GDPGR shock response of INF to INF shock response of INF to NPLL shock response of INF to EQA shock response of INF to ROA shock
0.0000 0.0081 0.0262 0.0024 0.0002
-0.0086 -0.0030 -0.0000 -0.0098 -0.0082

0 6 0 6 0 6 0 6 0 6
s s s s s
response of NPLL to GDPGR shock response of NPLL to INF shock response of NPLL to NPLL shock response of NPLL to EQA shock response of NPLL to ROA shock
0.0020 0.0017 0.0083 0.0299 0.0074
-0.0050 -0.0054 -0.0064 -0.0013 -0.0051

0 6 0 6 0 6 0 6 0 6
s s s s s
response of EQA to GDPGR shock response of EQA to INF shock response of EQA to NPLL shock response of EQA to EQA shock response of EQA to ROA shock
0.0056 0.0027 0.0123 0.0184 0.0293
-0.0034 -0.0054 -0.0073 -0.0027 -0.0021

0 6 0 6 0 6 0 6 0 6
s s s s s
response of ROA to GDPGR shock response of ROA to INF shock response of ROA to NPLL shock response of ROA to EQA shock response of ROA to ROA shock
Errors are 5% on each side generated by Monte-Carlo with 500 reps
Fig.4a. IRFs for all sample banks, (Model with 5 variables: GDPgr-INF-NPLs-ROA-EQA)
27
ii. Interest-Based relative to Interest-Free banks:
Fig.4b and Fig.4c reports the IRFs of IBBs and IFBs, respectively. On one hand, the
IBBs‟ credit risk responds negatively to GDPgr and INF shocks. In fact, the immediate
response to the INF is negative albeit it turns to be positive after the first year, suggesting
that an increased Inflation rate have a long term positive impact on credit risk. Also, IRFs
report that NPLs decrease with strongest capitalization and improved profitability.
However, note that the response to the profitability shock is larger in magnitude which
indicates that enhancing profitability occurs more on credit risk attenuation.

3.7854 0.1164 0.5270 1.1224 0.9303
-0.3536 -0.7346 -0.0601 -0.1793 -0.1624

0 6 0 6 0 6 0 6 0 6
s s s s s
0.6024 3.3779 0.2382 0.0231 0.3189
-0.2044 -0.0368 -0.1617 -0.5169 -0.0626

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0014 0.0062 0.0227 0.0039 0.0000
-0.0092 -0.0027 -0.0001 -0.0100 -0.0078

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0000 0.0014 0.0046 0.0168 0.0018
-0.0062 -0.0030 -0.0001 -0.0010 -0.0031

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0020 0.0006 0.0005 0.0065 0.0117
-0.0014 -0.0022 -0.0027 -0.0051 -0.0002

0 6 0 6 0 6 0 6 0 6
s s s s s
Fig.4b. IRFs for conventional banks, (Model with 5 variables: GDPgr-INF-NPLs-ROA-EQA)
On the other hand, the IFBs‟ credit risk responses are, almost the same of the IBBs‟
responses. We observe that credit risk decreases in response to a positive GDPgr shock and
increase in response to a positive shock on Inflation rate. Noteworthy, the peak response to
GDPgr shock take place after 3 years, while it converge towards the equilibrium for the rest
of the period. Effectively, this result highlight that for IFBs, an enhancement of the
economic cycle reduces credit risk in the short run of 3 years. Fig.4c further shows that the
response of NPLs to ROA and EQA shocks appear to be nearly the same. In fact, the effect
of one standard deviation shock of both ROA and EQA on NPLs is negative; NPLs display
a slight decrease then it turns to zero therefore and similarly the EQA ratio. This last
28
finding provides some evidence in favor of the moral hazard hypothesis. In this respect it is
evident that Islamic banks should improve their capital to tackle for toxic loans.

3.8120 0.2482 0.7335 0.3832 0.2838
-0.1235 -0.6987 -0.0625 -0.3592 -0.3828

0 6 0 6 0 6 0 6 0 6
s s s s s
0.5003 3.7757 0.6296 0.3276 0.3200
-0.7888 -0.1055 -0.4197 -0.6110 -0.3260

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0039 0.0131 0.0357 0.0062 0.0034
-0.0091 -0.0042 -0.0001 -0.0139 -0.0083

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0109 0.0108 0.0150 0.0526 0.0122
-0.0030 -0.0139 -0.0214 -0.0021 -0.0062

0 6 0 6 0 6 0 6 0 6
s s s s s
0.0241 0.0084 0.0310 0.0422 0.0557
-0.0051 -0.0182 -0.0189 -0.0084 -0.0023

0 6 0 6 0 6 0 6 0 6
s s s s s
Fig.4c. IRFs for Islamic banks, (Model with 5 variables: GDPgr-INF-NPLs-ROA-EQA)
Turning to the VDCs estimates of the two types of banks. For IBBs, GDPgr seems
to have the larger explanatory power for bad loans explaining around 14% of the variation
in NPLs, while INF explains 4.2 %. ROA and EQA account for 1.6% and 4.8%,
respectively. Similarly, in the case of IFBs, macroeconomic variables still have the larger
explanatory power but with a much smaller power than the IBBs. In particular, close to
6.5% of credit risk‟s forecast error variance is explained by INF rate disturbances, whereas
the GDPgr explains less than 0.75%. Regarding bank level variables, profitability only
explain 0.16% of the forecast error variance for NPLs after 30 years. Additionally, close to
0.5% of the forecast error variance in NPLs is explained by the capital adequacy ratio.
29
Table 10: Variance decomposition (s = 30 years)
Panel A: All sample banks

GDPgr 0.96583371 0.01995374 0.00751902 0.00630217 0.00039136
INF 0.00776707 0.98489841 0.00146082 0.00571038 0.00016332
NPL 0.08959344 0.04890394 0.81748208 0.02099232 0.02302822
EQA 0.01382585 0.01028138 0.00894282 0.96580604 0.00114392
ROA 0.01196264 0.01134668 0.01221219 0.27728652 0.68719197
Panel B: Conventional banks
GDPgr 0.89338834 0.0212611 0.01096759 0.04119014 0.03319282
INF 0.01384187 0.96760735 0.00060379 0.01594962 0.00199737
NPL 0.13889893 0.04240842 0.7542059 0.01621374 0.04827301
EQA 0.13036044 0.01062431 0.04177583 0.80736541 0.00987402
ROA 0.01487396 0.01816662 0.01759027 0.1638489 0.78552026
Panel C: Islamic banks
GDPgr 0.97645345 0.00586371 0.01682883 0.00014237 0.00071164
INF 0.0157761 0.97935801 0.0024296 0.00241846 0.00001783
NPL 0.00746981 0.06346006 0.92196855 0.00545549 0.00164609
EQA 0.00945549 0.01714798 0.07885379 0.88867307 0.00586966
ROA 0.06512239 0.01636681 0.0266093 0.30961054 0.58229096
7. Conclusion
The purpose of this paper is to empirically analyze the determinants of NPLs; the
managerial behavior, the bank-level factors along with macro and institutional environment
impact on Islamic bank credit risk exposure. We conduct a comparative analysis of Islamic
and conventional banks using a panel dataset of 117 banks located in MENA and South-
East Asian countries over the period 2005-2012. To explain behavior‟s differences of the
two type banks up to credit risk exposure we used two complementary methods: one-step
GMM analysis as well as panel vector autoregressive framework. Results from one-step
GMM indicate that not only bank-level factors affect credit risk but macro-economic and
institutional factors also matter. We find that NPLs rise with higher provisions, grater
capitalization and almost higher quality management. However, toxic loans decline with
larger credit growth, bigger bank size, improved profitability and mainly inside a growing
economy (i.e. higher GDP growth rate). Results also reveal that Islamic banks behave
differently to credit risk dilemma. In fact, interest-free banks differ from interest-based ones
because they shares profit with the investment account holder (under the profit and loss
30
sharing-PLS principle). The bank is, therefore, not liable for losses but investment
depositors bear part of bank credit risk. As a consequence, it is generally argued that
Islamic banks face lower credit risk levels. However, due to insufficient credit risk Shariah
compliant management tools, these banks tend to often mimic practices of interest-based
banks and therefore they tend to manage this risk in the same manner as it is managed by
these latter.
The panel VAR survey demonstrates that Islamic and conventional banks hold different
responses to various shocks. As for the investigation of the relationship efficiency-risk, the
results support the “bad management” hypothesis for conventional banks. This hypothesis
suggests that an increase in non-performing loans is preceded by a decrease in cost
efficiency. Moreover, results support the moral hazard and skimping hypotheses for both
banks‟ type. Regarding the macro-financial linkages assessment, on the whole, results
depicted that a positive shock to GDP growth, capitalization and profitability lead to an
improvement in loan portfolio quality which in turn result in lower credit risk. On the other
hand, higher inflation rate lead to the deterioration of the loan portfolio quality in the long
run.
Noteworthy, this study provides a significant contribution as its findings can give
policymakers, regulators and bank management bodies‟ better insight into the efficiency of
Islamic banking and its behavior toward credit risk. In fact, Islamic banking industry should
be discovered separately to the conventional industry; there are different entities that
behave differently and that need a specific regulatory environment to each of them. In
particular, it is necessary to cheek risks specific to each contract (i.e. Mudharabah,
Musharakah, Murabaha, Istisna‟a…) separately in order to develop and construct particular
and innovative risk management tools.
In order to extend the literature on Islamic banking system and exactly on non-
performing loans, we sought to explore the risk sensitivity of these banks by estimating the
shadow price and the cost of bad loans under different risk vectors. In addition, we plan to
incorporate corporate governance and the regulatory framework in our future research.
These lines of research are, in our knowledge, not yet explored in the context of Islamic
banks.
31
References
Abbas, A., Zaidi, S.A.H., Ahmad, W., Ashraf, R.U., 2014. Credit risk exposure and Performance of
Banking Sector of Pakistan. J. Basic. Appl. Sci. Res. 4 (3), 240-245.
Ahmad, N., Ariff, M., 2007. Multi-country study of bank credit risk determinants. Int. J. Bank.
Finance. 5, (1), 135-152.
Al-Wesabi, H.A.H., Ahmad, N.H., 2013. Credit risk of Islamic banks in GCC countries. Int. J.
Bank. Finance. 10 (2), 120-136.
Ananzah, E., Othman, M., 2010. Test The Efficiency of Credit RiskManagement on The Quality
of Credit Portfolios in The Jordanian Islamic Banks. Paper Presented At The Seventh
International Scientific Conference Of The Philadelphian Private University.
Arellano, M. and Bover, O. (1995) „Another look at the instrumental variable estimation of error
component models‟, J. Econom, Vol.68, 29-51.
Badar, M., Javid, A.Y., 2013. Impact of macroeconomic forces on nonperforming loans: An
empirical study of commercial banks in Pakistan. WSEAS. Tran. Bus. Econ. 10 (1), 40-48.
Beck, R., Jakubik, P., Piliou, A., 2013a. Non-Performing Loans: What matters in addition to the
economic cycle? European Central Bank Working paper 1515.
Beck, T., Demirguk-Kunt, A., Merrouche, O., 2013b. Islamic vs. Conventional banking: Business
model, efficiency and Stability. J. Bank. Finance. 37, 433-477.
Berger, A.N., DeYoung, R., 1997. Problem loans and cost efficiency in Commercial banks. J. Bank.
Finance. 21, 849-870.
Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel data
models. J. Econom. 87, 115-143.
Boudriga, A., Boulila Taktak, N., Jellouli, S., 2009. Banking supervision and nonperforming loans:
A cross country analysis. J. Financ. Econ. Polic. 1 (4), 286-318.
Boudriga, A., Boulila Taktak, N., Jellouli, S., 2010. Bank specific, Business and institutional
environment determinants of banks nonperforming loans: Evidence from MENA countries.
Economic research forum, working paper 547.
Boumedienne, A., 2011. Is credit risk really higher in Islamic banks? J. Credit. Risk, 7 (3), 97-129.
Castro, V., 2013. Macroeconomic determinants of the credit risk in the banking system: The case of
GIPSI. Economic Modelling. 31, 672-683.
Causse, G., 2012. Islamic finance: An alternative finance or an antidote to the crisis of capitalism?‟
Int. Symp. Econ. Theory. Econom. 22, 173-196.
32
Chaibi, H., Ftiti, Z., 2015. Credit risk determinants: Evidence from a cross-country study. Res. Int.
Bus. Finance. 33, 1-16.
Chong, B.S., Liu, M.H., 2009. Islamic banking: interest-free or interest-based? Pac. Basin. Finance.
J. 17, 125-144.
Das, A., Ghosh, S., 2007. Determinants of credit risk in Indian State owned Banks: An Empirical
Investigation. Econ. Issues. 12 (2), 27-46.
El-Gamal, M.A., 2006. Islamic finance: Law, Economics and practice. Cambridge University Press,
Cambridge, New York.
Epinoza, R., Prasad, A., 2010. Nonperforming loans in the GCC banking system and their
macroeconomic effects. IMF working paper 224.
Erico, L., Sundararajan, V., 2002. Islamic financial institutions and products in the global financial
system: key issues in risk management and challenges ahead. International Monetary Fund.
Farhan, M., Sattar, A., Chaudhry, A.H., Khalil, F., 2012. Economic determinants of nonperforming
loans: Perception of Pakistani bankers. Eur. J. Bus. Manag. 4 (19), 87-99.
Festic, M., Beko, J., 2008. The banking sector and macroeconomic indicators: Some evidence for
Hungary and Poland. Our Economy, Professional paper, N.5-6, pp.118-125.
Fiordelisi, F., Marques-Ibanez, D., Molyneux, P., 2011. Efficiency and risk in European banking. J.
Bank. Fin. 35 (5), 1315-1326.
Goldlewski, C.J., 2004. Bank capital and Credit risk taking in Emerging Market economies. J.
Bank. Fin. 6 (2), 128-145.
Hasan, M., Dridi, J., 2010. The effect of the global crisis on Islamic and conventional banks: A
comparative study‟. International Monetary Fund.
Hasan, I., Wall, L.D., 2004. Determinants of the loan loss allowance: Some cross-country
comparisons. Financ. Rev. 39 (1), 129-152.
Jakubik, P., Reininger, T., 2013. Determinants of nonperforming loans in Central, Eastern and
Southeastern Europe. Focus. Eur. Econ. Integr. Q3, 48-66.
Jimenez, G., Saurina, J., 2006. Credit cycles, credit risk and prudential regulation. Int. J. Cent.
Bank. 2 (2), 65-98.
Kaufman, D., Kraay, A., Mastruzzi, M., 2008. Governance Matters VII: Aggregate and Individual
Governance indicators for 1996-2005. World Bank Policy Research Working Paper 4654.
Khan, F., 2010. How Islamic is Islamic banking? J. Econ. Behav. Organ. 76, 805-820.
33
Khan, T., Ahmed, H., 2001. Risk management: an analysis of issues in Islamic Financial Industry.
Islamic Development Bank-Islamic Research and Training Institute, Occasional Paper 5, Jeddah.
Khemraj, T., Pasha, S., 2009. The determinants of Non-performing loans: An Econometric case
study of Guyana. Presented at the Caribbean Centre for Banking and Finance Bi-annual
conference on Banking and Finance, St Augustine, Trinidad.
Klein, N., 2013. Non-Performing loans in CESEE: Determinants and impact on Macroeconomic
performance. IMF working paper 72.
Koutsomanoli-Filippaki, A., Mamatzakis, E., 2009. Performance and Merton-type default risk of
listed banks in the EU: a panel VAR approach. J. Bank. Finance. 33 (11), 2050-2061.
Lee, C.C., Hsieh, M.F., 2013. The impact of bank capital on profitability and risk in Asian banking.
J. Int. Money. Finance. 32, 251-281.
Louzis, D.P., Vouldis, A.T., Metaxas, V.L., 2012. Macroeconomic and bank-specific determinants
of non-performing loans in Greece: A comparative study of mortgage, business and consumer
loan portfolios. J. Bank. Finance. 36, 1012-1027.
Love, I., Turk-Ariss, R., 2014. Macro-financial linkages in Egypt: A panel analysis of economic
shocks and loan portfolio quality. J. Int. Finance. Markets. Inst. Money. 28, 158-181.
Love, I., Zicchino, L., 2006. Financial development and dynamic investment behaviour: evidence
from panel vector autoregression. Q. Rev. Econ. Finance. 46, 190-210.
Makri, V., Tsagkanos, A., Bellas, A., 2014. Determinants of Non-Performing loans: the case of
Euro-zone. Panoeconomicus. 2, 193-206.
Mamatzakis, E., Matousek, R., Vu, A.N., 2015. What is the impact of bankrupt and restructured
loans on Japenese bank efficiency? J. Bank. Finance. doi: 10.1016/j.jbankfin.2015.04.010.
Messai, A.S., Jouini, F., 2013. Micro and Macro determinants of non-performing loans. Int. J. Econ.
Financ. Issues. 3 (4), 852-860.
Monokroussous, P., Thomakos, D., 2014. What explain the rise in non-performing loans (NPLs) in
the euro area following the outbreak of the global financial crisis? Greece case study. Eurobank
Global Markets Research: Greece macro monitor. February 14.
Nkusu, M., 2011. Nonperforming loans and macro-financial vulnerabilities in advanced economies.
IMF working paper 161.
Nurul-Kabir, M., Worthington, A., Gupta, R., 2015. Comparative credit risk in Islamic and
convetional bank. Pac. Basin. Finance. 34, 327-353.
Reinhart, C., Rogoff, K., 2010. From financial crash to debt crisis. NBER Working paper 15795.
34
Saeed, M. Izzeldin, M., 2014. Examining the relationship between default risk and efficiency in
Islamic and conventional banks. J. Econ. Behave. Organ.
http://dx.doi.org/10.1016/j.jebo.2014.02.014
Salas, V., Saurina, J., 2002. Credit risk in two institutional regimes: Spanish commercial and
savings banks. J. Financ. Serv. Res. 22, 203-224.
Sandararajan, V., Errico, L., 2002. Islamic financial institutions and products in the global financial
system: key issues in risk management and challenges ahead. International Monetary Fund,
Working paper 02/192.
Setiawan, C., Hassan, T., Hassan, M. K., 2013. Nonperforming loans and bank efficiency of
conventional and Islamic banks in the Organization of Islamic Cooperation (OIC) countries.
Presented at the 9th International conference on Islamic economics and finance, Istanbul,
Turkey.
Sharma-Poudel, R.P., 2013. Macroeconomic determinants of credit risk in Nepalese banking

industry. Proceedings of 21st international business research conference 10-11 June, Ryerson
University, Toronto, Canada, ISBN: 978-1-922069-25-2.
Shingjergji, A., 2013. The impact of bank specific variables on the non performing loans ratio in the
Albanian banking system. Res. J. Finance. Account. 4 (7), 148-152.
Skarica, B., 2014. Determinants of nonperforming loans in Central and Eastern European Countries.
Financ. Theory. Pract. 38 (1), 37-59.
Tan, Y., 2015. The impact of risk and competition on bank profitability in China. J. Int. Finance.
Markets. Inst. Money. doi: 10.1016/j.intfin.2015.09.003
Tehulu, T.A., Olana, D.R., 2014. Bank-specific determinants of credit risk: Empirical evidence
from Ethiopian banks. Res. J. Finance. Account. 5 (7), 80-85.
Zribi, N., Boujelbene, Y., 2011. The factors influencing bank credit risk: the case of Tunisia. J.
Account . Tax. 3 (4), 70-78.
35
Appendices
Apeendix.1: Cost efficiency Estimation
To estimate cost efficiency, we opt for the stochastic frontier approach (SFA) proposed by
Battese and Coelli (1995).This approach uses a parametric technique to estimate the
characteristics of a best-practice bank from the cost function. A stochastic cost frontier is
estimated in the following way:
( ) ( ) and [1]
Where ln ( ) denotes observed total cost for bank i at year t for country j, Y is a vector of
input prices; X is a vector of output prices and is a vector of unknown parameters to be
estimated. and are the two components of the error term ; is the stochastic
error that captures the effect of noise and measurement error and constitute the cost
inefficiency term and its is composed by a vector of exogenous (Z) variables and a random
error (e).
The cost function in [1] is estimated via a maximum likelihood procedure where the
inefficiency component is assumed to follow a truncated normal distribution, i.e.
. To empirically implement our cost frontier, we opt for the Translog
functional form as used by Saeed and Izzeldin (2014) and which is specified as follows:
∑ [∑ ∑ ∑∑ ]
∑∑
and are the parameters to be estimated.
The intermediation approach is applied as it is widely used in the relevant literature

examining banks‟ efficiency (Viverita et al., 2007; Srairi, 2010; Assaf et al., 2011). The
data used in order to estimate cost efficiency is presented in Table (4).
Table.A.1: Cost efficiency variables
36
Variable Definition Description
Dependent variable
C Total Cost Interest expenses+ Non-interest expenses
Outputs and inputs prices
Y1 Price of labor Personal expenses divided by total assets
Y2 Price of fund Interest expenses divided by total deposits
Y3 Price of physical capital Other administration expenses+ Other operating
expenses divided by fixed assets
P1 Net total loans Total loans net of provision
P2 Other earning assets Deposits and short term funds+ other interest bearing
liabilities+ other non interest bearing liabilities
37
Table A.2: Frontier estimation results of the cost function
Variables Coefficient p-Value

alfa01 0.8172823 (0.000)
alfa02 0.503817 (0.000)
alfa11 0.168895 (0.000)
alfa12 -0.1831145 (0.000)
alfa22 0.1719205 (0.000)
Beta01 1.149556 (0.000)
Beta02 -0.5165589 (0.000)
Beta11 0.0325057 (0.270)
Beta12 -0.0363162 (0.111)
Beta22 0.0346234 (0.091)
Gama11 -0.0134604 (0.245)
Gama12 0.0463416 (0.000)
Gama21 -0.0506823 (0.000)
Gama22 0.050244 (0.000)
Constant -0.4374137 (0.294)
Exogenous variables
Dt -0.2047115 (0.378)
DYear 2005 -1.095367 (0.320)
DYear 2006 -1.063366 (0.267)
DYear 2007 -0.7831004 (0.311)
DYear 2008 -0.5555719 (0.367)
DYear 2009 -0.4118799 (0.394)
DYear 2010 0.0118992 (0.967)
DYear 2011 0
DYear 2012 0
DUAE -10.25992 (0.043)
DKuwait -11.78126 (0.117)
DSaudi-arabia 6.30428 (0.325)
DBahrain -22.93548 (0.150)
DQatar -18.36918 (0.067)
DJordan 1.9371 (0.348)
DTunisia -12.77339 (0.690)
DTurkey -6.706257 (0.014)
DBangladesh -9.35221 (0.223)
DIndonesia 0
DlnZ1 -0.0489752 (0.068)
DZ2 -2.486485 (0.279)
DlnZ3 -0.6593628 (0.007)
DZ4 -7.697366 (0.064)
DZ5 1.551431 (0.210)
DlnZ6 4.471557 (0.050)
DZ7 -0.0468975 (0.991)
DlnZ8 6.734771 (0.140)
Constant -59.24838 (0.106)
Ln (σ2) -1.650741 (0.000)

σ2 0.1919077
Γ 0.9186854
σu2 0.1763028
σ v2 0.0156049
Note: The dummies highlighted in italics have been omitted because of colinearity
38
Appendix.2: Modes of Islamic banking financing
Table A.3: Islamic banking contracts
Contract Description
Mudharabah Mudharabah is an arrangement whereby a party or investor possessing a capital
(rabbul-mal) put forward funds to his/her partner (Mudharib) for trading
purposes. The benefits must be shared on a pre-agreed basis between both parties.
Musharakah Musharakah is a joint-venture arrangement. This kind of partnership means a

financial participation in which the bank participates together with one of its
clients in a commercial-industrial undertaking operation or bid. The losses are
born by all the partners according to their participations. As for profits, they
should be distributed among partners according to a pre-agreed ratio.
Murabahah In a murabahah contract, the financial institution buys the product on client‟s
request. It then sells it back to the client (for cash or in installments). Evidently,
the bank will sell the product at a higher price than that paid originally.
Ijarah Ijarah is a contract that relates to a specific benefit derived from the ownership of
a good or product for a known cost.
Salam Salam is a sale contract whereby the seller undertakes to supply certain goods to
the purchaser at deferred a future date, in exchange for an advanced price fully
paid on the spot. The Salam contract creates a moral obligation on the Salam
seller to deliver the goods. Once signed, this contract can never be revoked or
cancelled.
Istisna’ Istisna‟ is a form of sale through which a commodity is traded or transacted

before it comes into existence. It involves ordering a manufacturer to make or
provide a specific commodity for the transaction. Islamic banks enter into a back
to back Istisna‟ structure which actually includes two contracts. The first contract
is drawn between the Islamic bank and the manufacturer, whereby the bank acts
as a buyer of the manufactured assets. The second contract is established between
the Islamic bank and the customer through which the bank would act as the seller
of the manufactured asset.
In order to be compliant with Shariah rules, the contract must specify the features
of the goods to be produced as well as the delivery date.
39
*Graphical Abstract

J Ribaf 2016 03 014

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

J Ribaf 2016 03 014

Uploaded by

Copyright:

Available Formats

Accepted Manuscript

Title: Credit risk, managerial behaviour and macroeconomic

Author: Awatef Louhichi Younes Boujelbene

To appear in: Research in International Business and Finance

Received date: 2-4-2015

First author: Awatef Louhichi

Affiliation: Faculty of Economics and Management of Sfax, Tunisia

Second Author: Younes Boujelbene

Affiliation: Faculty of Economics and Management of Sfax, Tunisia

To achieve our purpose, we applied two complementary approaches namely one-step

Actually, this study is intended to help clarify certain misunderstanding relevant to

2. A comparative overview of credit risk in Islamic and conventional banking

It is worth noting that Islamic banking institutions have experienced an unprecedented

Contract Associated Credit risk

In a broader study spanning 26 advanced economies and covering period of 1998-2009,

4. Data and variables description

Table .2: Banks by country and region

Region Country Islamic banks Conventional banks All banks

Bank capitalization is approximated by the equity to assets ratio (EQA). According to

Variables Notation Description Sources

We employ two complementary methods to assess credit risk determinants. First, we

5.2.Panel vector auto-regression model

6. Empirical results and Interpretations

Table 4: Descriptive statistics of variables’ model by bank’s business model (average

All sample banks Conventional banks Islamic banks

Obs. Mean S.D. Obs. Mean S.D. Obs. Mean S.D.

Islamic banks Conventional banks

Fig.1. NPLs ratio by years and business model

Year All sample banks Islamic banks Conventional banks

2006 0.89 0.82 0.96

2007 0.88 0.86 0.93

2008 0.87 0.84 0.94

2009 0.86 0.84 0.94

2010 0.84 0.82 0.93

2011 0.84 0.85 0.94

2012 0.84 0.86 0.92

Mean 0.86 0.84 0.94

Islamic banks Conventional banks

Fig.2. Cost efficiency by region and business model

6.1.Dynamic panel estimation results:

All sample banks Conventional banks Islamic banks

*Denote significance at 10% level

Profitability as approximated by ROA reveals a negative and statistically significant

Banks capitalization is positively related to credit risk solely within conventional

Regarding the macroeconomic factors, GDPgr is found to be highly significant and

6.2.Panel VAR analysis

6.2.1. Risk- efficiency issue

The response of interest-based banks to credit chocks is expected to be different from

Table 7: Main results of a 2-variables VAR model

* Denote significance at 10% level

All sample banks Conventional banks Islamic banks

-0.0003 0.0000 -0.0001 -0.0013 -0.0024 -0.0154

-0.0023 0.0000 -0.0051 0.0000 -0.0076 -0.0006

i. All sample banks

ii. Interest-Based relative to Interest-Free banks

All sample banks Conventional banks Islamic banks

We investigate the macroeconomic and dynamic consequences of an increase in NPLs

Panel A: All sample banks

*Denote significance at 10% level

We proceed now to the analysis of mutual interactions between individual variables.

i. All sample banks:

Second, we analyses various interactions among bank-level variables. Results show

Additionally, in table (10) (panel A) we present variance decompositions results to