Вы находитесь на странице: 1из 8

Economic Modelling 55 (2016) 127–134

Contents lists available at ScienceDirect

Economic Modelling

journal homepage: www.elsevier.com/locate/ecmod

Business cycle and bank lending procyclicality in a dual banking system


Mansor H. Ibrahim
School of Graduate Studies, International Centre for Education in Islamic Finance (INCEIF), Lorong Universiti A, 59100 Kuala Lumpur, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: The paper studies bank lending behaviour over the business cycle in a dual banking system, Malaysia, with
Accepted 19 January 2016 the objective of ascertaining whether Islamic banks have a role in stabilizing credit. The study makes use of
Available online xxxx unbalanced panel data of 21 conventional banks and 16 Islamic banks covering mostly the period 2001–2013.
Applying dynamic GMM estimators, we find the aggregate loans by banks to be pro-cyclical in conformity with
Keywords:
existing studies. However, when we segregate the lending/financing behaviour of conventional and Islamic
Dual banking system
Lending decisions
banks, the cyclicality of bank lending seems to be true only for conventional banks. As for the Islamic banks,
Business cycle the business cycle does not seem to affect their financing decisions. Indeed, there is indication that the Islamic
Malaysia banks in general and the full-fledged Islamic banks in particular can even be counter-cyclical in their financing
decisions. This conclusion is fairly robust to a different loan measure, alternative model specifications, and to
an alternative business cycle measure. Hence, our results provide further support to the “stability” view of the
Islamic banks in that they have the ability to stabilize credit.
© 2016 Elsevier B.V. All rights reserved.

1. Introduction risk-taking and the fundamental links between Islamic banks' assets
and liabilities and real activities. This “stability” view of the Islamic
The eruption of the global financial crisis and its real consequences, banking system has received empirical supports from several recent
the magnitude of which is considered to be the worst since the great studies, notably Cihak and Hesse (2010), Hasan and Dridi (2010), Beck
depression, have placed financial intermediaries under microscopic et al. (2013), and most recently Belanes et al. (2015) and Farooq and
lens of both policymakers and economists. Acknowledging that disrup- Zaheer (2015). Evaluating the Z-score of 77 Islamic banks and 397 con-
tions in the banking system have substantial economic costs and banks ventional banks, Cihak and Hesse (2010) find small Islamic banks to be
are affected differently by the crisis, various studies have attempted to more stable than small conventional banks and large Islamic banks.
identify bank-specific characteristics that make them vulnerable or Hasan and Dridi (2010) also find Islamic banks to be more resilient at
resilient using a spectrum of bank performance measures such as the early stage of the global financial crisis in their examination of 120
profitability, non-performing loans and bank loans. As regards with Islamic and conventional banks in 8 countries. In Beck et al. (2013),
bank lending behaviour, the global financial crisis has served as a test the Islamic banks are further noted to have better asset quality, higher
case for linking bank lending procyclicality to bank-specific characteris- intermediation ratio and are better capitalized. Belanes et al. (2015)
tics such as bank capital and bank ownership. The main inquiry is: assess the efficiency performance of 30 Islamic banks from the GCC
which types of banks (e.g. private versus public banks, and domestic region during the period 2005–2011. During the crisis period, these
versus foreign) tend to exhibit relative stability in their lending behav- Islamic banks remain largely efficient with several Islamic banks to be
iour during the recent cyclical downturns? only marginally affected. Finally, looking at bank deposit and lending
The Islamic banking sector is no exception to this scrutiny, driven behaviour in Pakistan, Farooq and Zaheer (2015) demonstrate less
mainly by its significant growth since the turn of the 21st century susceptibility of Islamic bank branches of mixed banks to deposit
especially in the Middle East and Southeast Asia and its increasing withdrawals during September–October 2008 financial panic. Further,
acceptance in non-Muslim world. The view prevailing within the circle the Islamic banks tend to grant more new loans and their lending
of Islamic banking professionals is that the Islamic banking system is decision is less sensitive to variations in deposits during the period.
stable and resilient to financial and economic disruptions on the basis These findings supporting the “stability” view of the Islamic banking
that it is different. As elaborated by Farooq and Zaheer (2015), the rela- system notwithstanding, the stability and resiliency of the Islamic
tive stability of Islamic banks is rooted in their distinct features, namely, banking system remain a contentious issue. Indeed, the arguments
the prohibition of interest rate, speculative activities and of excessive and evidence for no differential performance or even less stability of
Islamic banks are not hard to find. First, while the theoretical model of
the Islamic banking business differs from the conventional banking
E-mail addresses: mansorhi@inceif.org, mansorhi@hotmail.com. model, Islamic banks in operations today are still far away from an

http://dx.doi.org/10.1016/j.econmod.2016.01.013
0264-9993/© 2016 Elsevier B.V. All rights reserved.
128 M.H. Ibrahim / Economic Modelling 55 (2016) 127–134

ideal Islamic banking model. In practice, they are not distinguishable Muslim countries with a dual-banking system. Hence, with our interest
from the conventional banks since the profit-and-loss sharing (PLS)- to evaluate comparative pro-cyclicality of conventional and Islamic
based assets highlighted as a distinct feature of Islamic banks make up banks, Malaysia seems to be a natural choice.
only a small fraction of the Islamic banks' assets (Chong and Liu, 2009; And third, in line with Bertay et al. (2015) and a much earlier study
Khan, 2010). Abdul-Rahman et al. (2014) and Azmat et al. (2015) by Micco and Panizza (2006), we focus on bank lending procyclicality
further question the ability of Islamic banks to heighten PLS-based over the business cycle instead of the lending behaviour during the
business activities under the present setting. Second, it is well acknowl- crisis. The relative resiliency of Islamic banking model during the recent
edged that financing by banks depends critically on funding sources and global financial crisis documented by various studies may not be related
asset compositions. Thus, Islamic banks may be more affected by to its Islamic structure or business model. Instead, it may be due to the
adverse shocks since the Islamic money markets are less developed facts that Islamic banks are relatively small in scale and their business
and hence limited access by the Islamic banks to alternative sources of activities are more domestically oriented and hence are less exposed
funds other than deposits (Farooq and Zaheer, 2015). In addition, the to cross-border or international transactions. We believe that a more
assets of Islamic banks tend to be more concentrated and their hedging meaningful assessment of the Islamic banks is to look at how they adjust
instruments are limited due to restrictions by Islamic laws (Beck et al., their lending decisions over cyclical phases of real activities. From the
2013), features that add to potential instability in the face of adverse aforementioned studies, whether Islamic banks are more or less pro-
shocks. And finally, several studies have contested the above “stability” cyclical remains an open issue. The finding by Hasan and Dridi (2010)
findings by noting insignificant difference between the two banking that Islamic banks were more adversely affected once the financial crisis
systems; see, for instance, Bourkhis and Nabi (2013). The higher- brought negative repercussion on the real sector suggests the pro-
intermediation costs (Beck et al., 2013) and larger capital buffers cyclical behaviour of Islamic bank financing to be more intense than
(Daher et al., 2015) maintained by Islamic banks also raise doubt that of the conventional bank lending. By contrast, the finding by
whether they can remain competitive in an increasingly integrated Farooq and Zaheer (2015) that Islamic banks are less prone to liquidity
global financial market. The findings by Hasan and Dridi (2010) that shocks or to variations in deposits hints on potential stabilizing effect of
the profitability of Islamic banks is more adversely affected once the Islamic banks' financing decision. Thus, there is a need to look further at
financial crisis is filtered through to the real sector and by Cihak and these apparently contradicting findings.
Hesse (2010) that large Islamic banks tend to be less stable than large The rest of the paper is structured as follows. As a precursor to the
conventional banks add further to scepticism to the “stability” view. analysis, the next section provides brief background information on
In this paper, we empirically examine bank lending behaviour over the Islamic banking model and Islamic banking development in
the business cycle in a dual banking system, i.e. Malaysia, with the Malaysia. Then, Section 3 describes empirical models and data.
objective of ascertaining whether the Islamic banks have a role in stabi- Section 4 tabulates and discusses estimation results. The last section,
lizing credit. We contribute to the literature in three respects. First, we Section 5, concludes with a summary of the main findings and their
add to the strand of research emphasizing heterogeneous responses of implications.
banks with different banks' characteristics during economic downturns.
More particularly, our study is related to several recent studies that have 2. Background of Islamic banking
assessed the importance of ownership in explaining lending stability
during the crisis periods (Brei and Schclarek, 2013; Cull and Martinez The beginning of the modern Islamic banking can be dated back to
Peria, 2013; Fungacova et al., 2013; Albertazzi and Bottero, 2014; Ferri the establishment of the Mit-Ghamr Islamic Saving Associations
et al., 2014; Bertay et al., 2015). We add to this literature by bringing (MGISA) in Egypt in 1963 to mobilize savings of Muslims and provide
into the fore the implication of the Islamic banking system on lending returns in accordance with the Islamic principles. While the bank
procyclicality. Consequentially, we also add to the on-going research came to an end in 1967 due to political reasons, its success in attracting
on the relative stability of the Islamic finance system which, as elaborat- savings is beyond doubt (Chachi, 2005). Since then, several Muslim
ed above, has received mixed supports. Undoubtedly, the Islamic majority countries took a step to introduce the Islamic-based banking
banking system will continue to play increasing intermediation role system with a main aim to meet the religious obligations of the
and have a significant force in the global financial scene. Hence, further Muslims. These include the establishments of The Dubai Islamic Bank
empirical verification of the issue is much needed such that the argu- in the United Arab Emirates, Faisal Islamic Bank in Sudan, and Kuwait
ments in favour of the Islamic banking system as a viable alternative Finance House in Kuwait in the 1970s. In Malaysia, saving mobilization
are placed on a concrete ground. of Muslims begins with the establishment of the Pilgrims Fund Corpora-
Second, many existing studies rely on bank-level panel data assem- tion, an institution founded in 1963 to provide saving and investment
bled from many countries. In the standard panel-data approaches, avenues for Muslims with an exclusive aim to support their expenditure
omitted country-specific heterogeneities are normally accounted by during the Pilgrimage period. A decade later, the first Islamic bank
the inclusion of the so-called country-specific effects. However, the in Malaysia was set up. Starting with humble beginning, the Islamic
approach may be inadequate when applied to countries with vast banking system has now been accepted by many Muslim countries as
differences in economic structure, political environment, culture, and well as by several non-Muslim majority countries.
regulations (see also Huang et al., 2015). In the case of the Islamic While playing a similar role of financial intermediation, Islamic
banking, even some progresses have been made on harmonization of banks differ from their conventional counterparts in that their opera-
laws and regulations, interpretation of the shari'ah also differs across tions are governed by Islamic laws and principles. More specifically,
countries. To mitigate this difficulty, the present analysis employs as elaborated by Hussain et al. (2015), Islamic finance is guided by
bank-level panel data of only a single country, i.e. Malaysia. The choice the principles of equity, participation and ownership. Based on these
of Malaysia is straightforward. By June 2013, Malaysia was ranked principles, Islamic finance are free from predetermined payments
third in the share of Islamic banking assets (% of the global Islamic (riba) above the principal amount of loans, deposits and other financial
banking assets) behind Iran and Saudi Arabia (Hussain et al., 2015). transactions. The agreement among contracting parties must also be
Supported by well-planned government's strategies and comprehen- free from excessive uncertainty or gharar. In addition, the contractual
sive regulatory framework, Malaysia is viewed to have one of the arrangements of Islamic banking transactions are tied fundamentally
most developed Islamic banking markets and, together with its to productive activities through sale-based transactions (murabaha)
pioneering efforts in the development of Islamic capital markets, is or, ideally, profit-and-loss sharing agreements (mudaraba and
considered as a leading centre of Islamic finance. Financial market musyaraka). The financing of Islamic banks must also be confined to
development-wise, Malaysia is more advanced as compared to other activities allowable by Islam. Accordingly, activities involving the
M.H. Ibrahim / Economic Modelling 55 (2016) 127–134 129

production, consumption or sales of tobacco, alcohol, pork products, procyclicality of bank lending, it is important to control for the sources
pornography materials, gambling and the like are strictly prohibited. of funding given the earlier stated argument that Islamic banks have
In Malaysia, the development of the Islamic banking sector has been less access to alternative sources of funding. In addition to the real
most progressive. Starting with a single Islamic bank in 1983, i.e. Bank GDP growth, we also include the inflation rate to capture macroeco-
Islam Malaysia Berhad (BIMB), the Islamic banking sector in Malaysia nomic conditions of the country.
now comprises 16 Islamic banks. BIMB was the sole provider of the In Eq. (1), real GDP growth is the key variable. Its coefficient captures
Islamic financial services from its inception year to 1993. In 1993, the the adjustments of bank loans to macroeconomic shocks or cycles
government of Malaysia introduced the “Islamic Banking Scheme” to (Micco and Panizza, 2006; Bertay et al., 2015). The finding that it is
allow conventional banks to participate in the provisions of Islamic positive and significant would suggest bank lending pro-cyclicality.
financial services through the so-called Islamic windows and conse- Meanwhile, if it turns out to be negative, then bank lending is taken to
quentially increase the number of players in the market. Right after be counter-cyclical. Given our interest to determine whether Islamic
the 1997/1999 Asian financial crisis, the second Islamic bank, known bank financing is more or less pro-cyclical, we interact the Islamic
as Bank Muamalat Malaysia Berhad, came into existence in 1999. bank dummy with real GDP growth and add to the equation as:
Indeed, since the Asian financial crisis, the Islamic banking sector has
become more competitive by the establishment of Islamic bank subsid- ΔLit ¼ γΔLit−1 þ ðβ0 þ β1 IBi Þ  Δyt þ θX it−1 þ πin f t þ α i þ εit ð2Þ
iaries to replace Islamic windows and by the setup of foreign Islamic
banks. At present, Malaysia has a total of 2 full-fledged domestic Islamic where IB equals 1 if bank i is an Islamic bank and 0 otherwise. We treat
banks, 3 full-fledged foreign Islamic banks and 11 Islamic bank full-fledged Islamic banks and Islamic bank subsidiaries of conventional
subsidiaries. banks as one as well as separately in our analysis. In Eq. (2), β0 repre-
The growth of the Islamic banking sector in Malaysia has been sents the lending cyclicality of conventional banks while β0 + β1 the fi-
impressive especially after the Asian financial crisis. Over the years, nancing cyclicality of Islamic banks. If Islamic banks are less pro-cyclical
the Islamic banking sector has played increasingly significant role in and hence are able to smooth credit over the business cycle, we expect
mobilizing savings and meeting financial needs of the private sector. that β1 b 0.
In year 2000, the ratio of Islamic financing to total banking loans and Statistically, the traditional panel estimators of Eqs. (1) and (2) are
advances was slightly above 5%. The corresponding ratio for Islamic problematic due to non-zero correlation between the individual-fixed
deposit-aggregate deposit was below 7%. At the end of 2014, these effect and lagged dependent variable. Hence, we apply the system gen-
ratios surpassed 20%. At present, the share of the Islamic banking assets eralized method of moments (GMM) estimator proposed by Arellano
in the total banking assets is well above 20%. Encouraged by this rapid and Bover (1995), and Blundell and Bond (1998), which is viewed to
growth and by the aspiration to be the global hub of Islamic finance, be superior in handling dynamic panel modelling. The first-difference
Malaysia now sets for a more ambitious target to increase Islamic GMM estimator by Arellano and Bond (1991) involves first differencing
financing's share of total financing to 40% by 2020. This means that to wipe out the individual-specific effect and employing lagged-level
the Islamic banking system will play an even more important role in variables as instruments to address the correlation between the error
the future. terms and the explanatory variables. Arellano and Bover (1995) and
Despite much discussion on the ability of the Islamic banking system Blundell and Bond (1998) further suggest using both level and first-
in injecting stability into the financial system, only recently that it has differenced regressions as a system such that potential information in
been subject to scholarly inquiry. We contribute further to this line of the level relationship is not filtered out and the potential biases and im-
inquiry by investigating the relative procyclicality of Islamic financing precision associated with the first-difference GMM due to weak instru-
in a dual banking system. ments are reduced. This is known as the system GMM. In the system
GMM, the level regression is instrumented with lagged first-
3. Empirical models differenced variables while the first-differenced regression is instru-
mented with the lagged level variables. We opt for the two-step system
The empirical specification is designed to assess bank lending GMM estimators and use the Windmeijer (2005) correction to mini-
procyclicality and verify whether Islamic banks are less or more pro- mize the downward bias of the two-step standard errors. Still, the
cyclical. Taking leads from Micco and Panizza (2006) and Bertay et al. two-step first-difference GMM results are also reported for comparison.
(2015), we first specify the basic bank lending equation as: In addition, a variety of robustness checks, the detail of which to be ex-
plicated later in the next section, are also conducted.
ΔLit ¼ γΔLit−1 þ βΔyt þ θX it−1 þ πin f t þ α i þ εit ð1Þ We gather Malaysia's bank-level data from Bureau van Dijk's
Bankscope. In arriving at our final data sample, we only include banks
where Δ is the first difference operator, L is the natural logarithm of CPI- that have data over five consecutive years and the years must include
deflated gross loan, y is the natural logarithm of real GDP, X is a vector of 2008 and 2009. This is to ensure that the recent cyclical downturn expe-
bank-specific variables, inf is the inflation rate, αi is bank-specific effect, rienced by Malaysia is included in the sample. We also place the re-
and ε is the common error term. In the above specification, we include quirement that the banks must be in operations for the full year. This
the lagged-dependent variable to capture potential dynamics in real means that the years that banks began operations or ceased to exist
loan growth due to for example loan growth targets set by banks. We due to mergers mid-year are excluded. Finally, we also exclude bank-
consider the following bank-specific variables: the natural logarithm year observations of the merged banks. With these criteria, our data
of real assets, equity to asset ratio, and deposit to total liabilities ratio. are unbalanced comprising 21 conventional and 16 Islamic banks (5
They are included with lagged one to alleviate the endogeneity problem. full-fledged Islamic banks and 11 Islamic bank subsidiaries) and cover-
We acknowledge that, in the bank lending literature, many bank- ing the period 2001 to 2013. The classification of banks to conventional
specific variables have been considered. However, we include only and Islamic and to domestic and foreign is from Malaysia's central bank,
these three variables, which are deemed most relevant, to keep the Bank Negara Malaysia. Annual data for the macro variables are from the
specification parsimonious and to reduce potential multicollinearity World Development Indicators.
issue from including numerous bank-specific variables. Real assets and Table 1 provides descriptive statistics of bank loans and bank-
equity to asset ratio are the most common bank-specific variables to specific variables, i.e. the CPI-deflated total assets expressed in natural
control for bank size and bank capitalization. The inclusion of bank logarithm, banks' equity to total asset ratio, and deposit to liabilities
deposit to total liability is to control for the funding ratio or Malaysian ratio. The following facts regarding the Malaysian banking industry
banks' dependence on deposits. With our objective to evaluate the are apparent. Conventional banks are on average substantially larger
130 M.H. Ibrahim / Economic Modelling 55 (2016) 127–134

Table 1
Descriptive Statistics.

Variables Full sample Conventional banks Islamic banks

Mean SD Mean SD Mean SD

Loan measures
Gross Loans 8349.81 12714.87 10946.53 15856.55 2940.00 3551.02
% growth 0.1187 0.2398 0.0795 0.2275 0.2105 0.2439
Net Loans 8087.46 13362.19 10586.39 15454.65 2851.61 3505.25
% growth 0.1229 0.2431 0.0847 0.2316 0.2123 0.2468
Bank-specific variables
Real assets (log) 8.5776 1.4510 8.7997 1.6174 8.1124 0.8519
Equity-asset ratio 0.1051 0.0683 0.1134 0.0746 0.0877 0.0488
Deposit-liabilities ratio 0.7321 0.1815 0.7068 0.1877 0.7852 0.1554

than Islamic banks, as reflected by the sizes of their real assets and loans. 4. Results
However, over the sample period, the growth rates of both loan
measures are far more rapid for the Islamic banks. While the growth 4.1. Basic results
rate of conventional bank lending was roughly 8% per year over 2001–
2013, the corresponding growth rate of Islamic financing was well Table 2 provides basic estimation results of models (1) and (2) using
over 20% per year over the same period. Interestingly, the conventional the two-step system GMM procedure. The odd regression numbers in-
banks are better capitalized with equity to asset ratio equalling to 11.3 clude only bank-specific variables as controlled variables. Meanwhile,
as compared to only 8.8 for the Islamic banks. As we have expected, the even regression numbers add the inflation rate as an additional
the funding ratio (i.e. deposit to liabilities ratio) is slightly higher for controlled variables. We treat all explanatory variables to be weakly
the Islamic banks. exogenous, which is sensible since lagged-one bank-specific variables
Fig. 1 depicts Malaysia's macroeconomic performance, i.e. real GDP are used and it is unlikely that financing shocks at bank-level have
growth and inflation rate. Over the period 2001–2013, Malaysia record- impacts on real GDP growth and the inflation rate. We limit the instru-
ed an average annual growth rate of 4.63%. Its low growth in 2001 and ments of the lagged explanatory variables such that the number of
again in 2009 is noticeable. At the turn of the 21st century, Malaysia instruments is reasonable given our cross-sectional sample size. Diag-
was affected by the slowdown in the global demand for the electronic nostic statistics reported at the bottom of the table indicate adequacy
and electrical appliances. Meanwhile, the global financial crisis and oil of the system GMM estimation. The autocorrelation tests of the first-
price escalation were the two key factors driving Malaysian economy differenced residuals suggest the presence of autocorrelation of order
into a negative real growth in 2009. While Malaysia recovered quickly 1 but fail to reject the null of no autocorrelation of order 2. This means
in 2010, its growth performance has slowed down again in recent the residuals in Eqs. (1) and (2) are free from the autocorrelation prob-
years. Malaysia's inflation record has been low averaging 2.23% per lem. In addition, the Sargan test statistics validate the instruments used
year over 2001–2013; but, it witnessed an upward trend from 2001 to in the estimation.
2008 before it dropped to 0.58% in 2009, the year with negative growth. In regressions (1) and (2), bank size and funding ratio are found to
From these macroeconomic facts, Malaysia was not immediately hit by be significant. By contrast, the coefficient of capitalization is indistin-
the global financial crisis but it was not insulated. As the crisis affected guishable from 0. These results prevail throughout other regressions re-
the real economy and became global, Malaysia's real performance ported in the table. In line with Bertay et al. (2015); Brei and Schclarek
turned recessionary. (2013), and Brei et al. (2013), bank size carries a negative coefficient.
Against these backdrops of Malaysia's dual banking system and Thus, smaller banks tend to have higher loan growth due possibly to
recent macroeconomic performance, we make an attempt to assess strong lending relationships between small banks and firms. Moreover,
whether bank lending accentuates cyclical swings of real activities by the larger banks may gear towards non-intermediation activities, which
examining pro-cyclicality of bank loans and to examine whether there limit their loan growth. The positive and significant coefficients of the
are differences between Islamic financing and conventional lending funding ratio or deposit-liabilities ratio further suggest the dependence
behaviour over the business cycle. of Malaysia's bank financing on deposits. A similar finding is also
documented by Bertay et al. (2015). While both Brei and Schclarek
(2013) and Brei et al. (2013) find capitalization to be significant,
Bertay et al. (2015) fail to document its significance in their regressions.
Turning to our key variable, we find evidence supporting the
cyclicality of bank lending in Malaysia, which is consistent with Micco
and Panizza (2006) and Bertay et al. (2015). The results from regres-
sions (1) and (2) suggest that a 1 percentage point increase in real
GDP growth is related to roughly 1.2 to 1.3 percentage point increase
in the growth of real gross loans. In regressions (3) and (4), we add
the Islamic bank interactive dummy and document significant
difference in the pro-cyclicality of conventional bank lending and
Islamic bank financing. Significant difference in bank lending behaviour
over the business cycle is also observed when we consider only the full-
fledged Islamic bank interactive dummy as indicated in regressions
(5) and (6). Interestingly, once we treat the full-fledged Islamic banks
and Islamic bank subsidiaries of conventional banks separately and
enter them in the estimation, i.e. regressions (7) and (8), we find only
the coefficient of the full-fledged Islamic bank interactive dummy to
Fig. 1. Real GDP Growth and Inflation over 2001–2013. be negative and statistically significant. Meanwhile, the interaction
M.H. Ibrahim / Economic Modelling 55 (2016) 127–134 131

Table 2
System GMM Estimation — Baseline Results.

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

ΔLit − 1 0.0620 0.0651 0.0532 0.0629 0.0516 0.0591 0.0512 0.0623


(0.388) (0.407) (0.424) (0.398) (0.458) (0.467) (0.421) (0.388)
Δyit 1.1705⁎⁎ 1.3068⁎ 1.7475⁎⁎⁎ 2.0815⁎⁎⁎ 1.5654⁎⁎⁎ 1.7892⁎⁎⁎ 1.8175⁎⁎⁎ 2.0473⁎⁎⁎
(0.020) (0.067) (0.006) (0.001) (0.000) (0.004) (0.001) (0.000)
Δyit × IBi – – −1.8467⁎ −2.3886⁎⁎ – – – –
(0.052) (0.033)
Δyit × IBFi – – – – −2.8069⁎ −2.9081⁎⁎ −2.6567⁎⁎ −3.1389⁎⁎
(0.078) (0.040) (0.040) (0.035)
Δyit × IBSi – – – – – – −1.6285 −1.5850
(0.379) (0.377)
Lassetit − 1 −0.1987⁎⁎⁎ −0.2094⁎⁎⁎ −0.2000⁎⁎ −0.1877⁎⁎⁎ −0.2035⁎⁎⁎ −0.2123⁎⁎⁎ −0.1975⁎⁎⁎ −0.1857⁎⁎⁎
(0.003) (0.001) (0.017) (0.004) (0.003) (0.003) (0.007) (0.002)
Capitalit − 1 −2.2404 −2.2934 −2.3465 −2.2593 −2.2114 −2.1904 −2.3164 −2.2000
(0.178) (0.147) (0.159) (0.210) (0.207) (0.258) (0.187) (0.186)
Depositit − 1 0.6537⁎⁎ 0.6448⁎⁎ 0.6494⁎⁎ 0.6244⁎⁎⁎ 0.6434⁎⁎ 0.6240⁎⁎ 0.6376⁎ 0.6054⁎⁎
(0.027) (0.012) (0.047) (0.008) (0.039) (0.036) (0.051) (0.011)
Inft – −0.0054 – −0.0057 – −0.0055 – −0.0056
(0.707) (0.710) (0.754) (0.687)
Constant 1.5140⁎⁎⁎ 1.6186⁎⁎⁎ 1.5351⁎⁎ 1.4426⁎⁎ 1.5538⁎⁎⁎ 1.6308⁎⁎⁎ 1.5183⁎⁎ 1.4340⁎⁎⁎
(0.006) (0.003) (0.022) (0.013) (0.007) (0.010) (0.011) (0.006)

P-values
AR(1)/AR(2) 0.043/0.721 0.043/0.715 0.041/0.586 0.042/0.522 0.041/0.619 0.043/0.606 0.040/0.556 0.039/0.502
Sargan 0.352 0.406 0.477 0.523 0.359 0.385 0.423 0.477

Note: numbers in parentheses are P-values.


⁎⁎⁎ Indicates significance at 1% level.
⁎⁎ Indicates significance at 5% level.
⁎ Indicates significance at 10% level.

between Islamic bank subsidiaries and economic growth is not signifi- full-fledged Islamic banks in particular can even be counter-cyclical in
cant. We repeat the estimation of the baseline models using the first their financing decisions.
difference GMM and provide the results in Table 3. All preceding results
prevail under the first-difference GMM estimator, although there are 4.2. Robustness checks
few exceptions.
Taken together, these results tend to be favourable to the “stability” To add credence to the aforementioned results supporting the
view of the Islamic banks in that they have the ability to stabilize credit. smoothing role of Islamic financing over the business cycle, we perform
Further, the results seem to be more supportive for the stabilizing role of a variety of robustness checks. The first robustness check employs net
the full-fledged Islamic banks. Indeed, based on the estimated coeffi- loan growth in place of gross loan growth. The results presented in
cients, there is an indication that the Islamic banks in general and the Table 4 using the system and first-difference GMM estimators further

Table 3
First-Difference GMM Estimation — Baseline Results.

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

ΔLit − 1 0.0594 0.0609 0.0597 0.0615 0.0552 0.0499 0.0588 0.0582


(0.418) (0.425) (0.361) (0.388) (0.398) (0.467) (0.350) (0.390)
Δyit 0.8667⁎ 0.9532⁎⁎ 1.6349⁎⁎⁎ 1.7498⁎⁎⁎ 1.2352⁎⁎⁎ 1.3109⁎⁎⁎ 1.6444⁎⁎⁎ 1.7590⁎⁎⁎
(0.051) (0.027) (0.000) (0.000) (0.002) (0.003) (0.000) (0.000)
Δyit × IBi – – −2.0149⁎⁎ −2.1102⁎⁎ – – – –
(0.043) (0.043)
Δyit × IBFi – – – – −2.2245 −2.4420⁎ −2.5713⁎⁎ −2.8770⁎
(0.114) (0.091) (0.036) (0.028)
Δyit × IBSi – – – – – – −1.5192 −1.6399
(0.218) (0.192)
Lassetit − 1 −0.3326⁎⁎⁎ −0.3442⁎⁎⁎ −0.2984⁎⁎⁎ −0.3072⁎⁎⁎ −0.3367⁎⁎⁎ −0.3451⁎⁎⁎ −0.3110⁎⁎⁎ −0.3197⁎⁎⁎
(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Capitalit − 1 −1.5829 −1.6085 −1.5434 −1.5745 −1.7264 −1.6818 −1.6361 −1.6210
(0.319) (0.268) (0.292) (0.268) (0.226) (0.192) (0.255) (0.234)
Depositit − 1 0.4654 0.4675 0.4482 0.4486 0.4802⁎ 0.4843⁎ 0.4431 0.4506
(0.165) (0.160) (0.187) (0.198) (0.097) (0.079) (0.146) (0.143)
Inft – −0.0031 – −0.0030 – −0.0004 – −0.0020
(0.808) (0.803) (0.967) (0.858)
Constant 2.7831⁎⁎⁎ 2.8776⁎⁎⁎ 2.5001⁎⁎⁎ 2.5704⁎⁎⁎ 2.8176⁎⁎⁎ 2.8739⁎⁎⁎ 2.6171⁎⁎⁎ 2.6763⁎⁎⁎
(0.000) (0.000) (0.001) (0.001) (0.000) (0.000) (0.000) (0.000)

P-values
AR(1)/AR(2) 0.052/0.810 0.048/0.811 0.048/0.629 0.045/0.626 0.048/0.706 0.047/0.730 0.045/0.605 0.043/0.610
Sargan 0.293 0.304 0.311 0.310 0.429 0.483 0.364 0.374

Note: numbers in parentheses are P-values.


⁎⁎⁎ Indicates significance at 1% level.
⁎⁎ Indicates significance at 5% level.
⁎ Indicates significance at 10% level.
132 M.H. Ibrahim / Economic Modelling 55 (2016) 127–134

Table 4
GMM Estimation Results — Net Loans.

System GMM First-difference GMM


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

ΔLit − 1 0.0071 −0.0030 −0.0003 0.0037 0.0205 0.0242 0.0143 0.0210


(0.926) (0.963) (0.996) (0.950) (0.768) (0.702) (0.823) (0.732)
Δyit 1.2680 2.0581⁎⁎⁎ 1.6997⁎⁎⁎ 2.0383⁎⁎⁎ 0.9088⁎⁎ 1.7377⁎⁎⁎ 1.3180⁎⁎⁎ 1.7271⁎⁎⁎
(0.148) (0.001) (0.007) (0.002) (0.036) (0.000) (0.001) (0.000)
Δyit × IBi – −2.2652⁎⁎ – – – −2.3078⁎⁎ – –
(0.034) (0.029)
Δyit × IBFi – – −2.7940⁎⁎⁎ −3.2338⁎ – – −2.6616⁎ −3.0532⁎⁎
(0.009) (0.059) (0.055) (0.016)
Δyit × IBSi – – – −1.1653 – – – −1.6274
(0.477) (0.235)
Lassetit − 1 −0.2142⁎⁎⁎ −0.2026⁎⁎ −0.2203⁎⁎⁎ −0.2119⁎⁎ −0.3371⁎⁎⁎ −0.3183⁎⁎⁎ −0.3459⁎⁎⁎ −0.3300⁎⁎⁎
(0.008) (0.028) (0.009) (0.029) (0.000) (0.000) (0.000) (0.000)
Capitalit − 1 −2.5414 −2.5152 −2.5509 −2.5139 −1.5282 −1.5869 −1.6637 −1.6344
(0.145) (0.181) (0.177) (0.187) (0.213) (0.204) (0.144) (0.178)
Depositit − 1 0.5831⁎⁎⁎ 0.5456⁎⁎ 0.5701⁎⁎ 0.5534⁎⁎ 0.4353 0.4471 0.4561⁎ 0.4324
(0.008) (0.033) (0.019) (0.031) (0.159) (0.151) (0.075) (0.116)
Inft −0.0032 −0.0030 −0.0038 −0.0044 −0.0002 −0.0001 0.0013 0.0001
(0.802) (0.800) (0.744) (0.717) (0.987) (0.991) (0.899) (0.990)
Constant 1.7414⁎⁎ 1.6560⁎⁎ 1.7948⁎⁎ 1.7281⁎⁎ 2.8332⁎⁎⁎ 2.6687⁎⁎⁎ 2.8964⁎⁎⁎ 2.7782⁎⁎⁎
(0.015) (0.040) (0.015) (0.041) (0.000) (0.000) (0.000) (0.000)

P-values
AR(1)/AR(2) 0.053/0.445 0.051/0.311 0.045/0.342 0.045/0.289 0.053/0.559 0.048/0.364 0.049/0.436 0.046/0.343
Sargan 0.396 0.551 0.393 0.455 0.306 0.377 0.547 0.473

Note: numbers in parentheses are P-values.


⁎⁎⁎ Indicates significance at 1% level.
⁎⁎ Indicates significance at 5% level.
⁎ Indicates significance 10% level.

Table 5
System GMM Estimation Results – Extended Specification.

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

ΔLit − 1 0.0513 0.0639 0.0556 0.0771 0.0546 0.0513 0.0490 0.0721


(0.541) (0.454) (0.603) (0.464) (0.483) (0.640) (0.609) (0.485)
Δyit 2.2269⁎⁎⁎ 2.3637⁎⁎⁎ 2.2170⁎⁎ 2.3085⁎⁎ 2.2946⁎⁎⁎ 2.0428 2.1121⁎⁎ 2.3076⁎⁎
(0.001) (0.005) (0.026) (0.020) (0.006) (0.169) (0.024) (0.018)
Δyit × IBi −2.4290⁎⁎ −2.2087 −2.0452⁎ −1.8185 – – – –
(0.026) (0.167) (0.086) (0.134)
Δyit × IBFi – – – – −3.9634⁎⁎ −2.9504 −3.1723⁎ −2.7162
(0.013) (0.162) (0.056) (0.121)
Δyit × IBSi – – – – −1.2967 0.2944 −0.4039 −0.6270
(0.410) (0.961) (0.768) (0.739)
Lassetit − 1 −0.2230⁎⁎⁎ −0.2398⁎⁎⁎ −0.2352⁎⁎⁎ −0.2155⁎⁎⁎ −0.2261⁎⁎⁎ −0.2700⁎⁎⁎ −0.2360⁎⁎⁎ −0.2270⁎⁎⁎
(0.005) (0.007) (0.006) (0.009) (0.005) (0.003) (0.003) (0.002)
Capitalit − 1 −1.9228 −2.0834 −2.1937 −1.8833 −1.7774 −1.9917 −2.0243 −1.8635
(0.340) (0.340) (0.342) (0.370) (0.306) (0.366) (0.3180 (0.349)
Depositit – 1 0.4529 0.4207⁎⁎ 0.4469⁎⁎ 0.3875⁎ 0.3888 0.3660⁎⁎ 0.3864⁎ 0.3474⁎
(0.138) (0.050) (0.042) (0.055) (0.127) (0.046) (0.067) (0.098)
Lassetit − 1 × IBi – 0.0334 – – – 0.0328 – –
(0.517) (0.563)
Capitalit − 1 × IBi – – 0.7439 – – – 0.7819 –
(0.667) (0.694)
Depositit − 1 × IBi – – – 0.1900 – – – 0.2020
(0.673) (0.664)
Inft −0.0025 −0.0003 −0.0032 −0.0027 −0.0025 −0.0010 −0.0024 −0.0033
(0.908) (0.986) (0.886) (0.888) (0.888) (0.954) (0.905) (0.850)
Rt −0.0105 −0.0264 −0.0128 −0.0282 −0.0090 −0.0232 −0.0142 −0.0300
(0.836) (0.503) (0.772) (0.416) (0.849) (0.556) (0.733) (0.373)
FBi −0.3442⁎ −0.2758 −0.3316 −0.2773 −0.3783 −0.3045 −0.3353 −0.2686
(0.100) (0.180) (0.128) (0.197) (0.101) (0.171) (0.129) (0.228)
Constant 2.0287⁎⁎⁎ 2.1184⁎⁎⁎ 2.1411⁎⁎⁎ 1.9488⁎⁎⁎ 2.0899⁎⁎⁎ 2.3875⁎⁎⁎ 2.1688⁎⁎⁎ 2.0629⁎⁎⁎
(0.003) (0.005) (0.004) (0.008) (0.003) (0.001) (0.002) (0.001)

P-values
AR(1)/AR(2) 0.048/0.513 0.047/0.471 0.060/0.493 0.051/0.488 0.040/0.451 0.060/0.534 0.057/0.478 0.051/0.471
Sargan 0.506 0.626 0.579 0.609 0.513 0.742 0.688 0.674

Note: numbers in parentheses are P-values.


⁎⁎⁎ Indicates significance at 1% level.
⁎⁎ Indicates significance at 5% level.
⁎ Indicates significance at 10% level.
M.H. Ibrahim / Economic Modelling 55 (2016) 127–134 133

substantiate the potential role of especially full-fledged Islamic banks in the average HP-cyclical component over the 4 quarters of the year. The
smoothing aggregate credit. From the table, we find the coefficients results from using this alternative measure of business cycle, termed
of Islamic bank interactive dummy and of full-fledged Islamic bank output gap or hpgap, are presented in Table 6. Results from regressions
interactive dummy to be negative and significant at better than 10% (1) and (2) from the table strengthen the procyclicality of bank lending
significance level. In line with the preceding results, their magnitudes in general for Malaysia. Further, in line with preceding findings, the
are larger in absolute values than the coefficients of the real GDP interaction between hpgap and Islamic bank dummy carries negative co-
growth. In other words, in contrast to the cyclicality of conventional efficients and they are significant at conventional levels of significance
bank lending, the evidence hints on the counter-cyclicality of Islamic (see regressions 3 and 4). Thus, the ability of Islamic banks to stabilize
financing. As for other controlled variables, except for few instances, aggregate credit receives further confirmation. However, when we con-
the log of real assets and funding ratio remains significant throughout sider only the full-fledged Islamic bank dummy, the estimation results
alternative specifications of net loan growth equations. As in the base- indicate significance in only one of the two regressions (regressions 5
line results, capitalization and inflation carry insignificant coefficients. and 6). Finally, when the full-fledged Islamic bank dummy and Islamic
Our second robustness check involves extensions of the baseline bank subsidiary dummy are interacted independently with output gap,
model specification. More specifically, we include foreign ownership none turns out to be significant (regressions 6 and 8).
dummy and money market rate as additional controlled variables. The From these robustness exercises, the preponderance of evidence
lending behaviour of foreign banks vis-à-vis domestic banks has been tends to hint on the role played by Islamic banks in stabilizing aggregate
much discussed in the literature. Meanwhile, the interest rate variable credit during economic downturns. We have no evidence whatsoever
is to control for the bank lending channel of monetary transmission that the Islamic banks are more procyclical in their financing decisions.
mechanisms. Following Micco and Panizza (2006), we further interact Instead, there is ample indication that Islamic financing can be counter-
Islamic bank dummy with bank-specific variables. The results are cyclical. In short, our results provide further support to the “stability”
presented in Table 5. Among these additional and interactive dummy view of the Islamic banks in that they have the ability to stabilize credit.
variables, none turn out to be significant. The real assets remain signifi-
cant while the deposit-liabilities ratio is significant in most regressions.
Central to the present theme, some evidence is further uncovered to 5. Conclusion
support the ability of Islamic banks to stabilize aggregate credit. Namely,
we document the significance of the interactive Islamic bank dummy This paper takes Malaysia, a country with a dual banking system, as a
(regression 1) and of the full-fledged Islamic bank interactive dummy case study to assess the relative procyclicality of Islamic bank financing
(regression 5). In addition, we also document their significant in the ex- and conventional bank loans during the recent cyclical downturns.
tended model having capitalization-Islamic bank dummy interaction Employing dynamic panel approaches and a bank-level panel data set
(regressions 3 and 7). In the remaining regressions, Islamic banks are of 37 banks over the period 2001–2013, we affirm the procyclicality of
found to be no more procyclical than conventional banks. bank lending for the case of Malaysia. However, when distinction is
Our final robustness check employs the Hodrick–Prescott (HP) filter made between Islamic financing and conventional loans, the prepon-
to obtain an alternative measure of business cycle but maintains the derance of evidence tends to confine bank procyclicality only to the lat-
baseline controlled variables. One main criticism of the HP filter is that ter. As regards Islamic financing, it is found to be no more pro-cyclical
it could lead to imprecise end-point estimation. To alleviate this prob- than the conventional bank lending. Indeed, there is some evidence to
lem, we use the data up to 2014. More precisely, we apply the HP filter indicate that Islamic banks in general and full-fledged Islamic banks in
to the log of quarterly GDP seasonally adjusted using data from 2001 to particular can even be counter-cyclical in their financing decisions.
2014. Then, we define the state of the business cycle for a given year as These conclusions hold under a variety of empirical settings including

Table 6
Estimation Results — HP Filtered Business Cycle.

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

ΔLit − 1 0.0494 0.0514 0.0353 0.0421 0.0461 0.0440 0.0328 0.0388


(0.502) (0.515) (0.589) (0.545) (0.537) (0.543) (0.589) (0.543)
Δygapit 1.6997⁎⁎⁎ 2.8129⁎⁎ 2.7088⁎⁎⁎ 3.9351⁎⁎⁎ 2.0234⁎⁎⁎ 2.9646⁎⁎ 2.5017⁎⁎⁎ 3.6414⁎⁎⁎
(0.005) (0.013) (0.000) (0.003) (0.002) (0.016) (0.008) (0.005)
Δygapit × IBi – – −3.6286⁎⁎ −3.4887⁎ – – – –
(0.038) (0.060)
Δygapit × IBFi – – – – −3.0046⁎⁎ −1.5061 −2.2969 −2.2327
(0.034) (0.494) (0.406) (0.440)
Δygapit × IBSi – – – – – – −3.7069 −3.3143
(0.135) (0.251)
Lassetit − 1 −0.2095⁎⁎⁎ −0.2241⁎⁎⁎ −0.2168⁎⁎⁎ −0.2142⁎⁎⁎ −0.2128⁎⁎⁎ −0.2345⁎⁎⁎ −0.2234⁎⁎⁎ −0.2187⁎⁎⁎
(0.002) (0.001) (0.003) (0.003) (0.003) (0.002) (0.003) (0.002)
Capitalit − 1 −2.1393 −2.2000 −2.1559 −2.1796 −2.1843 −2.2962 −2.2031 −2.1943
(0.221) (0.211) (0.185) (0.156) (0.217) (0.196) (0.167) (0.131)
Depositit − 1 0.6847⁎⁎ 0.6605⁎⁎ 0.6579⁎⁎⁎ 0.6104⁎⁎⁎ 0.6896⁎⁎ 0.6491⁎⁎ 0.6494⁎⁎⁎ 0.6033⁎⁎⁎
(0.026) (0.033) (0.009) (0.009) (0.027) (0.025) (0.007) (0.009)
Inft – −0.0146 – −0.0152 – −0.0148 – −0.0145
(0.357) (0.312) (0.342) (0.281)
Constant 1.6301⁎⁎⁎ 1.8066⁎⁎⁎ 1.7040⁎⁎⁎ 1.7549⁎⁎⁎ 1.6585⁎⁎⁎ 1.9112⁎⁎⁎ 1.7701⁎⁎⁎ 1.7979⁎⁎⁎
(0.004) (0.003) (0.007) (0.006) (0.005) (0.006) (0.007) (0.004)

P-values
AR(1)/AR(2) 0.043/0.807 0.047/0.788 0.040/0.749 0.042/0.713 0.043/0.764 0.046/0.770 0.039/0.772 0.041/0.737
Sargan 0.398 0.464 0.480 0.512 0.414 0.512 0.522 0.549

Note: numbers in parentheses are P-values.


⁎⁎⁎ Indicates significance at 1% level.
⁎⁎ Indicates significance at 5% level.
⁎ Indicates significance at 10% level.
134 M.H. Ibrahim / Economic Modelling 55 (2016) 127–134

an alternative loan measure, alternative extensions of baseline specifi- Bertay, A.C., Demirguc-Kunt, A., Huizinga, H., 2015. Bank ownership and credit over the
business cycle: is lending by state banks less procyclical? J. Bank. Financ. 50 (326–
cation and an alternative measures of economic cycle. 229).
From these results, we may conclude that Islamic banks can play a Blundell, R., Bond, S., 1998. Initial conditions and moment restrictions in dynamic panel
stabilizing role during the time of adverse shocks and, hence, provide data models. J. Econ. 87, 115–143.
Bourkhis, K., Nabi, M.S., 2013. Islamic and conventional banks' soundness during the
further support to the “stability” view of the Islamic banking system. De- 2007–2008 financial crisis. Rev. Financ. Econ. 22, 26–77.
spite the observational equivalence between Islamic and conventional Brei, M., Schclarek, A., 2013. Public bank lending in crisis times. J. Financ. Stability 9,
banks (Chong and Liu, 2009; Khan, 2010) and limited alternative 820–830.
Brei, M., Gambacorta, L., von Perter, G., 2013. Bank rescue packages and bank lending.
sources of funds and hedging instruments in the face of adverse shocks,
J. Bank. Financ. 37, 490–505.
Islamic banks in Malaysia have been able to smooth or even increase Chachi, A., 2005. Origin and development of commercial and Islamic banking operations.
their financing during the recent downturn experienced by Malaysia J.KAU. Islam. Econ. 18 (2), 3–25.
Chong, B.S., Liu, M., 2009. Islamic banking: interest-free or interest-based? Pac. Basin
as a consequence of the global financial crisis. We believe that, along
Financ. J. 17, 125–144.
the line of Farooq and Zaheer (2015), the Islamic principles governing Cihak, M., Hesse, H., 2010. Islamic banks and financial stability: an empirical analysis.
Islamic banking operations together with religiosity concerns may J. Financ. Serv. Res. 38, 95–113.
have made Islamic banks to be more stable. However, to be more Cull, R., Martinez Peria, M.S., 2013. Bank ownership and lending patterns during the
2008–2009 financial crisis: evidence from Latin America and Eastern Europe.
concrete, we also believe that further analyses are needed. Perhaps, J. Bank. Financ. 37, 4861–4878.
the issue of Islamic financing procyclicality should be extended to Daher, H., Masih, M., Ibrahim, M.H., 2015. The unique risk exposures of Islamic banks' cap-
other dual banking systems or should cover more banks under different ital buffers: a dynamic panel data analysis. J. Int. Financ. Mark. Inst. Money 36, 36–52.
Farooq, M., Zaheer, S., 2015. Are Islamic banks more resilient during financial panics? Pac.
jurisdictions. Econ. Rev. 20 (1), 101–124.
Ferri, G., Kalmi, P., Kerola, E., 2014. Does bank ownership affect lending behaviour?
References Evidence from the euro area. J. Bank. Financ. 48, 194–209.
Fungacova, Z., Herrala, R., Weill, L., 2013. The influence of bank ownership on credit
Abdul-Rahman, A., Abdul-Latiff, R., Muda, R., Abdullah, M.A., 2014. Failure and potential of supply: evidence from the recent financial crisis. Emerg. Mark. Rev. 15, 136–147.
profit-loss sharing contracts: a perspective of new institutional economics (NIE) Hasan, M., Dridi, J., 2010. The effects of the global crisis on Islamic and conventional
theory. Pac. Basin Financ. J. 28, 136–151. banks: a comparative study. IMF Working Paper #WP/10/201.
Albertazzi, U., Bottero, M., 2014. Foreign bank lending: evidence from the global financial Huang, H.-C. (River), Fang, W., Miller, S.M., Yeh, C.-C., (2015). The effect of growth
crisis. J. Int. Econ. 92, S22–S35. volatility on income inequality. Econ. Model. 45, 212–222.
Arellano, M., Bond, S.R., 1991. Some tests of specification for panel data: Monte Carlo Hussain, M., Shahmoradi, A., Turk, R., 2015. An overview of islamic finance. IMF Work
evidence and application to employment equations. Rev. Econ. Stud. 58, 277–297. Papers #WP/15/120.
Arellano, M., Bover, O., 1995. Another look at the instrumental variables estimation of Khan, F., 2010. How ‘Islamic’ is Islamic Bank? J. Econ. Behav. Organ. 76, 805–820.
error components model. J. Econ. 68, 29–51. Micco, A., Panizza, U., 2006. Bank ownership and lending behaviour. Econ. Lett. 93,
Azmat, S., Skully, M., Brown, K., 2015. Can Islamic banking ever become Islamic? 248–254.
Forthcom. Pac. Basin Financ. J. Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient two-
Beck, T., Demirguc-Kunt, A., Merrouche, O., 2013. Islamic vs. conventional banking: step GMM estimators. J. Econ. 126, 25–51.
business model, efficiency and stability. J. Bank. Financ. 37, 443–447.
Belanes, A., Fitti, Z., Regaieg, R., 2015. What can we learn about Islamic banks efficiency
under the subprime crisis? Evidence from GCC region. Pac. Basin Financ. J. 33, 81–92.

Вам также может понравиться