Sei sulla pagina 1di 13

Pacific-Basin Finance Journal 48 (2018) 99–111

Contents lists available at ScienceDirect

Pacific-Basin Finance Journal


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

How profitability differs between conventional and Islamic banks:


T
A dynamic panel data approach
⁎,1
Halit Yanikkayaa, , Nihat Gumusb, Yasar Ugur Pabuccuc
a
Department of Economics, Gebze Technical University, P.K. 141, 41400 Gebze, Kocaeli, Turkey
b
Department of Management, Ibn Haldun University, Ulubatlı Hasan Cd. No: 2, 34494 Başakşehir, Istanbul, Turkey
c
Kuveyt Turk Participation Bank Research & Development Center, Kule Plaza K:9, 42060 Selçuklu Konya, Turkey

AR TI CLE I NF O AB S T R A CT

JEL classification: This paper analyzes and compares the dynamics for the profitability of conventional banks and
E44 Islamic banks in the Organization of Islamic Cooperation countries and the United Kingdom
G21 between 2007 and 2013 using a sample of 74 Islamic and 354 conventional commercial banks.
Z12 “Net interest margin” and “return on asset” are employed as variables representing the profit-
Keywords: ability and several new explanatory variables are introduced such as, the usage of self-service
Net interest margin banking channels, penetration of financial services, crude oil/agriculture price indexes and asset
Return on asset ratio of non-Murabahah assets of Islamic Banking. Dynamic panel data estimates indicate that
Islamic banking
almost all explanatory variables of profitability for conventional and Islamic banks are different
Banking profitability
implying that profitability of Islamic banks relies on the different dynamics than that of con-
ventional ones. Both profitability measures are not persistent over time and neither of them has
significant relationship with the country specific macroeconomic variables. Estimation results
imply the importance of new product and alternative channel development in enhancing the
profitability of Islamic banks. Moreover, our analysis shows that the usage of products which
promotes more risk sharing as compared to the products based on Murabahah structure can
contribute to the performance of Islamic banks.

1. Introduction

The purpose of this paper is to compare and contrast the dynamics for the profitability of Islamic Banks (IBs) and conventional
banks (CBs) so as to contribute to the analysis of current Islamic banking phenomenon in the Organization of Islamic Cooperation
(OIC) countries and the United Kingdom (UK). Islamic finance, as an alternative approach to conventional finance practices, has
gained a remarkable momentum in recent years. According to the ICD Thomson Reuters (2015) and Islamic Financial Services Board
(2016) data the global size of Islamic financial assets is predicted to grow from 861 USD billion to more than 1.88 USD trillion from
2008 to 2015. This corresponds to a compound annual growth rate around 12%. As the flagship sector of Islamic finance industry,
Islamic banking activities constitute around 74% of all Islamic financial assets as of the end 2014.
As per the county-wise distribution of Islamic finance assets, three countries come forward, namely Malaysia, Saudi Arabia, and
Iran. These three countries host almost two thirds of Islamic financial assets worldwide. Gulf countries like the United Arab Emirates
(UAE), Kuwait, Qatar, and Bahrain are other prominent markets in terms of Islamic finance holding almost one fourth of global assets.
Countries, like Turkey, Indonesia, Bangladesh, Pakistan and Sudan are other important markets. Recent developments in markets like


Corresponding author.
E-mail addresses: halityanikkaya@gtu.edu.tr (H. Yanikkaya), nihat.gumus@ihu.edu.tr (N. Gumus), yasar.pabuccu@kuveytturk.com.tr (Y.U. Pabuccu).
1
The first author acknowledges support from the Turkish Academy of Sciences.

https://doi.org/10.1016/j.pacfin.2018.01.006
Received 29 July 2017; Received in revised form 15 January 2018; Accepted 24 January 2018
Available online 02 February 2018
0927-538X/ © 2018 Elsevier B.V. All rights reserved.
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

the UK and Luxemburg displayed that Islamic finance is not a phenomenon just confined to Muslim majority countries given the
recent trend in some Western developed markets.
Besides the prohibition on interest receiving and taking (Riba), gambling (Maisir), excessive uncertainty (Gharar) and restriction
on investing on some sectors producing products prohibited under Islamic jurisprudence, what makes Islamic finance peculiar is that
even the financial contracts should be established based on assets through the principle of profit-loss-sharing (PLS). The requirements
of these principles are to be fulfilled through some contractual obligations which are constructed upon buying and selling of
(Murabahah), leasing (Ijarah) of or partnership in (Musharakah/Mudharabah) in an asset or a portfolio of assets (Iqbal and Mirakhor,
2011). It is argued that it is this PLS structure contributing to the capacity of Islamic banks in absorbing external shocks as compared
to their conventional counterparts (Khan and Mirakhor, 1989) and to the capacity of adding to the overall economic growth by
enabling long-term funding (Chapra, 1992; Imam and Kpodar, 2015). The advantageous nature of the PLS model employed by IBs has
empirically supported by Hasan and Dridi (2010).
Despite those conclusions about the business model of Islamic banking the daily practice of IBs has always been a matter of
discussion. Theoretically, it is assumed that the business model that should be employed by IBs which promotes profit and risk
sharing both at the asset and liability side through rather asset based or asset-backed, partnership-like contractual structures.
However, in practice, the contractual relationships constructed between IBs and their customers on the asset side are mostly based on
transactional structures such as Murabahah, which leads to the conclusion that the behavior of IBs are alike to the CBs (Khan, 2010;
and Azmat et al., 2015). On the liability side the relationships are usually established similar to the ones at CBs. This situation does
also support the view that the operations of IBs are very similar to the CBs so that there would be no difference between the dynamics
determining their performances (El-Hawary et al., 2007).
Besides the issue of the difference between the theory and practice of Islamic banking there are issues how Islamic banks fit in the
overall banking and finance environment in terms of regulatory framework. Whether Islamic banks should be subject to different
regulatory framework as compared to conventional banks, how the regulatory framework should be established and how different
regulations affect the efficiency and profitability of IBs are some of the important themes (Song and Oosthuizen, 2014; Mejía et al.,
2014; Bitar, 2014; Solé, 2007). The conclusions point into the differences in the regulatory practices across countries where Islamic
finance activities are available, into the need for enhanced standards for the regulation of Islamic finance activities and into the
different effects of similar regulatory frameworks on the efficiency and performance of IBs as compared to conventional financial
institutions. Meanwhile one important point in this regard is the need for standardization in the Sharia governance practices for
various Islamic finance activities. (Godlewski et al., 2014).
Questions such as of what are the dynamics of the performance of IBs and whether there are differences in those dynamics as
compared the CBs have started to be asked more frequently due the recent growth trend in Islamic finance and banking industry.
However, the aforementioned issues make it hard to analyze Islamic banking, their model, profitability and efficiency. Some em-
pirical comparisons of IBs and CBs conclude that Islamic Banks have better asset quality, are better capitalized but have few business
model differences (Beck et al., 2013), are more resilient to the financial crises (Hasan and Dridi, 2010; Beck et al., 2013) and can be
distinguished via financial ratios (Olson and Zoubi, 2008). Khediri et al. (2015) also differentiate IBs via credit and insolvency risk,
operating leverage and off-balance sheet activities but profitability and liquidity ratios are not very much different.
The main purpose of this paper is to contribute to the related literature by investigating the dynamics of performance for IBs in
comparison to the CBs. Employing a dynamic panel data approach for a sample of 74 IBs and 354 CBs in the OIC countries and the UK
for the period between 2007 and 2013, the study aims to reveal;

• key factors affecting both conventional and Islamic bank profitability using the dynamic panel data approach,
• how usage of self-service banking channels, level of financial penetration, Islamic Finance development within a country affect the
profitability of banks, and
• how the breakdown of instruments of Islamic banks contributes to their profitability.
To a great extent, some findings of the study regarding the dynamics of CBs in the OIC countries and the UK turned out to be
mostly in line with previous studies. However, our results for IBs imply important differences as respect with the existing literature.
For instance, the estimation results imply that the profitability ratios used in the analysis turned out to be not persistent over time. In
addition, unlike some studies in the literature, some explanatory variables used to explain the variations in the performance of IBs are
found out to be insignificant in the analysis. For instance, the key determinants of net interest margin (NIM)2 such as operation costs
and loan to assets seem to have no significant effects for IBs. Further, considering the fixed effect estimations, the most widely used
technique in the literature for IBs, and the GMM estimation results are substantially different. One of the main contributions of the
study has thus been the usage of panel data method for the analysis, which provides the substantial evidence that the analysis of the
performance dynamics of IBs is sensitive to the methodologies employed.
The rest of the paper is organized as follows: Section 2 summarizes the existing literature on the performance of IBs as compared
to CBs while Section 3 describes the data set, statistical model, and variables. Section 4 reports the descriptive statistics and discusses

2
Even though IBs do not engage in interest-bearing contracts and operations, the NIM concept can also be applied to them since NIM is a ratio related to the
difference between revenues from financing operations and costs of those activities. In case of IBs, the revenues stand for revenues generated via investing and lending
activities while the cost of financing may include the profits distributed to depositors or investment account holders. Therefore, within the study the performance
measure of NIM has been used both for IBs and CBs in order for keeping the terminology simple and intact.

100
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

the empirical results on the determinants of return on asset and bank margins. Finally, Section 5 concludes while pointing out to some
implications for further research.

2. Literature review

In the empirical literature, banking profitability or efficiency is mainly measured by return on asset (ROA), return on equity (ROE)
and net interest margin (NIM) and modeled through similar bank specific and country specific variables mostly in single step re-
gressions.
Theories explaining profitability mostly focus on market conditions (such as concentration) and persistency of profits. The
structured-conduct-performance hypothesis asserts that if a market is in the equilibrium, profits of companies tend to converge.
However, high market shares and well-differentiated products can lead to non-competitive profits. On the other hand, the efficient
structure hypothesis suggests that managerial and scale efficiencies are the key factors for profitability. Another effect of market
concentration is that it leads to persistency in profit rates. Other than market concentration, persistency can also arise from reg-
ulations such as entry barriers preventing banking market to reach the equilibrium.
One of the earliest study comes from Short (1979) and it shows that higher concentration increases the profit rates of the banks.
Bourke (1989) can be regarded as one of the pioneering empirical analysis of overall profitability of CBs, which in general use ROA
and/or ROE as performance measures. Bourke (1989) models overhead costs, capital ratios, liquidity-credit risk and external
factors (concentration, ownership, interest rates) as the determinants of profitability and finds internal factors and concentration as
the positive determinants of overall bank profitability. Reiterating Bourke's study for the banks operating in the European Union
(EU) countries, Molyneux and Thornton (1992) employ ROE as a profitability measure and confirm the results of the Bourke's
analysis. Goddard et al. (2004) find out that that capital to asset ratio has a positive effect on the profitability of the banks in the EU
region. Pasiouras and Kosmidou (2007), by focusing on the bank ownership as foreign or domestic, confirm the positive effect of
capital adequacy ratios on the bank performance while reporting the negative impact of cost to income ratio on the performance. In
addition, the study concludes that GDP growth and inflation have opposite effects on ROA for both domestic and foreign banks.
Focusing on the performance of Greek banks by employing a GMM method, Athanasoglou et al. (2008), conclude that ROAs of
Greek banks are mostly determined by bank-level variables. As another study using the GMM methodology, Dietrich and
Wanzenried (2014) analyze a large sample including the data of 10,165 commercial banks from 118 countries for the period
between 1998 and 2012. The main finding of the study is that the determinants of bank profitability vary depending on the country
income levels. For instance, the equity to assets ratio has a significant impact on ROA for just high income countries. While the
foreign ownership is found to be positively linked to performance in low income countries, it turned to be negatively linked for
high income countries.
On the other hand, the literature deploying NIM as a performance measure have started with the dealership model of Ho and
Saunders (1981). The two step dealership model considers a bank as a risk averse dealer that maintains a pure margin depending on
risk-neutral spread (result of deposit and loan arrivals to the bank), size of transactions, level of bank's risk aversion and variance of
interest rates. A number of studies, such as Angbazo (1997), Saunders and Schumacher (2000), Brock and Suarez (2000), and
Valverde and Fernández (2007), employ this model to explain NIM. They find operation costs, credit and default risk, implicit interest
payments, capital to asset ratios as significant determinants of NIM. Hawtrey and Liang (2008) show that implicit interest payments
and operation costs are positively related and cost to income and size are negatively related with NIM.
Another strand of the literature, such as Demirguc-Kunt and Huizinga (1998), Demirguc-Kunt and Huizinga (2000), Kasman et al.
(2010), López-Espinosa et al. (2011), Dietrich and Wanzenried (2014), analyze the determinants of NIM with single step regressions
for CBs. Demirguc-Kunt and Huizinga (1998) investigate how a bank's profitability is related with taxation, the structure of the
financial system, and with some legal and institutional indicators. The study concludes that banks in developing countries have
higher interest and deposit insurance reducing interest margins. Demirguc-Kunt and Huizinga (2000) find that for developing
countries the financial systems tend to be more bank-based, the profitability rates of banks are higher, and the overhead costs are
significantly related with NIM. Maudos and Solís (2009) utilize the GMM to explain NIM of Mexican banks between 1993 and 2005.
They show that major portion of NIM comes from operating costs and market power. Non-interest income is also significant but its
effect is very low.
Kasman et al. (2010) inquire for any relationship between financial reform processes and the level of NIM of banks among new
members and the EU candidates. The paper suggests that the determinants of the net interest margin are different across East and
Central European countries. López-Espinosa et al. (2011) question whether the accounting standards and macro-economic variables
have an impact on NIM for a sample of the banks from 15 developed and emerging markets covering the period from 1999 to 2008
and find that accounting practices in compliance with International Financial Reporting Standards (IFRS) have a negative impact on
NIM. Dietrich and Wanzenried (2014) find that banks in low income countries depend heavily on interest income and have larger
margins.
The literature on the profitability of Islamic banks is relatively premature as compared to the studies on conventional banks.
Bashir (2003) analyzes the determinants of profit before taxes for eight Middle East countries for a period between 1993 and 1998.
The high cost of capital levels in high income Muslim countries reduce the profitability of banks. GDP per capita, inflation, and
foreign ownership are positive determinants. Hassan and Bashir (2003) investigate the factors affecting the profitability of Islamic
banks from 21 countries for a period covering 1994 to 2001 by using ROA, ROE, non-interest income margin as dependent variables.
It is concluded that the profitability of Islamic banks responds positively to the increases in capital and negatively to loan ratios. The
relationship of GDP growth is found to be significant and positively related with ROA and ROE while the GDP per capita is a

101
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

significant determinant for ROA but the overhead cost turns to be not significant. Haron (2004) makes a panel data analysis of five
profitability measures. The study concludes that all types of funds collected through current, saving, or investment accounts are
positively related to the profitability of Islamic banks. The level of interest rates, inflation and bank size have significant positive
impacts on the profits of both conventional and Islamic banks.
Kasman et al. (2010) examine the profitability of African IBs for the period from 1999 to 2009 and find that bank capital and size
(positive), operation cost (negative), economic growth, inflation and banking industry concentration (positive) are significant de-
terminants of the profitability. Noor and Ahmad (2011) analyze the relationship between the ROE and the efficiency factors of 78
Islamic banks from 25 countries for the period covering 1992 to 2009. The fixed effect estimation results indicate that more profitable
banks have higher operation costs, higher equity to asset ratio. Masood and Ashraf (2012) report that operating expense, non-
performing loans are negatively and banks size positively related to ROA for 25 IBs from 12 countries.
Zeitun (2012) studies the profitability of around 40 IBs and CBs from the Gulf Cooperation Council (GCC) countries for the period
from 2002 to 2009 using ROA and ROE as performance measures. The analysis concludes that operating costs and inflation have
negative significant effects on ROA and ROE for both types of banks. Level of equity and GDP growth are found to have positive
effects on CBs but their impacts are insignificant for IBs. Fatnassi et al. (2014) examine the performance of Islamic and conventional
banks of the GCC region via capital and risk levels using data from 2003 to 2011 for a sample of 65 conventional and 48 Islamic banks
with the GMM method. Study shows that the ROA and ROE levels of IBs are persistent and profitability levels of highly capitalized IBs
are lower.
Studies using NIM as the performance measure of IBs are quite rare in the literature. As a recent study, Sun et al. (2014) focus on
the determinants of net interest margin for conventional and Islamic Banks from OIC countries for the years between 1997 and 2010
using a fixed effect model. Operational costs and lagged NIM, and capital adequacy are significantly positive determinants for both
types of banks' performance. However, size (positive), loan loss provisions (negative), liquidity (positive) are stated to have sig-
nificant impact only on the NIM of Islamic Banks. On the other hand, the study finds out that risk aversion (negative), management
efficiency (negative), implicit interest payments (positive), and Lerner index (positive) are significant determinants of the perfor-
mance of just CBs. Sun et al. (2017) repeats the previous NIM study for OIC banks between 1999 and 2010 with the GMM method.
Unlike the previous research with the fixed effects, lagged NIM is no longer significant for CBs. Operation costs are also not significant
for both types of banks. Only Lerner index and lagged NIM are significant variables for IBs. Fatnassi et al. (2014) by using GMM,
report that NIM of IBs are not persistent. Equity to asset, loan loss provisions are negative and loan to asset, inflation, size of total
loans are positive determinants of profitability for IBs.
One of the main specifications of the previous empirical studies on CB and IB performance is that they either focus on a specific
region or contain just limited number of banks by utilizing mostly static panel data techniques. There is no cross-country study with
bank and country specific variables. Our study substantially contributes the literature by gathering a dataset for larger number of
countries and banks from a different data source and employing dynamic panel estimation methods with some new explanatory
variables. Our results mostly confirm the literature focusing on CBs. Profitability measures are persistent (Goddard et al., 2004;
Athanasoglou et al., 2008), banks tend to add their operation costs to their margins (Brock and Suarez, 2000; Valverde and
Fernández, 2007; Demirguc-Kunt and Huizinga, 2000), and credit risk negatively related with profitability (Athanasoglou et al.,
2008; Dietrich and Wanzenried, 2014). On the other hand our results differ from the previous studies examining IBs. Profitability of
IBs are not persistent and not significantly related with macroeconomic variables.

3. Data and methodology

The main data source in our study is the Financial Times Banker Database. The Banker Database provides financial data of 5000
leading banks of the world from more than 160 countries and the data have been normalized for regional reporting. Year end result is
used in the dataset and in case of mid-quarter reporting we move the data of the last 6 months to the next year. We empirically
examine 74 IBs and 354 conventional commercial banks from the OIC member countries and the UK for the period from 2007 to
2013. We extensively have searched the central bank websites of sample countries and the Thomson Reuters Zawya database to
determine Islamic Banks. CBs with Islamic Banking windows are assumed to be as conventional banks as the major part of their
income come from interest-related operations. Table 1 reports the number of Islamic and conventional banks for each country within
the dataset. For country-level macroeconomic and financial variables the World Bank World Development Indicators database is
used.
Our study employs a large number of bank-specific variables. We use Loan Loss Provision (LLP) for Credit Risk, Loan to Asset
Ratio for Liquidity, Risk Weighted Assets for a bank's general risk aversion level and bank size. We use two different measures of
capital adequacy, namely the equity to asset ratio and Basel Capital Adequacy Ratio.3 Non-interest margin, Lerner index, Foreign/
Local ownership and lagged NIM, lagged ROA, operation costs are the other idiosyncratic determinants that are supposed to have
impact on the profitability. We exclude non-interest margin from the estimates for ROA as non-interest revenues are the directly
included within the calculation of ROA. Non-Murabahah Asset Ratio is a bank-level variable used for the analysis of the performance
of IBs. The data concerning the non-murabaha levels of the sample IBs is collected from the Islamic Banks Information System (IBIS)
that is developed by Islamic Development Bank.

3
We use two different measures of capital adequacy equity to asset and Basel capital adequacy ratio but we report only the regressions with Basel capital adequacy
ratio as results do not alter significantly.

102
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 1
Country specific variables and number of banks.

Country Inflation GDP Regulator Banking Service Ratio of Online # of banks, Islamic Financial # # of
Name (annual %) Growth Quality Accessibility (1 if Borrowers to Banking POS and ATM Development of CBs
(annual %) high) Savers (%) Usage (%) (1 if high) Indicator IBs

Albania 2.86 2.50 0.21 0 136.73 14.69 1 4.77 1 14


Azerbaijan 4.13 3.98 −0.38 0 353.82 1.29 1 4.07 1 27
Bahrain 1.90 3.58 0.70 1 61.52 17.51 1 76.41 11 18
Bangladesh 8.02 5.92 −0.85 0 132.72 16.71 0 24.92 5 31
Egypt, Arab 10.09 3.27 −0.37 0 154.50 1.16 0 20.43 4 29
Rep.
Guinea 21.35 3.91 −1.00 0 68.12 3.41 0 1.47 2 7
Indonesia 5.08 5.96 −0.32 0 49.45 3.84 1 27.57 7 73
Iran, Islamic 18.09 2.11 −1.62 1 145.91 24.45 1 16.84 17
Rep.
Iraq 2.88 5.54 −1.05 0 118.84 4.98 0 4.40 1 2
Jordan 3.77 3.19 0.23 0 361.33 2.41 0 36.39 3 10
Kuwait 4.24 2.63 0.08 1 55.17 19.45 1 38.00 4 7
Lebanon 2.84 8.94 0.03 0 89.09 2.99 1 17.42 18
Malaysia 1.98 4.66 0.54 0 57.79 11.78 1 93.18 15 48
Pakistan 12.44 2.57 −0.61 0 45.81 3.40 0 34.39 5 24
Qatar 0.12 10.78 0.66 0 49.46 10.97 1 39.58 4 7
Saudi Arabia 4.46 5.21 0.11 0 78.39 17.75 1 30.64 5 10
Senegal 0.71 3.50 −0.05 0 53.52 6.25 0 5.73 1 11
Sudan 23.92 −0.28 −1.37 0 56.11 26.00 0 27.83 18
Tunisia 4.97 2.85 −0.22 0 77.52 4.78 1 14.95 2 23
Turkey 7.66 4.75 0.37 0 219.29 9.63 1 13.50 4 44
United Arab 0.99 2.57 0.55 1 48.10 20.00 1 57.44 7 19
Emirates
United 3.18 0.88 1.69 1 40.39 49.35 1 16.16 3 96
Kingdom
Yemen, Rep. 14.18 −4.28 −0.69 0 48.67 0.96 0 11.07 3 5

Variable sources are in the text.


Due to limited data points in data set 74 Islamic banks and 354 Interest based banks analyzed.

As per country-specific variables, our study employs inflation rate, GDP growth rate, interest rate volatility,4 exchange rate
volatility and regulator quality as overall economic-financial indicators. Regulatory quality is the ability of the government to for-
mulate and implement sound policies and regulations that promote private sector development.5 We also use several country-specific
variables to proxy financial inclusion and infrastructure development level. In this regard, the ratio of borrowers to savers is the ratio
of population borrowed from a financial institution to population saved in a financial institution during the past 12 months. Banking
service coverage is a dummy variable takes “1” if more than 50% of people (as the % of the population over the age 15) have bank
account or credit/debit cards (Data derived from the World Bank Findex Database). Number of banking branches, POS and ATM
Machines is a dummy variable takes ‘1’ if average number of banks, POS and ATM machines per 100.000 people is larger than the
median of the dataset. Usage of self-service banking channels is the ratio of population who make electronic payments or mobile
banking transactions for the last year. This variable is not a bank- level variable as we do not have data on each and every bank's
internet channel. Instead we use the overall internet-mobile banking penetration in the country. We use price indices of Crude Oil,
Agriculture and Metal and Minerals to observe how Islamic and conventional banks' profitability and margins respond to commodity
price changes. All country level variables and price indices are derived from the World Bank World Development Indicators data-
bases. Central bank interest rates are retrieved from www.tradingeconomics.com and exchange rates are retrieved from
www.oanda.com. We also utilize a variable on Islamic Finance Development provided by Thomson Reuters to measure the maturity
of Islamic banking in a country. Table 2 describes the variables and expected relations.

3.1. Econometric model

Our study employs a single step dynamic model similar to Maudos and Solís (2009) and Athanasoglou et al. (2008) to analyze the
determinants of profitability for both CBs and IBs. We describe two equations: one for NIM and one for ROA. These specifications
include lagged levels of dependent variables in order to reflect the persistent nature of profitability.

4
We use three measures for interest rate volatility: standard deviation, Z-Score and slope and only report the estimation results for standard deviation as the results
for the other variables of volatility are almost the same.
5
We also employ industrial production index in the estimates. Since it has insignificant coefficients and also substantially reduces the dataset, we excluded it during
the analysis.

103
Table 2
H. Yanikkaya et al.

Variable definitions.

Variable Measured Effect Definition -Calculation Expected Impact (Bank Expected Impact (ROA)
Margin)

Bank Specific Variables


Net interest margin(NIM) Dependent Variable IB NIM: (Financing Revenue − Financing Expense) / Total Asset
Return on Asset (ROA) CB NIM: (Interest Revenue − Interest Expense) / Total Asset
ROA: Total Profit Before Tax / Total Asset
Lagged Profitability Previous Year's Effect Lagged NIM or Lagged ROA Positive for both Positive for both
Equity/Asset Capital Adequacy Total Equity / Total Asset CB: (+) IB: (NS) CB: (+) IB: (+)
BIS Capital Adequacy Ratio Capital Adequacy (Tier1 Capital + Tier2 Capital) / Risk Weighted Asset CB: (+) IB: (+) CB: (+) IB: (+)
Loan Loss Provision Credit Risk Provision, Impairment charges for loan loss / Total Loans Positive for both Negative for both
Loan to Asset Ratio Liquidity Total Loan / Total Asset CB: (+) IB: (NS) Positive for both
Operations Cost to Total Asset Operation Cost Staff and Administration Expenses / Total Asset Positive for Both Negative for both
Size (Log(Asset)) Size Log(Total Asset) Positive for Both Positive for Both
Risk Weighted Asset to Total Asset Overall Riskiness Total Risk Weighted Asset as defined by Basel Committee / Total Asset CB: (+) IB: (NS) Positive for Both
Non-Interest Income Effect of income Fees & Non Interest & Non Finance Income / Total Asset Negative for Both —Excluded—
charges
Lerner Index Market Power (Total Income − Operation Costs) / Total Income CB: (+) IB: (NS) CB: (+) IB: (NS)
Foreign Ownership Foreign Ownership Dummy variable takes 1 if bank is a foreign subsidiary CB: (+) IB: (NS) CB: (+) IB: (NS)
Non-Murabaha Asset Ratio Islamic Banking Instruments Ratio of Non-murabaha (non-cost plus sales) assets to total assets IB: (+) IB: (+)

104
Country Specific Variables
Inflation: Consumer Price Index Inflation Effect Annual Inflation Rate Positive for Both Positive for Both
GDP Growth Macroeconomic Effect Annual percentage growth rate of GDP at market prices based on constant local currency Positive for Both Positive for Both
Regulatory Quality Regulation Ability of the government to formulate and implement sound policies and regulations that Negative for both Negative for both
permit and promote private sector development.
Country Dummies Regional Effects Dummy variable takes 1 for GCC, Malaysia and UK banks Ambiguous Ambiguous
Ratio of Borrowers to Savers Overall Financial Structure The ratio of population borrowed from a financial institution to population saved in a CB: (+) IB: (NS) Positive
financial institution in past 12 months.(Data derived from Findex Database) For both
Banking Service Coverage Financial Service Penetration Dummy variable takes ‘1’ if more than 50% of people (% age 15+) have bank account or Negative for both Negative for both
credit/debit cards (Data derived from World Bank Findex Database)
Banking Branches, POS and ATM Financial Service Penetration Dummy variable takes ‘1’ if Average # of banks, POS and ATM machines per 100,000 Negative for Both Negative for Both
Machines people is high. (Data derived World Bank Findex Database. Countries with bigger value
than median take ‘1’)
Usage of self-service banking channels Self-service Banking Ratio of population who make electronic payments or mobile banking transactions for the Not Significant for both CB: (+)
prevalence last year. (Data derived from World Bank Findex Database) IB: (NS)
Islamic Financial Development Indicator Islamic Banking Development Composite weighted index that measures the overall development of the Islamic finance CB: (NS) IB: (+) CB: (NS) IB: (+)
provided by Thomson Reuters.
Price Indices for Crude Oil, Agriculture Global Prices Agriculture, 2010 = 100, nominal$ Crude oil, Dubai, $/bbl, nominal$ Metals and CB: (NS) IB: (−) CB: (NS) IB: (−)
and Metal&Minerals minerals, 2010 = 100, nominal$ reported in World Bank database

(+): Positive relation expected (−): Negative Relation expected (NS): No significant relation expected.
Pacific-Basin Finance Journal 48 (2018) 99–111
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 3
Descriptive statistics and t-tests of bank specific variables.

Variable Conventional Banks Islamic Banks t-Test for Equal Means

Obs Mean Std. Dev. Min Max Obs Mean Std. Dev. Min Max t Value p Value

⁎⁎
Bank Margin 1243 3.14 2.12 −3.10 33.32 226 2.86 1.83 −4.74 14.96 2.06 0.040
ROA 1241 1.48 2.35 −29.97 19.68 225 0.84 3.78 −34.31 5.43 2.45⁎⁎ 0.015
Non int Margin 1243 1.51 1.37 −9.94 13.38 226 1.34 1.84 −14.68 13.18 1.31 0.191
BIS Capital Adequacy Ratio 1221 19.74 12.91 0.56 204.19 224 19.65 11.60 4.05 86.35 0.098 0.922
Equity to Asset 1243 10.87 7.10 −10.25 73.32 226 13.62 10.56 1.60 75.84 −3.75⁎⁎⁎ 0.000
Loan Loss Provision 1243 1.20 2.67 −29.62 46.24 226 2.35 12.82 −2.47 191.25 −1.3457 0.180
Loan to Asset 1243 61.39 18.23 0.44 155.41 226 66.23 15.53 7.95 91.75 −4.18⁎⁎⁎ 0.000
Operation Costs 1243 2.27 1.58 −3.83 17.62 226 2.13 1.57 0.53 13.11 1.224 0.222
Size 1243 3.73 0.82 1.75 6.44 226 3.69 0.60 1.68 4.90 0.76 0.450
Risk Weighted Asset Ratio 1243 69.09 21.75 12.50 205.53 226 75.55 25.06 5.95 227.57 −3.636⁎⁎⁎ 0.000
Bank Lerner Index 1243 49.10 46.26 −586.78 793.67 226 78.74 317.99 −351.38 3971.28 −1.398 0.163

H0: Equal means.


Two sided p-value reported The test allows for the variance to be different between the two groups.
⁎⁎⁎
Means are not equal at 1% level.
⁎⁎
Means are not equal at 5% level.

k j
NIM = ξNIMit − 1 + ∑ γ kBSitk + ∑ β jCSitj + ηi + υit
k=1 j=1

k j
ROAit = ξROAit − 1 + ∑ γ kBSitk + ∑ β jCSitj + ηi + υit
k=1 j=1

NIM: Bank Margin ROA: Return on Asset


NIMit-1: One period Lagged NIM ROAit−1: One period Lagged ROA
BS: Bank Specific Variables CS: Country Specific Variables
ηi: Bank Specific error terms υit_: Error Term factor
it: bank i at time t
We actually run several set of regressions. First, bank-specific variables and macroeconomic variables are regressed. We then
sequentially add financial penetration and inclusion variables, region dummies and commodity price indexes.

4. Empirical results

4.1. Descriptive statistics

Table 3 reports the summary statistics. ROA and NIM of CBs are larger on average than IBs and t-statistic suggests that the means
are not equal at the 5% level. As it can be seen from Table 3, for half of variables, namely for non-interest margin, BIS capital
adequacy ratio, loan loss provision, operation costs, size, Lerner index, t-tests imply equal means for both CBs and IBs.
IBs have 25% more equity to asset ratio implying that they are better capitalized, which is consistent with Fatnassi et al. (2014),
Beck et al. (2013), and Sun et al. (2014). IBs have larger loan to asset, and risk weighted asset ratios. These results might be related to
profit-loss sharing structure and their limited usage of hedging and investment instruments due to Sharia-compliance considerations.

4.1.1. GMM estimation results


We first estimate the fixed effect regressions to compare our results with the previous literature.6 We then estimate the same
specifications with the System General Methods of Moments (GMM). We primarily base our conclusions on the GMM estimations
because the fixed-effect estimations eliminate time-persistent cross-section information and fail to solve a number of problems
common to panel data. For example, one of the most common problems in these fixed-effect regressions is the potential endogeneity
of the dependent variable. This issue urges researchers to be more cautious when evaluating the results on the determinants of
profitability. GMM estimator takes into account of potential endogeneity of the explanatory variables by utilizing the panel nature of
the data. Moreover, since we expect that NIM and ROA are substantially determined by their lagged values, we utilize the GMM

6
We do not report the fixed effect results due to the space considerations but are available upon request.

105
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 4
GMM Estimates: CBs NIM.

CBs Net Interest Margin Macro and Financial Financial Infrastructure Region Dummies

Coef t Coef t Coef t

Lagged NIM 0.345 2.69 ⁎⁎⁎ 0.326 2.44 ⁎⁎ 0.321 2.36 ⁎⁎


Non Interest Margin 0.056 0.42 0.048 0.36 0.046 0.35
Loan Loss Provision -0.052 -2.93 ⁎⁎⁎ -0.056 -3.42 ⁎⁎⁎ -0.057 -3.38 ⁎⁎⁎
Loan to Asset 0.012 1.5 0.012 1.43 0.011 1.42
Size 0.317 1.2 0.250 0.98 0.243 0.99
Operation Cost 0.457 2.45 ⁎⁎ 0.404 2.29 ⁎⁎ 0.437 2.27 ⁎⁎
Basel Capital Adequacy Ratio -0.010 -0.63 -0.009 -0.55 -0.009 -0.56
Bank Lerner Index 0.006 3.55 ⁎⁎⁎ 0.006 2.13 ⁎⁎ 0.005 2.1 ⁎⁎
Risk Weighted Asset to TA -0.001 -0.12 -0.003 -0.34 -0.005 -0.65
Foreign Ownership 1.522 2.11 ⁎⁎ 1.493 2.09 ⁎⁎ 1.402 1.87 ⁎
Inflation -0.047 -1.78 ⁎ -0.042 -1.71 ⁎ -0.044 -1.79 ⁎
GDP Growth -0.045 -3.38 ⁎⁎⁎ -0.031 -2.64 ⁎⁎⁎ -0.038 -3.32 ⁎⁎⁎
Regulator Quality -0.710 -3.28 ⁎⁎⁎ -0.774 -2.4 ⁎⁎ -0.445 -1.22
Interest Rate Volatility 0.175 1.27 0.127 1.05 0.129 0.9
Exchange Rate Volatility 0.000 -0.13 0.000 0.36 0.000 1.17
Ratio of Borrowers to Savers 0.002 1.45 0.002 0.99
Banking Service Accessibility 0.586 1.39 0.116 0.37
# of Banking Branches, POS and ATM Machines 0.189 0.74 0.008 0.03
Usage of self service banking channels -0.010 -0.98 0.017 0.73
Islamic Financial Development Indicator -0.006 -1.47 -0.010 -1.1
Crude Oil Prices Index 0.000 0.16 0.000 0.22
Agriculture Prices Index 0.001 0.14 -0.001 -0.16
Metal Minerals Prices Index 0.000 0.1 0.002 0.71
GCC Dummy 0.211 0.47
Malaysia Dummy -0.022 -0.03
UK Dummy -1.415 -1.1
AR1 0.011 0.016 0.015
AR2 0.373 0.453 0.551
Hansen 0.393 0.439 0.481
# of Instruments 223 228 231
# of Groups 353 352 352
Observation 1115 1110 1110

Note: Dependent variable: Net interest margin. The reported t-statistics are based on robust standard errors. AR order 1 (2) are tests for first (second)-order serial
correlation, asymptotically N(0, 1). These test the first-differenced residuals in the system GMM estimators. The Hansen test is a test of overidentification restrictions.
Under the null hypothesis, the test statistic is distributed as a chi-squared in the number of overidentifyng restrictions, p-values are presented. System GMM results are
two-step estimates. The two-step standard errors are computed in accordance to the Windmeijer (2005) finite-sample correction.

Significant at 10%.
⁎⁎
Significant at 5% level.
⁎⁎⁎
Significant at 1% level.

estimation that provides a solution to this problem. Note that since we have smaller sample size especially for IBs, using full in-
struments in estimations is impossible due to the instrument proliferation problem. To obtain robust results, one solution to this
instrument proliferation problem is to use “collapse” option in the GMM estimations for IBs (see, Roodman, 2009).
Tables 4 to 7 report the GMM estimations for NIM and ROA, respectively. As it can be observed on the tables, insignificant AR(2)
tests suggest that error terms don't have the second-order autocorrelation. Insignificant Hansen test of over-identifying restrictions
(with high p values) implies that the models are correctly specified, considering that there are no evidences of correlation between
instruments and errors. Given the discussion above, we conclude that our estimations are robust and consistent.

4.1.2. GMM Results for NIM


Table 4 reports the GMM estimations for NIM of CBs. The statistically significant and positive estimated coefficients on lagged
NIM clearly imply the persistency of NIM for CBs. Operation costs has significantly positive effects on NIM. CBs tend to reflect
operation costs to bank margins. Lerner index –as expected- positive significant meaning monopoly power increases NIM of CBs. LLP
also has a significant and negative effect on NIM. There is also evidence showing foreign banks in the OIC and the UK tend to have
larger bank margins. However, non interest margin, Basel capital adequacy ratio, risk weighted assets turn out to be insignificant in
the GMM estimates. Insignificance of capital adequacy ratios and risk weighted assets can be plausible because the latest crisis might
push all banks to have similar capital and risk level.
Regarding the effect of country-specific variables on NIM for CBs, inflation, regularity quality, and GDP growth are found to have
negative impacts on NIM. None of the interest and exchange rate volatility variables have significant relationship with NIM for CBs.
All financial inclusion and structural variables are insignificant for CBs, implying that the overall financial system accessibility and
alternative distribution channels don't have any significant effects on NIM for the CBs from sample countries. Price indexes for
commodities along with region dummies also don't also show any association with NIM for CBs. These results imply that, as far as CBs

106
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 5
GMM Estimations: IBs NIM.

IBs Net Interest Margin Macro and Financial Financial Infrastructure Region Dummies Islamic Modes of Finance

Coef t Coef t Coef t Coef t

Non Interest Margin -0.103 -1.15 -0.108 -1.05 -0.129 -1.17 -0.238 -1.61
Loan Loss Provision -0.022 -4.12 ⁎⁎⁎ -0.023 -4.56 ⁎⁎⁎ -0.023 -3.82 ⁎⁎⁎ -0.020 -2.1 ⁎⁎
Loan to Asset 0.013 1.22 0.008 0.68 0.013 1.26 0.017 1.48
Size 1.534 2.26 ⁎⁎ 1.570 1.81 ⁎ 1.597 1.74 ⁎ 1.696 2.23 ⁎⁎
Operation Cost 0.057 0.46 0.010 0.06 0.016 0.15 0.113 1.07
Equity to Asset/ BIS Cap 0.026 1.12 0.017 0.65 0.031 1.35 0.011 0.75
Bank Lerner Index 0.000 -0.98 0.000 -0.55 0.000 -0.65 0.000 -1.73 ⁎
Risk Weighted Asset to TA -0.017 -3.13 ⁎⁎⁎ -0.016 -2.62 ⁎⁎ -0.015 -2.02 ⁎⁎ 0.000 0.03
Foreign Ownership -0.501 -0.5 -0.642 -0.42 -0.280 -0.23 0.726 0.42
Inflation 0.057 1.79 ⁎ 0.051 1.74 ⁎ 0.021 0.53 0.037 1.64
GDP Growth 0.013 0.6 0.018 0.63 0.007 0.17 0.025 0.9
Regulator Quality -0.190 -0.64 0.063 0.2 0.579 0.65 0.784 1.33
Interest Rate Volatility 0.306 0.48 -0.184 -0.42 -0.281 -0.5 0.118 0.31
Exchange Rate Volatility -0.001 -0.29 0.001 1.25 0.000 0.02 -0.001 -0.95
Ratio of Borrowers to Savers -0.001 -0.19 -0.007 -1.1 -0.004 -0.79
Banking Service Accessibility -0.119 -0.25 -0.718 -0.76 0.084 0.09
# of Banking Branches, POS and ATM Machines -1.191 -1.26 -0.370 -0.34 -1.461 -1.8 ⁎
Usage of self service banking channels -0.011 -0.5 0.045 0.94 -0.012 -0.12
Islamic Financial Development Indicator -0.003 -0.27 0.015 0.27 -0.014 -0.59
Crude Oil Prices Index -0.019 -3.7 ⁎⁎⁎ -0.017 -2.37 ⁎⁎ -0.024 -3.56 ⁎⁎⁎
Agriculture Prices Index 0.025 2.72 ⁎⁎⁎ 0.025 2.77 ⁎⁎⁎ 0.031 2.2 ⁎⁎
Metal Minerals Prices Index -0.010 -1.52 -0.012 -1.74 ⁎ -0.011 -1.28
GCC Dummy -2.800 -2.82 ⁎⁎⁎ -0.612 -0.65
Malaysia Dummy -3.477 -1.07 -0.534 -0.25
UK Dummy -4.371 -0.89 -0.880 -0.21
Non Murabaha Asset Ratio 0.027 1.63
AR1 0.269 0.465 0.548 0.034
AR2 0.319 0.312 0.327 0.888
Hansen 0.761 0.577 0.483 0.511
# of Instruments 67 72 75 56
# of Groups 74 74 74 60
Observation 223 223 223 170

Note: Dependent variable: net interest margin. The reported t-statistics are based on robust standard errors. AR tests for first (second)-order serial correlation,
asymptotically N(0, 1). These test the first-differenced residuals in the system GMM estimators. The Hansen test is a test of overidentification restrictions. Under the
null hypothesis, the test statistic is distributed as a chi-squared in the number of overidentifyng restrictions, p-values are presented. System GMM results are two-step
estimates. The two-step standard errors are computed in accordance to the Windmeijer (2005) finite-sample correction. Collapse option in the GMM estimations for
IBs.

Significant at 10%.
⁎⁎
Significant at 5% level.
⁎⁎⁎
Significant at 1% level.

are considered, to a large extent, our results are in line with the existing literature using NIM as the measure of bank performance.
Table 5 reports the GMM estimations for NIM of IBs. Due to the insignificant estimates for lagged NIM, we exclude this variable
from the estimations.7 Unlike the estimates for CBs, NIM is not affected from their lagged values for IBs, implying that NIM does not
show any persistency for IBs. Lower market shares and higher rate of growth of overall Islamic banking sector can be explanations for
the insignificance of lagged values. For IBs, LLP is negative and significant in all cases and it is the only variable having same effects
as compared to CBs. Statistically significant and positive coefficients of bank size imply that bank scale matters for IBs. Risk weighted
assets have the significantly negative effect on NIM. Operation costs also have insignificant coefficients, implying that IBs could not
reflect their costs to their margins. Of the five macroeconomic variables, only inflation have significant and positive coefficient for
IBs. Even significance is 10%, there is evidence that IBs consider inflation rates for interest margin determination. Additionally,
similar to the results for CBs, financial infrastructure variables don't have any significant relationships with NIM for IBs.
Moreover, unlike the case for CBs, commodity prices are found to affect the determination of NIM for IBs. Statistically significant
coefficients indicate that higher crude oil prices reduce, higher agriculture prices raise NIM for IBs. The negative relationship between
oil prices and NIM can be explained by referring to the fact that increasing oil prices will expand available funds in most of the oil-
exporting OIC counties and thus reduce the margins charged by IBs. Similarly, positive effect of agriculture prices on NIM could be
inferred as the result of the fact that the sample countries are usually net importers of agriculture products. Note that NIM has
significant relationships with commodity prices but not with other macroeconomic variables. Of three regional dummies, GCC
dummy is only significant one, implying that IBs in the GCC countries operate with lower margins.

7
Estimation results are not reported due to space considerations. We have almost similar results except the much smaller sample sizes.

107
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 6
GMM Estimations: CBs ROA.

CBs ROA Macro and Financial Financial Infrastructure Region Dummies

Coef t Coef t Coef t

Lagged ROA 0.065 1.85 ⁎ 0.060 1.78 ⁎ 0.065 1.9 ⁎


Loan Loss Provision -0.494 -8.32 ⁎⁎⁎ -0.501 -7.86 ⁎⁎⁎ -0.510 -8.56 ⁎⁎⁎
Loan to Asset 0.007 0.67 0.002 0.2 0.004 0.41
Size 0.209 0.6 0.220 0.61 0.093 0.26
Operation Cost -0.115 -0.47 -0.161 -0.66 -0.126 -0.46
Equity to Asset / BIS TOTAL -0.010 -0.47 -0.014 -0.63 -0.015 -0.58
Bank Lerner Index 0.004 0.67 0.004 0.79 0.004 0.8
Risk Weighted Asset to TA -0.010 -0.93 -0.014 -1.16 -0.021 -1.34
Foreign Ownership 1.185 1.39 1.491 1.99 ⁎⁎ 1.479 1.46
Inflation -0.063 -1.95 ⁎ -0.051 -1.6 -0.047 -1.3
GDP Growth -0.031 -1.92 ⁎ -0.004 -0.22 -0.032 -1.88 ⁎
Regulator Quality -0.922 -4.15 ⁎⁎⁎ -1.607 -3.96 ⁎⁎⁎ -0.827 -2.35 ⁎⁎
Interest Rate Volatility 0.548 3.3 ⁎⁎⁎ 0.390 2.12 ⁎⁎ 0.345 1.62
Exchange Rate Volatility -0.001 -0.73 0.000 0.08 0.001 1.25
Ratio of Borrowers to Savers 0.005 2.41 ⁎⁎ 0.004 2.46 ⁎⁎
Banking Service Accessibility 0.909 2.19 ⁎⁎ -0.113 -0.26
# of Banking Branches, POS and ATM Machines 0.282 0.87 -0.084 -0.21
Usage of self service banking channels 0.023 1.73 ⁎ 0.088 2.97 ⁎⁎⁎
Islamic Financial Development Indicator 0.002 0.5 -0.001 -0.1
Crude Oil Prices Index 0.005 1.62 0.005 1.73 ⁎
Agriculture Prices Index -0.012 -1.43 -0.012 -1.59
Metal Minerals Prices Index 0.006 0.97 0.008 1.43
GCC Dummy 0.333 0.49
Malaysia Dummy -0.781 -0.83
UK Dummy -3.531 -2.02 ⁎⁎
AR1 0.01 0.014 0.013
AR2 0.34 0.313 0.317
Hansen 0.526 0.474 0.515
# of Instruments 202 207 210
# of Groups 351 350 350
Observation 1117 1112 1112

Note: Dependent variable: Return on Asset. The reported t-statistics are based on robust standard errors. AR tests for first (second)-order serial correlation, asymp-
totically N(0, 1). These test the first-differenced residuals in the system GMM estimators. The Hansen test is a test of overidentification restrictions. Under the null
hypothesis, the test statistic is distributed as a chi-squared in the number of overidentifyng restrictions, p-values are presented. System GMM results are two-step
estimates. The two-step standard errors are computed in accordance to the Windmeijer (2005) finite-sample correction.

Significant at 10%.
⁎⁎
Significant at 5% level.
⁎⁎⁎
Significant at 1% level.

The last column of Table 5 report that non-Murabahah asset ratios have no significant relationship with NIM for IBs. Non-
Murabahah assets might be related with non-interest revenue portion of the IBs. Islamic Finance development also level has no
significant effect on NIM.

4.1.3. GMM results for ROA


As far as ROA is considered as the measure of bank performance, GMM analysis for CBs results in fewer number of significant
variables as compared to NIM results. The results indicate that there is persistency in ROA for CBs, which is an expected observation
given earlier studies. However, the level of persistency is lower than the persistency level for NIM. LLP has also a significant and
negative coefficient for ROA. Similar to the estimations of NIM, foreign ownership is also positively linked to the ROA levels of CBs.
Unlike the case for NIM, operational costs are found to have no significant impact on ROA. This might be due to fact that most of the
operational costs are already reflected in the margins of CBs but cancelled out when calculating overall profitability. Other bank-
specific variables included within the analysis such as Loan to Asset, Size, Basel Capital Ratio, Bank Lerner Index, Risk Weighted Asset
are found to have non-significant coefficients.
Considering the macroeconomics and financial variables, it is observed that interest rate volatility has a positive impact on the
ROA of CBs. The level of the impact is higher as compared to the one for NIM indicating that CBs do not increase their margins with
interest rate volatility. They rather take measures with other instruments. CBs take good advantage of interest rate volatility to earn
income. Macroeconomic variables have similar impacts on ROA as with the case for NIM. Regulator quality is found to have a
negative impact on the profitability. As it is in the analysis of NIM, the coefficients of inflation and GDP growth are found to be
insignificant. Whereas the coefficient exchange rate volatility is insignificant, financial structure and financial inclusion turned to be
significantly related to the ROA of CBs. The analysis indicates that the ratio of borrowers to savers has a strong and robust positive
relation with ROA of CBs. So, increases in the proportion of the borrowers within a country is increasing the ROA of CBs. The usage of

108
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Table 7
GMM Estimations: IBs ROA.

IBs ROA Macro and Financial Financial Infrastructure Region Dummies Islamic Modes of Finance

Coef t Coef t Coef t Coef t

Loan Loss Provision -0.194 -12.2 ⁎⁎⁎ -0.194 -14.01 ⁎⁎⁎ -0.192 -12.57 ⁎⁎⁎ -0.187 -64.37 ⁎⁎⁎
Loan to Asset -0.014 -0.42 -0.027 -0.92 -0.008 -0.31 -0.026 -1.15
Size 0.851 0.68 0.604 0.25 1.452 0.5 -0.193 -0.13
Operation Cost -0.803 -1.27 -0.719 -1.19 -0.804 -1.03 -0.540 -1.97 ⁎
Equity to Asset -0.004 -0.11 -0.028 -0.42 0.004 0.05 -0.052 -1.57
Bank Lerner Index -0.001 -3.93 ⁎⁎⁎ -0.001 -3.95 ⁎⁎⁎ -0.001 -3.89 ⁎⁎⁎ -0.002 -3.61 ⁎⁎⁎
Risk Weighted Asset to TA -0.003 -0.16 -0.002 -0.09 0.002 0.11 0.008 0.5
Foreign Ownership 2.761 0.98 3.607 1.14 3.001 0.9 0.938 0.35
Inflation 0.008 0.12 -0.015 -0.26 -0.043 -0.62 -0.038 -0.86
GDP Growth 0.063 1.38 0.038 0.87 0.010 0.14 0.042 1.03
Regulator Quality -0.934 -2.15 ⁎⁎ -1.525 -1.2 -0.338 -0.22 1.719 1.55
Interest Rate Volatility -0.280 -0.38 -0.052 -0.13 0.137 0.26 -0.265 -0.66
Exchange Rate Volatility 0.001 0.24 0.000 0.1 0.001 0.39 0.000 0.38
Ratio of Borrowers to Savers -0.013 -1.23 -0.016 -1.11 -0.005 -0.46
Banking Service Accessibility -1.275 -1.24 -2.540 -1.34 -0.827 -1.24
# of Banking Branches, POS and ATM Machines 1.005 0.34 0.497 0.21 -1.224 -1.01
Usage of self service banking channels 0.083 1.14 0.191 1.97 ⁎ 0.222 2.76 ⁎⁎⁎
Islamic Financial Development Indicator -0.011 -0.45 0.028 0.36 -0.074 -1.88 ⁎
Crude Oil Prices Index -0.003 -0.21 -0.005 -0.31 -0.011 -1.14
Agriculture Prices Index 0.009 0.42 0.022 0.58 0.015 0.85
Metal Minerals Prices Index -0.006 -0.54 -0.015 -0.71 -0.003 -0.29
GCC Dummy -3.049 -1 -0.807 -0.39
Malaysia Dummy -6.098 -0.82 1.896 0.75
UK Dummy -7.950 -1.56 -12.439 -2.16 ⁎⁎
Non Murabaha Asset Ratio 0.034 2.08 ⁎⁎
AR1 0.172 0.205 0.156 0.123
AR2 0.501 0.547 0.476 0.923
Hansen 0.417 0.533 0.14 0.529
# of Instruments 60 65 68 52
# of Groups 74 74 74 61
Observation 226 226 226 174

Note: Dependent variable: Return on Asset. The reported t-statistics are based on robust standard errors. AR tests for first (second)-order serial correlation, asymp-
totically N(0, 1). These test the first-differenced residuals in the system GMM estimators. The Hansen test is a test of overidentification restrictions. Under the null
hypothesis, the test statistic is distributed as a chi-squared in the number of overidentifyng restrictions, p-values are presented. System GMM results are two-step
estimates. The two-step standard errors are computed in accordance to the Windmeijer (2005) finite-sample correction. Collapse option in the GMM estimations for
IBs.

Significant at 10%.
⁎⁎
Significant at 5% level.
⁎⁎⁎
Significant at 1% level.

self-service banking channels is another variable that is positively related to the ROA levels of CBs. In other words, the more people
use self-service banking channels such as internet-mobile banking and electronic payment the more ROA ratios the CBs have. This
result points out to the importance of alternative distribution channels as respect with the performance of CBs. Furthermore, Banking
Service Coverage ratio which stands for the proportion of population having bank account has a positive impact on the level of ROA.
As respect with the relation of commodity prices and the ROA of CBs, it is observed that just the oil price indexes have a
significantly positive impact on the ROA levels, while other price indexes are found to be insignificant. Even if the mostly oil
exporting GCC region highly dominates the Islamic banking industry across OIC countries, as far as the OIC countries without oil
resources are considered, since any increase in oil prices increase their funding needs, the spreads of CBs will enlarge. Thus the ROA
levels of CBs will also rise in line with the increase in oil prices. As per the relationship of ROA levels of CBs to the dummy variables
standing for various regions, it is concluded that the estimated coefficient on UK dummy is significant and negative implying the
relatively more intense effect of global financial crisis on the CBs of the UK.
As far as the determinants of the ROA levels for IBs are considered, the number of significant explanatory variables is less than the
number of those for NIM, which is similar to the analysis for CBs. The non-persistency of ROA does also hold as in case of NIM for IBs.
The LLP and Lerner Index are observed to have negative and significant coefficients. Unlike the results for CBs, the significantly
negative coefficients on Lerner index imply that monopoly power does not generate more profit for IBs. This result may be due to the
lack of differentiated products and thus low market shares of IBs. In addition, this result may be explained by referring to the profit-
loss-sharing structure which might balance the income that would be generated through additional market power. The analysis
indicated that all macroeconomic variables employed within model are insignificant in explaining the ROA levels of IBs. In addition,
unlike the analysis of NIM, the price indexes are observed to have no significant relationship with the ROA levels of IBs. As in case of
CBs, we see positive impact of self-service banking channels on the ROA levels of IBs. as Again similar to CBs, the ROA of IBs are
found to be significantly less for the UK region which can be again explained by referring to the rather bigger impact of the global

109
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

crisis to the UK market. The insignificant coefficient of the level of Islamic finance development implies that increasing product
variety is more important to improve the profitability of IBs than the Islamic banking development level in a country. Finally, an
important finding of the analysis is that unlike NIM, the ROA of IBs is positively affected by the increased usage of non-Murabahah
financing.8

5. Conclusion

The remarkable increase in Islamic finance practice, especially in Islamic banking as the flagship sector of the industry, brought
about questions regarding the determinants of the profitability of Islamic banks. Although, in theory, Islamic banking relies upon the
profit-loss sharing structure of finance, in practice there is an ongoing argument that Islamic banks usually mimic conventional banks.
Extensive usage of Murabahah type of financing supports those arguments by arising the need of a comparative analysis of the
determinants of IBs and CBs. The main purpose of the study is to try to find a reasonable answer to this final question by using a
dynamic panel data approach for a sample of 74 IBs and 354 CBs in the OIC countries and the U.K. for the period between 2007 and
2013. Our study empirically analyzes and compares the profitability of Islamic Banks and conventional banks by employing two
different measures of profitability namely the net interest margin and the return on asset and a set of bank-level and country-level
explanatory variables.
Our GMM estimations for the CBs are mostly consistent with the previous findings in the literature. However, the results for IBs
results considerably different from the findings as most of the previous studies that mostly utilize the fixed effect methods. Thus, we
can safely claim that empirical conclusions of the previous studies are very much sensitive to the methods used. Moreover, almost all
determinants of profitability for CBs and IBs are different implying that profitability of IBs relies on the different dynamics than those
of CBs. CBs profitability ratios are persistent. In addition, CBs tend to reflect their operational costs to their customers by increased
margins. In case of CBs, foreign ownership is related with higher margins whereas the impact of foreign ownership turned to be
insignificant for IBs. Except some commodity prices, the profitability of IBs are found to be insignificantly related to the most of
macroeconomic variables used in the analysis. However, the profitability of CBs mostly responds negatively to most of the macro-
economic and financial variables except the case for the interest rate volatility. The profitability of CBs is found to increase together
with interest rate volatility due to the usage of alternative instruments to manage the associated risks.
In addition to the commonly used explanatory variables to estimate the dynamics of bank profitability, the study also deploys
several new or rarely used variables into the analysis such as variables measuring the level of financial inclusion and penetration, self-
service banking, and Islamic finance development of the countries. Better and more diffused financial infrastructure raises the
profitability of CBs but we see no relation with IBs. However, the usage of self-service banking channels increases profitability of the
both types of banks, implying that the importance of self-service banking for the higher profitability.
Furthermore, the study also makes a reference to the structure of financing used by Islamic banks to investigate the impact of
product structures which promotes more risk-sharing on the overall profitability. For this purpose, the ratio of non-Murabahah assets
is integrated into the model. The study concludes that the level of non-Murabahah assets positively impact the level of ROA for IBs.
This result has the implication that more usage of financing structures based more on the concept of risk sharing will positively linked
with the performance of IBs.
We believe that our study makes some important contributions to the literature. First of all, the study employs a GMM approach as
compared to the fixed effect models widespread within the literature. The results for IBs suggest that the analyses focusing on the
determinants of Islamic bank performance might be sensitive to the model selection. Secondly, the study reached the observation that
the dynamics affecting the performance of IBs may be different that those of CBs providing supporting evidence for the argument that
IBs not necessarily mimic the practice of CBs. A third conclusion is related to the usage of alternative self-banking tools. There is room
for IBs to enhance their profitability via providing alternative products and channels for their customers. A last but not the least
conclusion is related to the IBs' choice of financing structures provided to their clients. The study empirically supports the view that
more usage of non-Murabahah structures is expected to contribute positively to the overall performance of Islamic banks.

References

www.oanda.com, Accessed date: 7 January 2016.


www.tradingeconomics.com, Accessed date: 7 January 2016.
Angbazo, L., 1997. Commercial bank net interest margins, default risk, interest-rate risk, and off-balance sheet banking. J. Bank. Financ. 21 (1), 55–87.
Athanasoglou, P.P., Brissimis, S., Panayiotis, P., Delis, M.D., 2008. Bank-specific, industry-specific and macroeconomic determinants of bank profitability. J. Int.
Financ. Mark. Inst. Money 121–136.
Azmat, S., Skully, M., Brown, K., 2015. Can Islamic banking ever become Islamic? Pac. Basin Financ. J Available at. http://linkinghub.elsevier.com/retrieve/pii/
S0927538X15000359.
Bashir, A.H.M., 2003. Determinants of profitability in Islamic banks: some evidence from the Middle East. Islam. Econ. Stud. 11 (1), 31–57.
Beck, T., Demirgüç-Kunt, A., Merrouche, O., 2013. Islamic vs. conventional banking: business model, efficiency and stability. J. Bank. Financ. 37 (2), 433–447.
Bitar, M., 2014. Banking regulation, Stability and Efficiency of Islamic banks: What works best? A comparison with conventional banks. PhD Dissertation submitted in
L'unıversıté De Grenoble, France. Accessible at. www.theses.fr/2014GRENG017.pdf.

8
We also run the regression for CBs without the UK data (not reported here). We have qualitatively similar results. However, there are some differences especially
for NIM. Both inflation and Islamic Financial Development Indicator have now statistically significant and negative effect on ROA. Regarding NIM, loan to asset and
Islamic Financial Development (inflation and regulatory quality) Indicator are now statistically significant (insignificant). One can safely conclude that Islamic
financial development produces a competitive force that reduces the interest margins and profitability of CBs in the OIC countries.

110
H. Yanikkaya et al. Pacific-Basin Finance Journal 48 (2018) 99–111

Bourke, P., 1989. Concentration and other determinants of bank profitability in Europe, North America and Australia. J. Bank. Financ. 13, 65–79.
Brock, P.L., Suarez, L.R., 2000. Understanding the behavior of bank spreads in Latin America. J. Dev. Econ. 63 (1), 113–134.
Chapra, U., 1992. Islam and the Economic Challenge. The Islamic Foundation and International Institute of Islamic Thought.
Demirguc-Kunt, A., Huizinga, H., 1998. Determinants of commercial bank interest margins and profitability: some international evidence. World Bank Econ. Rev. 13,
379–408.
Demirguc-Kunt, A., Huizinga, H., 2000. Financial Structure and Bank Profitability. Policy ResearchWorking Paper Series 2430 The World Bank.
Dietrich, A., Wanzenried, G., 2014. The determinants of commercial banking profitability in low-, middle-, and high-income countries. Q. Rev. Econ. Finance 54 (3),
337–354. Available at. https://doi.org/10.1016/j.qref.2014.03.001.
El-Hawary, D., Grais, W., Iqbal, Z., 2007. Diversity in the regulation of Islamic Financial Institutions. Q. Rev. Econ. Finance 46 (5), 778–800.
Fatnassi, I., Hasnaoui, H., Ftiti, Z., 2014. The impact of bank capital on profitability and risk in GCC countries: Islamic vs. conventional. IPAG Bus. Sch. Work. Pap. 32
(1), 251–281.
Goddard, J.A., Molyneux, P., Wilson, J.O., 2004. Dynamics of growth and profitability in banking. J. Money, Credit, Bank. 36 (6), 1069–1090.
Godlewski, C.J., Turk, R., Weill, L., 2014. Do the Type of Sukuk and Choice of Shari'a Scholar Matter? International Monetary Fund. Working Paper No:14/147.
Haron, S., 2004. Determinants of Islamic bank profitability. Glob. J. Financ. Econ. 1 (1), 11–33.
Hasan, M., Dridi, J., 2010. The Effects of the Global Crisis on Islamic and Conventional Banks: A Comparative Study. IMF Working Paper, No: 10/201.
Hassan, M.K., Bashir, A.H.M., 2003. Determinants of Islamic Banking Profitability. 10th ERF Annual Conference, Moroccopp. 16–18.
Hawtrey, K., Liang, H., 2008. Bank interest margins in OECD countries. North Am. J. Econ. Financ. 19 (3), 249–260.
Ho, T.S.Y., Saunders, A., 1981. The determinants of bank interest margins: theory and empirical evidence. J. Financ. Quant. Anal. 16 (04), 581–600.
ICD Thomson Reuters, 2015. Islamic Finance Development Report: Global Transformation.
IFSB, 2016. Islamic Finance Services Industry Stability Report, Malaysia.
Imam, P., Kpodar, K., 2015. Is Islamic Banking Good for Growth? International Monetary Fund. Working Paper WP/15/81.
Iqbal, Z., Mirakhor, A., 2011. An Introduction to Islamic Finance: Theory and Practice, 2nd Ed. John Willey & Sons (Asia), Singapore.
Islamic Banks and Financial Institutions Information (IBIS) Available at: http://www.islamicfinancedata.org/, Accessed date: 15 June 2015.
Kasman, A., Tunc, G., Vardar, G., Okan, B., 2010. Consolidation and commercial bank net interest margins: evidence from the old and new European Union members
and candidate countries. Econ. Model. 27 (3), 648–655.
Khan, F., 2010. How “Islamic” is Islamic banking? J. Econ. Behav. Organ. 76 (3), 805–820.
Khan, S., Mirakhor, A., 1989. Islamic Banking: Experiences in the Islamic Republic of Iran and Pakistan. International Monetary Fund, IMF Working Paper No. WP/
89/12.
Khediri, K. Ben, Charfeddine, L., Youssef, S. Ben, 2015. Islamic versus conventional banks in the GCC countries: a comparative study using classification techniques.
Res. Int. Bus. Financ. 33, 75–98.
López-Espinosa, G., Moreno, A., Pérez de Gracia, F., 2011. Banks' net interest margin in the 2000s: a macro-accounting international perspective. J. Int. Money Financ.
30 (6), 1214–1233.
Masood, O., Ashraf, M., 2012. Bank-specific and macroeconomic profitability determinants of Islamic banks: the case of different countries. Qual. Res. Finan. Markets.
4 (2/3), 255–268.
Maudos, J., Solís, L., 2009. The determinants of net interest income in the Mexican banking system: an integrated model. J. Bank. Financ. 33 (10), 1920–1931.
Mejía, A.L., Aljabrin, S., Awad, R., Norat, M., Song, I., 2014. Regulation and Supervision of Islamic Banks. International Monetary Fund. Working Paper No:14/219.
Molyneux, P., Thornton, J., 1992. Determinants of European bank profitability: a note. J. Bank. Financ. 16 (6), 1173–1178.
Noor, M.A.N.M., Ahmad, N.H.B., 2011. Relationship between Islamic banking profitability and determinants of efficiency. IUP J. Manag. Econ. 9 (3), 43.
Olson, D., Zoubi, T.A., 2008. Using accounting ratios to distinguish between Islamic and conventional banks in the GCC region. Int. J. Account. 43 (1), 45–65.
Pasiouras, F., Kosmidou, K., 2007. Factors influencing the profitability of domestic and foreign commercial banks in the European Union. Res. Int. Bus. Financ. 21 (2),
222–237.
Roodman, D., 2009. A note on the theme of too many instruments. Oxf. Bull. Econ. Stat. 71 (1), 135–158.
Saunders, A., Schumacher, L., 2000. The determinants of bank interest rate margins: an international study. J. Int. Money Financ. 19 (6), 813–832.
Short, B.K., 1979. The relation between commercial bank profit rates and banking concentration in Canada, Western Europe and Japan. J. Bank. Financ. 3, 209–219.
Solé, J., 2007. Introducing Islamic Banks into Conventional Banking Systems. International Monetary Fund. Working Paper No:07/175.
Song, I., Oosthuizen, C., 2014. Islamic banking regulation and supervision: survey results and challenges. International Monetary Fund. Working Paper No 14 (220).
Sun, P.H., Hassan, M.K., Hassan, T., Ramadilli, S.M., 2014. The assets and liabilities gap management of conventional and Islamic banks in the organization of Islamic
cooperation (OIC) countries. Appl. Financ. Econ. 24 (5), 333–346.
Sun, P.H., Mohamad, S., Ariff, M., 2017. Determinants driving bank performance: a comparison of two types of banks in the OIC. Pac. Basin Financ. J. 42, 193–203.
The World Bank Global Findex. Available at: http://www.worldbank.org/en/programs/globalfindex.
The World Bank World Development Indicators. Available at: http://data.worldbank.org/indicator.
Valverde, S.C., Fernández, F.R., 2007. The determinants of bank margins in European banking. J. Bank. Financ. 31 (7), 2043–2063.
Windmeijer, F., 2005. A finite sample correction for the variance of linear efficient two-step GMM estimators. J. Econ. 126, 25–51.
Zeitun, R., 2012. Determinants of Islamic and conventional banks performance in GCC countries using panel data analysis. Glob. Econ. Financ. J. 5 (1), 53–72.

111

Potrebbero piacerti anche