Analysis on profitability and stability between Islamic banks and
conventional banks in the Gulf Cooperation Council
Abstract
In this thesis an analysis on profitability and stability between Islamic banks and conventional banks in the Gulf Cooperation Council from 2008 to 2017 is performed to determine if there is a difference between Islamic banks and conventional banks. Eight regressions will be conducted with return on assets, return on equity, z-score and non-performing loans as dependent variables. Two samples are used. The first sample contains the two biggest Islamic banks per country and two biggest
conventional banks if data was available. The second sample contains every fully Islamic bank and every fully conventional bank available with at least five observations. The results conclude that there isn’t a significant difference in return on assets and z-score. This could be because Islamic banking products have almost the same characteristics as conventional banking products. There is a significant difference in return on equity and non-performing loans. Islamic banks had a lower return on equity and a higher percentage of non-performing loans. This could be due to bad investments or because borrowers are less efficient with the money they borrowed.
Bachelor’s Thesis
Ward Rentenaar (11013117) 26 June 2018
2
Statement of originality
This document is written by Ward Rentenaar who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of contents
1. Introduction 4
2. Literature 7
2.1 Empirical results about the profitability of Islamic banks and conventional banks 7 2.2 Empirical results about the stability of Islamic banks and conventional banks 9
3. Methodology 11
4. Data 13
4.1 Sample one: Two biggest banks per country 13
4.2 Sample two: Every available bank 18
5. Results 24
5.1 Regression analysis using sample one 24
5.1.1 Regression with return on assets as dependent variable 24 5.1.2 Regression with return on equity as dependent variable 25 5.1.3 Regression with z-score as dependent variable 26 5.1.4 Regression with non-performing loans as dependent variable 27
5.2 Regression analysis using sample two 28
5.2.1 Regression with return on assets as dependent variable 28 5.2.2 Regression with return on equity as dependent variable 29 5.2.3 Regression with z-score as dependent variable 30 5.2.4 Regression with non-performing loans as dependent variable 31
5.3 Explanation for the obtained results 31
6. Conclusion and discussion 33
4
1. Introduction
Islamic financial institutions are expanding for over thirty years right now. First Islamic institutions started around the 1970’s. Nowadays, there are more than 300 financial institutions which conduct business according to the Sharia in more than 75 countries (El Qorchi, 2005). The Sharia is the Islamic Law, Islamic financial institutions have to comply with the Sharia and this is also why the Islamic financial system was established (Siddiqi, 2006).
During the financial crisis in 2008 some large banks went bankrupt. There was a lot of criticism on the banking sector those days. The Islamic banking sector got more attention during that time, there has been done much research due to the increased popularity.
Islamic banks are mainly operating in the Middle East, Southeast Asia and South Asia, but during the last few years Islamic banks also started conducting business in Europe, Britain even has an Islamic bank called Al-Rayan Bank. According to Mohamed and Goni (2017) assets in the Islamic finance industry grew to 2,2 trillion USD in 2016 and are expected to grow with an average
percentage of 9,5% per year to 3,8 trillion USD in 2022, Figure 1 shows this growth. In the GCC region there were 101 Islamic banks or Islamic windows in 2016, these banks had a 37% share in total banking assets and total global Islamic banking assets was estimated to be 5% of total global banking assets (Mohamed & Goni, 2017).
Figure 1: Global Islamic Financial Assets during 2012-2016 and expected growth to 2022 (Mohamed & Goni, 2017)
0 500 1000 1500 2000 2500 3000 3500 4000 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Total Assets Year
Global Islamic Financial Assets during 2012-2016 and
expected growth to 2022
One of the main differences between Islamic banking and conventional banking is the payment of interest (riba). Islamic banks are not allowed to pay and receive interest. Gambling or speculation (qimar), uncertainty (gharar) with respect to the transaction and investing in certain industries are also forbidden (haram), for example the weapon industry, but also the drugs, alcohol and pork industry are not allowed industries to invest in (Beck & Demirgüç-Kunt & Merrouche, 2013).
Nowadays the Islamic financial industry created many products which comply with the Sharia. According to Beck et al. (2013), most contracts are based on the profit- and loss-sharing principle. A mudaraba contract is one of the main contracts, in which the profits between the bank and their customers are shared using a specific ratio, losses on the other side, are only carried by the bank (Dar & Presley, 2000). Another sort of contract is the musharaka contract, the bank is not the only investor in this case. The profit- and loss-sharing principle still applies. The profits are shared between all investors using a ratio they agreed up on up front, and losses are shared between all investors according to the amount they invested (Ariff, 1988). Furthermore, there are also products that are not based on profit- and loss-sharing. One of the most common is the murabaha contract, this is a contract in which the bank buys an asset which is requested by the client and then sells it to the client with a fee who will pay it back in the future (Ariff, 1988). Because the bank bought the asset, it doesn’t make a return on just money lending (Beck et al., 2013). Another kind of leasing contract is the ijarah contract, this is almost the same as the contract just mentioned, except that in this contract the bank remains owner of the asset and rents it to the client with a fee (Beck et al., 2013).
In this thesis a comparative analysis is conducted between the profitability and stability of Islamic and conventional banks in the GCC region from 2008 up to and including 2017. Two samples are used. The first sample contains the two biggest Islamic banks per country to represent the whole GCC region and the two biggest conventional banks if there were two fully conventional banks per country. The second sample contains every fully Islamic bank and every fully conventional bank available on Datastream with five or more observations from the last ten years. Eight regressions are conducted, these regressions include a dummy variable for Islamic banks to distinguish the
difference between Islamic banks and conventional banks. Also, a few control variables are added to the regression to correct for the differences in determinants of profitability and stability. With the results of this regression it is possible to answer the following question: Did Islamic banks in the Gulf Cooperation Council perform better with respect to profitability and stability than conventional banks during the last ten years?
6 In the second section a literature review of existing papers about Islamic banks and
conventional banks relating to this thesis will be discussed. Results from previous research will also be discussed. In the third section the methodology will be discussed, this will contain the regression equation and information about the variables which are used. The fourth section discusses how the data was obtained and provides a summary of the statistics of the variables. In section five the results of the regression and the interpretation of these results will be discussed. The final section contains the conclusion and discussion.
2. Literature
In this section the existing literature on profitability and stability of Islamic banks and conventional banks is discussed. The first subsection contains the literature with respect to profitability and the second subsection contains the literature with respect to stability.
2.1 Empirical results about the profitability of Islamic banks and conventional banks There already are some papers written with empirical results about the difference in profitability between Islamic and conventional banks, these papers will be discussed in this subsection.
Latif, Abbas, Akram, Manzoor and Ahmad (2016) compared the performance of five Islamic banks and five conventional banks in Pakistan from 2006 to 2010. Their study included 12 financial ratios. They found there wasn’t much difference in the return on assets, return on equity and profit expense ratio between the two groups. They did find that Islamic banks are less risky and more solvent than conventional banks. Also, Islamic banks proved to be better in most of the liquidity ratios.
Rosly and Abu Bakar (2003) compared the performance between Islamic and conventional banks in Malaysia from 1996 to 1999. Islamic banks had a statistically significant higher return on assets than conventional banks. They concluded this is because the Islamic banks are subsidiaries from conventional banks, therefore the conventional banks carried most of the operating expenses. This resulted in a higher return on assets for Islamic banks.
Samad (2004) conducted research on the performance of Islamic and conventional banks in Bahrain from 1991 to 2001. The average return on assets was higher for Islamic banks, but they had a lower return on equity. However, these results weren’t statistically significant. Another finding was that credit performance of Islamic banks was superior. Samad (2004) concluded this was because Islamic banks hold more equity per capita. The loan to deposit ratio was significantly higher for Islamic banks. Therefore, Islamic banks are more liquid and have less exposure to liquidity risk.
Siraj and Pillai (2012) examined the performance of Islamic and conventional banks in the GCC region from 2005 to 2010. They chose six Islamic and six conventional banks and reviewed their performance with several financial ratios such as operating profit ratio, net profit ratio, return on assets, return on share capital and return on equity. They found that Islamic banks had a higher return on assets, but this wasn’t significant. The return on assets reduced from 2005 to 2010 so they concluded there needs to be research on the long run. They found the same results when comparing the return on equity.
8 Olson and Zoubi (2008) researched if Islamic and conventional banks can be distinguished using accounting ratios in the GCC region. One of their findings was that Islamic banks had a higher mean return on assets than conventional banks during 2000 to 2005, but this finding wasn’t statistically significant.
Loghod (2005) compared Islamic and conventional banks in the GCC region from 2000 to 2005 using a logit model. While comparing several profitability ratios he found that there were no significant differences between Islamic and conventional banks with respect to profitability. The research did conclude that Islamic banks had higher averages in each country in the GCC region except in the United Arab Emirates. He also found that there were significant differences between the liquidity ratios of every country in the GCC region except Kuwait. Islamic banks had higher cash to assets and cash to deposits ratio when compared with conventional banks and industry averages.
Islamic banks gained a lot of attention during the financial crisis, because they seemed less affected by the crisis. Parashar and Venkatesh (2010) confirmed this statement. They selected six Islamic banks and six conventional banks from the GCC region. They concluded that Islamic banks had significantly higher return on average assets during and before the crisis. Therefore, it can be concluded that Islamic banks were less affected by the financial crisis. Beck et al. (2013) also supported this statement.
Table 1 provides a summary about the results of the researches that have been discussed.
Country Study Variable Sample period More profitable
Pakistan Latif, Abbas, Akram,
Manzoor and Ahmad (2016)
ROA ROE
2006-2010 No difference
Malaysia Rosly and Abu Bakar (2003) ROA 1996-1999 Islamic banks
Bahrain Samad (2004) ROA
ROE
1991-2001 Islamic banks (not significant) GCC Siraj and Pillai (2012) ROA
ROE
2005-2010 Islamic banks (not significant)
GCC Olson and Zoubi (2008) ROA
ROE
2000-2005 Islamic banks (not significant)
GCC Loghod (2005) ROA
ROE
2000-2005 No difference
GCC Parashar and Venkatesh (2010)
ROA ROE
2006-2009 Islamic banks
2.2 Empirical results about the stability of Islamic banks and conventional banks
There has already been done some research about the stability of Islamic banks in
comparison to conventional banks. The results of these papers will be discussed in this subsection. Čihák and Hesse (2010) performed a cross-country empirical analysis of Islamic banks for 1993 to 2004. They included Islamic banks from The Middle East, Asia and Africa. Their research concluded that small Islamic banks were financially stronger than small conventional banks, however, large Islamic banks proved to be less strong than large conventional banks. Finally, they found that small Islamic banks were stronger than large Islamic banks.
Beck et al. (2013) also compared the stability of Islamic and conventional banks and found that Islamic banks were better capitalized and had a higher asset quality over the period 2005 to 2009. Hereby concluding Islamic banks were more stable and had a lower chance of going bankrupt.
Rajhi and Hassairi (2013) performed an analysis on Islamic and conventional banks from 16 countries during 2000 to 2008. They calculated the z-score and concluded that Islamic banks had a higher z-score than conventional banks and thus were more stable, except for small Islamic banks. This contradicts the research from Čihák and Hesse (2010) because they did find small conventional banks were financially stronger.
Tabash and Dhankar (2014) assessed the performance of Islamic banks during the financial crisis in 2008. Data from banks in Saudi Arabia from 2005 to 2010 was used to evaluate the stability of Islamic banks. The results conclude that Islamic banks are less exposed to liquidity risks. Another conclusion was that the Islamic banking sector has a large capacity to absorb financial shocks, therefore enjoying a higher stability than conventional banks.
Boumediene and Caby (2009) also performed an empirical analysis on the stability of Islamic banks during the financial crisis of 2008. Their research concluded that the volatility of Islamic banks increased during the crisis period, however not as much as conventional banks. Therefore,
concluding that Islamic banks are better at absorbing financial shocks.
10
Country Study Variable Sample period More stable
21 countries, Middle East, Asia and Africa
Čihák and Hesse (2010)
z-score 1993-2004 Small Islamic
banks > small conventional banks Large Islamic banks < large conventional banks Small Islamic banks > large Islamic banks 22 countries, mainly Middle East, Asia and Africa
Beck & Demirgüç-Kunt &
Merrouche (2013)
z-score 2005-2009 Islamic banks
16 countries, Middle East, North Africa and South East Asia
Rajhi and Hassairi (2013)
z-score 2000-2008 Islamic banks
10 countries, mainly Middle East, Asia and Africa Boumediene and Caby (2009) Volatility of shares 2005-2009 Islamic banks
Table 2: Summary of the existing literature on the stability of Islamic banks and conventional banks
There isn’t one main conclusion based on the results of the papers just discussed. Some of the research that has been done found a significant higher return on assets while others found no significant difference. This was the same for research with respect to stability. However, because research suggests that Islamic banks are better at absorbing shocks and therefore less affected by the recent financial crisis of 2008, it can be expected that the return on assets and the z-score will be higher for Islamic banks.
3. Methodology
To determine if there is a significant difference between the profitability and stability of Islamic banks and conventional banks, the following regression has been set up:
!"= $ + &'()**!+,-.*"+ &/ln( 456.- 7,,86,") + &:(;<.6=5"+ &>(8?5,=6<.6=5" + &@A?8<.6=BC8D?8B,8<.6=5"+ E"
This regression is run eight times since there are two samples and four dependent variables. To determine the difference in profitability, the return on assets (ROA) and return on equity (ROE) are used as dependent variables, the return on assets is calculated by dividing the net profit after tax by total assets. Return on assets is a measure often used to evaluate bank performance. According to Petersen & Schoeman (2008) the return on assets is an indicator to assess the efficiency of a bank. The return on equity is calculated by dividing the net profit after tax by total equity. It’s often used as a measure for profitability. This can also be seen from the literatures that has been discussed in section two, almost every research included the return on equity.
The z-score and non-performing loans (NPL) are used as dependent variables to determine the difference in stability. The z-score is used as a measure for stability and is widely used in recent literature according to Beck et al. (2013). The z-score is calculated by dividing the return on assets and the capital asset ratio (CAR) of a bank by the standard deviation of the return on assets of the bank. The performing loans are loans which aren’t paid back by the bank’s customers. The non-performing loans are measured as a percentage of total loans.
To determine the difference between Islamic and conventional banks, a dummy variable is included in the equation. This variable will be equal to one of it’s an Islamic bank and zero otherwise. When analyzing the regression results, it can be tested if the coefficient of this variable is different from zero, hereby determining if there’s a difference between the two groups of banks.
The regression also includes some control variables. To control for the size of the bank, the control variable total assets is included. The natural logarithm of total assets is chosen because the distribution is skewed. The outliers will be smaller by taking the natural logarithm. A positive relation between total assets and profitability is expected because of economies of scale. A negative relation is expected between the total assets and stability.
Another control variable is the debt/equity ratio. This is calculated by dividing the total debt by common equity. This ratio measures the leverage of the bank, it explains how much debt a bank is using to fund their assets relative to the value of the common equity. A negative relation between the debt/equity ratio and the profitability is expected because an increase in debt reduces the profitability. A negative relation between the debt/equity ratio and the stability is also expected, an increase in debt will increase the volatility of the bank and therefore reduce the stability.
12 The deposit ratio is also included as a control variable. This is calculated by dividing the total loans by the total deposits. A high ratio means that the bank may be illiquid when a lot of money is withdrawn at the same time. A low ratio means that the bank maybe could use their money more efficient and could use it to get a higher return. If a bank has a high deposit ratio it lends out more money, therefore it can make higher returns, thus a positive relation is expected between the deposit ratio and the profitability. A negative relation is expected between the deposit ratio and the stability.
Another control variable is the operating expense ratio (OER). This is defined as the
operating expenses divided by the operating income. This measures how much income is generated with each unit of costs. A negative relation is expected between the operating expense ratio and the profitability because a higher operating expense ratio implies more costs, hereby reducing the profitability. A negative relation between the operating expense ratio and the stability is also expected since an increase in expenses reduces the stability.
The mathematical formulas are as follows: FA7 = G86 +BH5*8 ∗ (1 − 4.D F.68) 456.- 7,,86, ∗ 100% FA; =G86 +BH5*8 ∗ (1 − 4.D F.68) 456.- ;N)=6! ∗ 100% O − ,H5<8 =FA7 + ( ;N)=6! 7,,86,) P(FA7) (8Q6 ;N)=6! <.6=5 =(R5BC 48<* (8Q6 + Sℎ5<6 48<* (8Q6 & V)<<8B6 W5<6=5B 5X R5BC 48<* (8Q6) V5**5B ;N)=6! ∗ 100% (8?5,=6 <.6=5 = 456.- R5.B, 456.- (8?5,=6,∗ 100% A?8<.6=BC 8D?8B,8 <.6=5 = A?8<.6=BC 8D?8B,8, A?8<.6=BC =BH5*8
4. Data
The data needed for this research is obtained from Datastream. Data is obtained for the period 2008-2017, hereby covering the last ten years prior to the date this thesis is written. In the GCC region exist a lot of banks which provide both Islamic and conventional banking, these banks aren’t used in the samples which is why data is scarce and therefore not every country is included in the samples with a conventional bank. Two samples are used in this thesis. The first sample contains the two biggest Islamic banks per country to represent the whole GCC region and the two biggest conventional banks if there were two fully conventional banks per country. Oman is excluded from both samples, since the Islamic banking sector is still in the early stages. Data before 2014 was scarce or had too many missing observations.
The second sample contains every fully Islamic bank and every fully conventional bank available on Datastream with five or more observations from the last ten years.
4.1 Sample one: Two biggest banks per country
The banks included in this sample are shown in Table 3.
Country
Islamic bank
Conventional bank
Bahrain Al Baraka Banking Group
Ithmaar Bank
Arab Banking Corporation Bank of Bahrain and Kuwait
Kuwait Ahli United Bank Kuwait
Kuwait Finance House
Gulf Bank of Kuwait National Bank of Kuwait
Qatar Qatar Islamic Bank
Masraf Al Rayan
Commercial Bank Qatar Al Khalij Commercial Bank
Saudi Arabia Al Rajhi Bank
Alinma Bank
United Arab Emirates Abu Dhabi Islamic Bank Commercial Bank of Dubai
Table 3: Overview of the represented Islamic banks and conventional banks in sample one
Table 4 shows a summary of the statistics during the period 2008-2017 with Islamic banks and conventional banks separated, Table 5 shows a summary of the statistics of both groups of banks. Figures 2 to 5 show the mean of the dependent variables per year for Islamic banks and conventional banks.
14
Variable
Obs
Mean
Std. Dev.
Min
Max
Obs
Mean
Std. Dev.
Min
Max
ROA
69
1,661304
1,639218
-4,48
6,8
58
1,271897
1,443973
-6,87
3,72
ROE
97
11,36557
9,553157
-33,03
27,91
58
6,427931
21,11952
-136,02
21,01
zscore
69
3,582667
2,509397 -2,248334
12,34628
58
3,598962
3,690506 -2,866959
12,40573
NPL
97
5,143402
4,970498
0
21,95
60
5,648167
5,955515
,34
28,59
TA
99
4,39e+07
3,98e+07
1963490
1,50e+08
60
2,83e+07
3,55e+07
2165088
1,38e+08
lnTA
99
16,93705
1,351278
14,49023
18,82857
60
16,47132
1,185621
14,58797
18,74601
TD_CE
99
150,7004
91,90676
0
521,27
60
272,78
300,5139
56,36
2402,87
TL_TD
99
126,8181
96,01031
67,89
997,42
60
129,0252
23,54938
77,39
209,68
OE
99
1853420
2109504
75254
8872704
60
1027968
1416149
90015
6237976
OI
99
1401047
2145587
-96773
9084587
60
339553,2
475358,7
-359516
1706672
OER
99
2,081634
17,88565
-80,8905
130,3812
60
3,619319
4,650323 -10,22669
28,23827
DummyIslam
100
1
0
1
1
60
0
0
0
0
Variable
Obs
Mean
Std. Dev.
Min
Max
ROA
127
1,483465
1,559266
-6,87
6,8
ROE
155
9,517935
15,09066
-136,02
27,91
zscore
127
3,590109
3,091899 -2,866959
12,40573
NPL
157
5,336306
5,35522
0
28,59
TA
159
3,80E+07
3,89E+07
1963490
1,50E+08
lnTA
159
16,76131
1,307192
14,49023
18,82857
TD_CE
159
196,7682
206,121
0
2402,87
TL_TD
159
127,6509
76,9787
67,89
997,42
OE
159
1541929
1915756
75254
8872704
OI
159
1000483
1790576
-359516
9084587
OER
159
2,661892
14,38927
-80,8905
130,3812
DummyIslam
160
0,625 0,4856429
0
1
Table 5: Summary of the statistics
Figure 2: Mean return on assets of Islamic banks and conventional banks
Figure 3: Mean return on equity of Islamic banks and conventional banks
-1,00 0,00 1,00 2,00 3,00 4,00 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Mean ROA Year
Mean ROA Islamic banks and conventional banks
Mean ROA Islamic banks Mean ROA conventional banks
-30,00 -20,00 -10,00 0,00 10,00 20,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean ROE Year
Mean ROE Islamic banks and conventional banks
16 Figure 4: Mean z-score of Islamic banks and conventional banks
Figure 5: Mean non-performing loans of Islamic banks and conventional banks
As shown in Figure 2, the mean return on assets of Islamic banks was higher in 2008 and from 2014 to 2017. The large difference in 2008 can be explained by the financial crisis of 2008. Conventional banks were affected more by this crisis, the higher return on assets of Islamic banks confirms the statements made by Tabash and Dhankar (2014) and Boumediene and Caby (2009). After 2008, the mean return on assets of conventional banks started increasing again while the return on assets of Islamic banks decreased. From 2010 to 2017 the return on assets of Islamic banks stayed between 1 and 2, mostly around 1,5 while the return on assets of conventional banks stayed around 1,5 but dropped almost to 1 for the last few years. The mean return on assets of Islamic banks varies between 1,26% and 3,11% while this is between -0,35% and 1,79% for conventional banks.
The mean return on equity shows a similar same pattern in the beginning of the sample period. The return on equity of Islamic banks was higher than the return on equity of conventional
0,00 1,00 2,00 3,00 4,00 5,00 6,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean z-score Year
Mean z-score Islamic banks and conventional banks
Mean zscore Islamic banks Mean zscore conventional banks
0,00 2,00 4,00 6,00 8,00 10,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean NPL Year
Mean NPL Islamic banks and conventional banks
banks in 2008, this has the same reason as for the return on assets. The return on equity followed the pattern of the return on assets after 2008, the return on equity of Islamic banks decreased, and it increased for conventional banks. From 2013 to 2017 the return on equity of Islamic banks was higher than the return on equity of conventional banks. The mean return on equity of Islamic banks varies between 7,66% and 19,27% while this is between -27,13% and 10,59% for conventional banks.
The mean score for Islamic banks and conventional banks is shown in Figure 4. The mean z-score was higher in 2008 for Islamic banks but decreased after this. It increased for conventional banks and from 2009 to 2013 it was higher than the mean z-score of Islamic banks. Even though the z-score of Islamic banks decreased since 2014, it was still higher after 2014 for every year except in 2017. The mean z-score varies between 3,13 and 5,56 for Islamic banks and between 2,91 and 4,48 for conventional banks.
The last dependent variable is the non-performing loans as a percentage of total loans. The mean percentage of non-performing loans is shown in Figure 5. From the figure it can be seen that the percentage of non-performing loans of conventional banks is more volatile and follows a different pattern than the percentage of non-performing loans of Islamic banks. This can also be seen in Table 4, the standard deviation of non-performing loans is higher for conventional banks. The mean percentage of non-performing loans varies between 3,29% and 6,45% for Islamic banks and between 2,77% and 8,13% for conventional banks.
The total assets varied between $1963490 and $150364007 for Islamic banks and between $2165088 and $138449052 for conventional banks. Taking the natural logarithm of total assets seemed to be better because the range of total assets is between $1963490 and $150364007 while this is between 14,49023 and 18,82857 for the natural logarithm, the outliers are thus smaller. The mean debt/equity ratio is 300.5139 for conventional banks and 150.7004 for Islamic banks. Islamic banks hold more equity, this could explain the difference. Islamic banks had a mean deposit ratio of 126.8181 while this was 129.0252 for conventional banks. The mean operating expense ratio is 2.081634 for Islamic banks and 3.619319 for conventional banks. This means Islamic banks have less expenses in relation to their income.
18
4.2 Sample two: Every available bank
The banks included in this sample are shown in Table 6.
Country
Islamic bank
Conventional bank
Bahrain Al Baraka Banking Group
Ithmaar Bank
National Bank Bahrain Bahrain Islamic Bank Gulf Finance House Khaleeji Commercial Bank
Arab Banking Corporation Bank of Bahrain and Kuwait Ahli United Bank Bahrain
Kuwait Ahli United Bank Kuwait
Kuwait Finance House Commercial Bank of Kuwait Kuwait International bank Warba Bank
Gulf Bank of Kuwait National Bank of Kuwait Al Ahli Bank of Kuwait
Qatar Qatar Islamic Bank
Masraf Al Rayan
Commercial Bank Qatar Al Khalij Commercial Bank Ahli Bank Qatar
Saudi Arabia Al Rajhi Bank
Alinma Bank Bank Aljazira Bank Albilad
United Arab Emirates Abu Dhabi Islamic Bank Commercial Bank of Dubai Ajman Bank
Table 6: Overview of the represented Islamic banks and conventional banks in sample two
Table 7 shows a summary of the statistics during the period 2008-2017 with Islamic banks and conventional banks separated, Table 8 shows a summary of the statistics of both groups of banks. One observation about the operating expense ratio is excluded since there was one
observation with operating expenses of 169594 and an operating income of 73, therefore getting an operating expense ratio of 2323. It’s excluded since this is a really large outlier and very unlikely. Figures 6 to 9 show the mean of the dependent variables per year for Islamic banks and
Variable
Obs
Mean
Std. Dev.
Min
Max
Obs
Mean
Std. Dev.
Min
Max
ROA
141
1,020638
3,790414
-27,6
11,49
83
1,36
1,245509
-6,87
3,72
ROE
191
6,352932
28,02584
-300,47
38,67
88
8,94875
17,69805
-136,02
25,96
zscore
141
3,902407
6,65799 -2,248334
29,181
83
4,036234
3,31688 -2,866959
12,40573
NPL
187
7,413316
10,60107
0
95,43
90
4,616444
5,136121
,34
28,59
TA
195
2,77e+07
3,53e+07
121441
1,50e+08
90
2,34e+07
3,03e+07
2165088
1,38e+08
lnTA
195
15,92383
1,815069
11,70718
18,82857
90
16,34645
1,096379
14,58797
18,74601
TD_CE
195
148,5267
130,3028
0
892,85
90
240,3533
253,2975
56,36
2402,87
TL_TD
193
152,2645
152,1219
49,85
1186,53
90
125,9938
21,27644
77,39
209,68
OE
194
1159423
1735521
5914
8872704
90
821627,4
1200413
90015
6237976
OI
195
769941,1
1668406
-209241
9084587
90
305852,7
409038,8
-359516
1706672
OER
195
4,533417
23,88974 -88,30502
163,887
90
3,291708
3,913876 -10,22669
28,23827
DummyIslam
200
1
0
1
1
90
0
0
0
0
20
Variable
Obs
Mean
Std. Dev.
Min
Max
ROA
224
1,146384
3,101161
-27,6
11,49
ROE
279
7,171685
25,22498
-300,47
38,67
zscore
224
3,951995
5,646186 -2,866959
29,181
NPL
277
6,504585
9,271713
0
95,43
TA
285
2,63e+07
3,38e+07
121441
1,50e+08
lnTA
285
16,05729
1,632751
11,70718
18,82857
TD_CE
285
177,5246
183,1199
0
2402,87
TL_TD
283
143,9098
126,6835
49,85
1186,53
OE
284
1052375
1591259
5914
8872704
OI
284
622870,8
1413417
-359516
9084587
OER
284
4,139917
19,85878 -88,30502
163,887
DummyIslam
290
,6896552
,4634345
0
1
Table 8: Summary of the statistics
Figure 6: Mean return on assets of Islamic banks and conventional banks
-2,00 -1,00 0,00 1,00 2,00 3,00 4,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean ROA Year
Mean ROA Islamic banks and conventional banks
Figure 7: Mean return on equity of Islamic banks and conventional banks
Figure 8: Mean z-score of Islamic banks and conventional banks
Figure 9: Mean non-performing loans of Islamic banks and conventional banks
-15,00 -10,00 -5,00 0,00 5,00 10,00 15,00 20,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean ROE Year
Mean ROE Islamic banks and conventional banks
Mean ROE Islamic banks Mean ROE conventional banks
0,00 1,00 2,00 3,00 4,00 5,00 6,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean z-score Year
Mean z-score Islamic banks and conventional banks
Mean zscore Islamic banks Mean zscore conventional banks
0,00 2,00 4,00 6,00 8,00 10,00 12,00 14,00 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Mean NPL Year
Mean NPL Islamic banks and conventional banks
22 The mean return on assets of Islamic banks and conventional banks is shown in Figure 6. The first thing that’s striking is the large difference in 2008. This difference can be explained by the financial crisis in 2008, the same reason as in the first sample. After 2008, the mean return on assets of conventional banks increased to 1,78% in 2011 and then decreased every year to 2017 except a small increase in 2014 and in 2017. Islamic banks had a large decrease in the mean return on assets since it’s negative in 2009 and 2010. In 2011 the return on assets started increasing again and from 2014 to 2017 they outperformed conventional banks, except for 2015. The mean return on assets of Islamic banks varies between -1,19% and 1,85% while this is between 0,48% and 1,78% for
conventional banks.
The mean return on equity of Islamic banks and conventional banks is shown in Figure 7. The return on equity of Islamic banks follows a similar pattern as the return on assets, except the
decrease in return on equity takes longer. In this case, the return on equity is decreasing from 2008 to 2011 and then starts increasing again. The return on equity of conventional banks was negative due to the financial crisis in 2008 but then increased again and stayed between 9% and 12%. The difference between the mean return on equity became a lot smaller from 2015 to 2017. The mean return on equity of Islamic banks varies between -11,36% and 18,43% while this is between -9,76% and 11,71% for conventional banks.
The mean z-score of Islamic banks and conventional banks is shown in Figure 8. The z-score decreased from 2008 till 2013 for Islamic banks, except in 2010 and 2011. It decreased in 2009 but increased again in 2010. Then it decreased till 2013 and stayed between 3 and 4 in the years after. From 2014 to 2017 the score was higher for Islamic banks. From Figure 8 it can be seen that the z-score of Islamic banks is more volatile and the standard deviation in Table 7 also confirms this. The mean z-score varies between 1,96 and 5,13 for Islamic banks and between 3,32 and 4,84 for conventional banks.
The mean percentage of non-performing loans of Islamic banks and conventional banks is shown in Figure 9. From the figure it can be seen that the percentage of non-performing loans of Islamic banks and conventional banks follows a similar pattern. However, the percentage of performing loans of Islamic banks is higher in every year except 2009. The mean percentage of non-performing loans varies between 2,82% and 11,56% for Islamic banks and between 2,16% and 6,29% for conventional banks.
The total assets varied between $121441 and $150364007 for Islamic banks and between $2165088 and $138449052 for conventional banks. Taking the natural logarithm of total assets also seemed to be better in this sample because the range of total assets is between $121441 and $150364007 while this is between 11,70718 and 18,82857 for the natural logarithm, the outliers are
thus smaller. The mean debt/equity ratio is 240,3533 for conventional banks and 148,5267 for Islamic banks. The debt/equity is also in this sample smaller for Islamic banks. The difference can also be explained for the same reason: Islamic banks hold more equity. Islamic banks had a mean deposit ratio of 152,2645 while this was 125,9938 for conventional banks. In this sample the deposit ratio of Islamic banks is larger while in sample one the deposit ratio is smaller. The mean operating expense ratio is 4,533417 for Islamic banks and 3,291708 for conventional banks. In this sample the result also changes, in sample one Islamic banks had a lower operating expense ratio while in this sample it’s higher. This means Islamic banks generate less income with every unit of costs and this could mean that they’re less efficient.
24
5. Results
This section will contain a regression analysis on the four dependent variables using the two samples to determine if there is a difference between both groups of banks.
5.1 Regression analysis using sample one
The regression that has been performed will be based on sample one in this subsection. In Table 9 the regression results based on sample one are given. Table 9 contains the coefficients for every regression. The values between the parentheses are the standard errors. Table 10 contains a summary of the regression outcomes.
ROA ROE z-score NPL
ln(Total Assets) 0,3458879*** (0,0799157) (0,4888683) 1,60738*** -0,4344119** (0,2134361) -1,396116*** (0,3108764) Total Debt % Common Equity
-0,0042528*** (0,0004495) -0,0623576*** (0,0031015) -0,0055587*** (0,0012004) 0,0013143 (0,0020068) Total Loans % Total Deposits 0,0088962*** (0,002909) (0,0190365) -0,0103834 (0,0077692) -0,0076177 -0,0155896*** (0,0050691) Operating Expense Ratio (0,0063954) -0,0089365 (0,0421645) 0,0136787 (0,0170806) -0,0093563 (0,0272772) -0,0266143 Islamic bank dummy
-0,2079839 (0,1992217) -3,736183*** (1,330342) -0,5739248 (0,5320747) 0,2579174 (0,8477672) Constant -4,341699*** (1,28562) (8,121191) -1,445023 13,2783*** (3,433593) 30,3933*** (5,267601) Number of observations 127 155 127 157 R-squared 0.5479 0,7615 0,1799 0,1952 Adjusted R-squared 0.5293 0,7535 0,146 0,1685 F-statistic 29,33 95,14 5,31 7,32
Table 9: Regression results from sample one (*p<0,10; **p<0,05; ***p<0,01)
ROA ROE z-score NPL
Islamic compared to conventional
No significant difference
Lower for Islamic banks
No significant difference
No significant difference Table 10: Summary of the regression outcomes
5.1.1 Regression with return on assets as dependent variable
The coefficients of the natural logarithm of total assets, the debt/equity ratio and the deposit ratio are two sided significant with an alpha of 1% in the regression with return on assets as dependent variable. Hereby it can be concluded that these variables are able to explain the return on assets. As expected, the size of the bank has a positive effect on the return on assets, this might
be due to economies of scale. The debt/equity ratio has a negative effect on the return on assets, this was also expected because an increase in debt reduces the profitability. The deposit ratio has a positive effect on the return on assets as expected because the profitability will increase if more money is lent. The operating expense ratio has a negative effect on the return on assets as expected but this effect isn’t significant. The coefficient of the dummy variable for Islamic banks is negative, this means that if the bank observed is an Islamic bank, it has a lower return on assets on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 9 it can be concluded that the p-value is 0,299 which is higher than 0,05 and therefore the coefficient of the dummy for Islamic banks isn’t statistically significant from zero and therefore the null hypothesis can’t be rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks don’t differ from conventional banks with respect to return on assets.
5.1.2 Regression with return on equity as dependent variable
The coefficients of the natural logarithm of total assets, the debt/equity ratio and the dummy for Islamic banks are two sided significant with an alpha of 1% in the regression with return on equity as dependent variable. Hereby it can be concluded that these variables are able to explain the return on equity. Same case as the regression on return on assets, it is expected that the size of the bank has a positive effect on the return on equity due to economies of scale. As expected, the size of the bank has a positive effect on the return on equity. The debt/equity ratio has a negative effect on the return on equity, this was also expected for the same reason as mentioned in the case of the return on assets regression, namely because an increase in debt reduces the profitability. The deposit ratio has a negative effect on the return on equity, this is not expected since an increase in loans should increase the profitability, however, the variable isn’t significantly different from zero, so no conclusions can be drawn from this variable. The operating expense ratio has a positive effect on the return on equity, this also isn’t expected because an increase in the operating expense ratio means the operating income declined or the operating expenses increased and therefore the return on equity should decrease, this variable also isn’t significantly different from zero, so no conclusions can be drawn again. The coefficient of the dummy variable for Islamic banks is negative, this means that if the bank observed is an Islamic bank, it has a lower return on equity on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
26
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 9 it can be concluded that the p-value is 0,006 which is lower than 0,05 and therefore the coefficient of the dummy for Islamic banks is statistically significant from zero and therefore the null hypothesis is rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks differ from conventional banks with respect to return on equity. To determine if the return on equity is significantly lower than zero, a second hypothesis is tested:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(< 0
Because this test is one tailed, the value of the two tailed test needs to be halved. The p-value will be 0,003 and therefore is still significant with an alpha of 5%. According to this sample, Islamic banks are less profitable in terms of return on equity.
5.1.3 Regression with z-score as dependent variable
The coefficient of the natural logarithm of total assets is two sided significant with an alpha of 5% and the coefficient of the debt/equity ratio is two sided significant with an alpha of 1% in the regression with the z-score as dependent variable. The coefficient of the natural logarithm of total assets is significantly lower than zero with a p-value of 0,022 and therefore a larger bank is less stable according to this sample, this in line with the conclusion from Čihák and Hesse (2010), who concluded that small Islamic banks were more stable than large Islamic banks. The difference is that it also holds for conventional banks according to this sample. The coefficient of the debt/equity ratio is significantly lower than zero with a p-value of 0,000. This is expected since it’s assumed that an increase in debt would reduce the stability and therefore the z-score due to the later repayments and the uncertainty that may arise with these repayments, also the payment of interest and the uncertainty that this may cause in the case of conventional banks decrease the stability. Hereby it can be concluded that these variables are able to explain the z-score. The deposit ratio and the operating expense ratio are both negative as expected but these variables aren’t significant. The coefficient of the dummy variable for Islamic banks is negative, this means that if the bank observed is an Islamic bank, it has a lower z-score on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 9 it can be concluded that the p-value is 0,283 which is higher than 0,05 and therefore the coefficient of the dummy for Islamic banks isn’t statistically significant from zero and therefore the null hypothesis can’t be rejected.
Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks don’t differ from conventional banks with respect to the z-score.
5.1.4 Regression with non-performing loans as dependent variable
The coefficients of the natural logarithm of total assets and the deposit ratio are two sided significant with an alpha of 1% in the regression with the percentage of non-performing loans as dependent variable. The coefficient of the natural logarithm of total assets is significantly lower than zero with a p-value of 0,000, this means that a larger bank has more loans that aren’t repaid and therefore a larger bank is less stable according to this sample. The coefficient for the deposit ratio is significantly lower than zero with a p-value of 0,001. This is not expected since it means that the more money is lent, the less the percentage of non-performing loans. However, we still can conclude these variables are able to explain the percentage of non-performing loans. The coefficient of the debt/equity ratio is positive but isn’t significantly different from zero. The coefficient of the operating expense ratio is negative as expected but this also isn’t significant. The coefficient of the dummy variable for Islamic banks is positive, this means that if the bank observed is an Islamic bank, it has a higher percentage of non-performing loans on average in this sample. To test if this
coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 9 it can be concluded that the p-value is 0,761 which is higher than 0,05 and therefore the coefficient of the dummy for Islamic banks isn’t statistically significant from zero and therefore the null hypothesis can’t be rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks don’t differ from conventional banks with respect to the percentage of non-performing loans.
28
5.2 Regression analysis using sample two
The regression that has been performed will be based on sample two in this subsection. In Table 11 the regression results based on sample two are given. Table 11 contains the coefficients for every regression. The values between the parentheses are the standard errors. Table 12 contains a summary of the regression outcomes.
ROA ROE z-score NPL
ln(Total Assets) 0,5647288*** (0,128442) 4,159604*** (0,8344944) (0,2492001) -0,1890162 -2,197555*** (0,3191683) Total Debt % Common Equity
-0,0048853*** (0,0010375) -0,0569451*** (0,0077363) -0,0045336** (0,002013) 0,0043619 (0,0029154) Total Loans % Total Deposits (0,0017078) 0,0031456* 0,0312785** (0,0127803) (0,0033135) -0,006133* (0,0044504) -0,007425* Operating Expense Ratio (0,0102842) -0,0057157 0,227919*** (0,0754775) (0,0199531) -0,0142125 -0,066984** (0,0264191) Islamic bank dummy (0,4214194) -0,6261937 -7,068221** (2,966697) (0,8176273) -0,3126299 2,719927** (1,12214) Constant -7,007682*** (2,149661) -49,72497*** (14,11036) (4,170719) 9,067695 40,64626*** (5,391422) Number of observations 221 276 221 276 R-squared 0.1722 0,263 0,0562 0,2019 Adjusted R-squared 0.1530 0,2493 0,0343 0,1871 F-statistic 8,95 19,27 2,56 13,66
Table 11: Regression results from sample two (*p<0,10; **p<0,05; ***p<0,01)
ROA ROE z-score NPL
Islamic compared to conventional
No significant difference
Lower for Islamic banks
No significant difference
Higher for Islamic banks
Table 12: Summary of the regression outcomes
5.2.1 Regression with return on assets as dependent variable
The coefficients of the natural logarithm of total assets and the debt/equity ratio are two sided significant with an alpha of 1% in the regression with return on assets as dependent variable. The coefficient of the deposit ratio is two sided significant with an alpha of 5%. Hereby it can be concluded that these variables are able to explain the return on assets. As expected, the same result as the in the regression from sample one is obtained, the size of the bank has a positive effect on the return on assets. This might also be due to economies of scale. The debt/equity ratio also has a negative effect, as explained in the analysis of sample one, this was expected because an increase in debt reduces the profitability. The outcome of the effect of the deposit ratio also is the same as in
sample one, the coefficient is positive and thereby a higher deposit ratio increases profitability. Only difference is that using sample two, the coefficient is only significant with an alpha of 10% because of the p-value of 0,067, instead of the p-value of 0,003 it had in the regression using sample one. The coefficient of the operating expense ratio is negative like in sample one, but it still is insignificantly different from zero. The coefficient of the dummy variable for Islamic banks is negative like in sample one, this means that if the bank observed is an Islamic bank, it has a lower return on assets on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 11 it can be concluded that the p-value is 0,139 which is higher than 0,05 and therefore the coefficient of the dummy for Islamic banks isn’t statistically significant from zero and therefore the null hypothesis can’t be rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks don’t differ from conventional banks with respect to return on assets.
5.2.2 Regression with return on equity as dependent variable
The coefficients of the natural logarithm of total assets, the debt/equity ratio and the operating expense ratio are two sided significant with an alpha of 1% in the regression with return on equity as dependent variable, while the coefficients of the deposit ratio and the dummy for Islamic banks are two sided significant with an alpha of 5%. Hereby it can be concluded that these variables are able to explain the return on equity. The size of the bank, the debt/equity ratio and the deposit ratio have the same effect on the return on equity as on the return on assets. The size of the bank and the deposit ratio have a positive effect while the debt/equity ratio has a negative effect on the return on equity. The effects regarding the size, the deposit ratio and the operating expense ratio are the same as in sample one, but the effect of the deposit ratio was negative in sample one and is positive in this sample, as expected. In this case, an increase in loans thus increases return on equity, contrary to sample one. The coefficient of the dummy variable for Islamic banks is negative, this means that if the bank observed is an Islamic bank, it has a lower return on equity on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 11 it can be concluded that the p-value is 0,018 which is lower than 0,05 and therefore the coefficient of the dummy for Islamic banks is statistically significant from zero and therefore the null hypothesis is rejected. Therefore, it
30 can be concluded that from this sample from 2008 to 2017 Islamic banks differ from conventional banks with respect to return on equity. To determine if the return on equity is significantly lower than zero, a second hypothesis is tested:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(< 0
Because this test is one tailed, the value of the two tailed test needs to be halved. The p-value will be 0,009 and therefore is still significant with an alpha of 5%. According to this sample, Islamic banks are less profitable in terms of return on equity.
5.2.3 Regression with z-score as dependent variable
The coefficient of the debt/equity is two sided significant with an alpha of 5% and the coefficient of the deposit ratio is two sided significant with an alpha of 10% in the regression with the z-score as dependent variable. The coefficient of the debt/equity ratio is significantly lower than zero with a p-value of 0,0125. This is expected since it’s assumed that an increase in debt would reduce the stability and therefore the z-score as explained in the regression with sample one. The deposit ratio is significantly lower than zero with a p-value of 0,033. This is expected since an increasing in lending should decrease the z-score because the stability reduces. Hereby it can be concluded that these variables are able to explain the z-score. The coefficient of the natural
logarithm of total assets was significantly different from zero in the regression with sample one but isn’t significant in this regression. This may be due to the fact that also small banks are included in this sample, therefore the standard deviation of the natural logarithm of total assets got bigger and this maybe reduces the ability to predict the z-score. It can be seen from Table 4 and 7 that the standard deviation of the natural logarithm of total assets increased in sample two. Also, the coefficient of the deposit ratio wasn’t significant in the regression with sample one but is significant in this sample. The operating expense ratio has the same effect in sample two as in sample one, namely a negative effect on the z-score as expected, but both coefficients are not significant. The coefficient of the dummy variable for Islamic banks is negative, this means that if the bank observed is an Islamic bank, it has a lower z-score on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From table 11 it can be concluded that the p-value is 0,703 which is higher than 0,05 and therefore the coefficient of the dummy for Islamic banks isn’t statistically significant from zero and therefore the null hypothesis can’t be rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks don’t differ from conventional banks with respect to the z-score.
5.2.4 Regression with non-performing loans as dependent variable
The coefficients of the natural logarithm of total assets is two sided significant with an alpha of 1%, while the coefficients of the operating expense ratio and the dummy for Islamic banks are two sided significant with an alpha of 5% and the coefficient of the deposit ratio is two sided significant with an alpha of 10% in the regression with the percentage of non-performing loans as dependent variable. The natural logarithm of total assets is significantly lower than zero with a p-value of 0,000, this result was also obtained in sample one and means that a larger bank has more loans that aren’t repaid and therefore a larger bank is less stable according to this sample. The coefficient of the debt/equity ratio is positive but isn’t significantly different from zero, same as in sample one. The coefficient for the deposit ratio is significantly lower than zero with a p-value of 0,048. This is not expected like in sample one since it means that the more money is lent, the less the percentage of non-performing loans. The coefficient of the operating expense ratio is negative as expected insignificant in sample one while it is significantly lower than zero with a p-value of 0,006 in this sample. The coefficient of the dummy variable for Islamic banks is positive, this means that if the bank observed is an Islamic bank, it has a higher percentage of non-performing loans on average in this sample. To test if this coefficient is significantly different from zero, a t-test is conducted with the following hypothesis:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(≠ 0
This hypothesis will be tested with an alpha of 5%. From Table 11 it can be concluded that the p-value is 0,016 which is lower than 0,05 and therefore the coefficient of the dummy for Islamic banks is statistically significant from zero and therefore the null hypothesis is rejected. Therefore, it can be concluded that from this sample from 2008 to 2017 Islamic banks differ from conventional banks with respect to the percentage of non-performing loans. To determine if the percentage of non-performing loans is significantly higher than zero, a second hypothesis is tested:
!
": %
&'(()*+,-(= 0 012 !
3: %
&'(()*+,-(> 0
Because this test is one tailed, the value of the two tailed test needs to be halved. The p-value will be 0,008 and therefore is still significant with an alpha of 5%. According to this sample, Islamic banks are less stable in terms of the percentage of non-performing loans.
5.3 Explanation for the obtained results
In both samples the coefficient of the dummy variable for Islamic banks is insignificant in the regressions with return on assets and the z-score as dependent variables. Therefore, it can be concluded Islamic banks don’t differ in return on assets and in z-score. In sample one the coefficient
32 of the dummy variable for Islamic banks with the non-performing loans is insignificant while it is significant in sample two.
One of the reasons for the findings with respect to the return on assets could be that the Islamic banking products are similar to conventional products, since Islamic financial contracts are mostly contracts with a profit markup. These contracts are similar to a debt contract that is offered by a conventional bank. Khan (2010) also supports this statement by concluding that Islamic banks provide almost the same products as conventional banks and that Islamic banking isn’t a better alternative than conventional banking.
The reason for the lower return on equity of Islamic banks could be bad investments that have been made. The bank only carries the loss of an investment since most Islamic financial products work according to the profit- and loss sharing principle. Therefore, their return on equity could decrease as a result of bad investments. Also, it could be moral hazard. Borrowers could be less efficient with the money obtained from the bank since they know that in the case of a loss the bank is the only one who carries the loss.
The explanation for the insignificant coefficient of the dummy variable in the regressions with z-score as dependent variable could be the same as for the return on assets, namely that Islamic banking products have the same characteristics as conventional banking products and therefore don’t differ in z-score.
The reason for the higher percentage of non-performing loans for Islamic banks could be the same reason as mentioned for the return on equity. Namely that Islamic banks could have made bad investments or that borrowers are less efficient with the money they borrowed and that they’re facing moral hazard. The reason that the dummy variable isn’t significant in sample one while it is significant in sample two could be because sample two contains more small sized Islamic banks, they could have a higher percentage of non-performing loans. This is true according to Table 4 and Table 7, since the mean of non-performing loans of Islamic banks is 5.143402 in sample one while it is 7,413316 in sample two.
6. Conclusion and discussion
In this thesis a comparative analysis is performed with respect to the profitability and stability of Islamic banks and conventional banks. Two samples are used. The first sample contains the two biggest Islamic banks per country to represent the whole GCC region and the two biggest conventional banks if there were two fully conventional banks per country. Oman is excluded from the sample, since the Islamic banking sector is still in the early stages. The second sample contains every fully Islamic bank and every fully conventional bank available on Datastream with five or more observations from the last ten years.
To determine if there is a significant difference, eight regressions are conducted with a dummy variable for Islamic banks and control variables total assets, debt/equity ratio, deposit ratio and the operating expense ratio.
From the regressions it can be concluded that there is no significant difference in return on assets in both samples between Islamic banks and conventional banks. This could be because Islamic banking products have almost the same characteristics as conventional banking products.
In both samples Islamic banks had a significant lower return on equity. This could be due to bad investments or because borrowers are less efficient with the money they borrowed and that they’re facing moral hazard.
Both samples concluded that there is no significant difference in the z-score. This could be due to the same reason that there is no significant difference in return on assets, namely that Islamic banking products have the same characteristics as conventional banking products.
In sample one there wasn’t a significant difference in percentage of non-performing loans while there was a significant difference in sample two. This could be because sample two contains more small sized Islamic banks, they could have a higher percentage of non-performing loans. The difference in sample two could be due to the same reason as for return on equity, namely that Islamic banks could have made bad investments or that borrowers are less efficient with the money they borrowed and that they’re facing moral hazard.
One of the main limitations from this thesis is the lack of data. There isn’t much data about the variables used in this thesis, especially there is little data on the return on assets. Another limitation is the lack of banks. Fully Islamic banks and fully conventional banks are hard to find nowadays, since most conventional banks also have a division that offers products which comply with the Sharia. Research that has been done in the past years was able to include more
conventional banks since many conventional banks did not offer those products a few years ago. Also, the sample period is important, since the Islamic financial sector is expected to grow with 9%
34 on average till 2022 according to Mohamed and Goni (2017). Therefore, it’s important to see if these results still hold in the long run, this is also mentioned by Siraj and Pillai (2012).
7. References
Čihák, M., & Hesse, H. (2008). Islamic Banks and Financial Stability: An Empirical Analysis. Journal of
Financial Services Research, 38(2), 95-113.
Ariff, M. (1988). Islamic banking. Asian-Pacific Economic Literature, 2(2), 48-64.
Beck, T., Demirgüç-Kunt, A., & Merrouche, O. (2013). Islamic vs. conventional banking: Business model, efficiency and stability. Journal of Banking & Finance, 37(2), 433-447.
Boumediene, A., & Caby, J. (2009, December 17). The Financial Stability of Islamic Banks during the
Subprime Crisis. Retrieved from SSRN: http://ssrn.com/abstract=1524775
Dar, H. A., & Presley, J. R. (2000). Lack of profit loss sharing in Islamic banking: Management and control imbalances. International Journal of Islamic Financial Services, 2(2), 3-18.
El Qorchi, M. (2005). Islamic Finance Gears Up. Finance & Development Magazine by IMF, 42(4), 46-49.
Khan, F. (2010). How ‘Islamic’ is Islamic Banking? Journal of Economic Behavior & Organization, 76, 805-820.
Latif, Y., Abbas, A., Akram, M. N., Manzoor, S., & Ahmad, S. (2016). Study of performance comparison between Islamic and conventional banking in Pakistan. European Journal of
Educational and Development Psychology, 4(1), 17-33.
Loghod, H. A. (2010). Do Islamic Banks Perform Better than Conventional Banks? Evidence from Gulf Cooperation Council countries. API Working Paper Series, 1011.
Mohamed, S., & Goni, A. (2017). Islamic finance development report 2017. ICD-Thomson Reuters. Olson, D., & Zoubi, T. A. (2008). Using accounting ratios to distinguish between Islamic and
conventional banks in the GCC region. The International Journal of Accounting, 43, 45-65. Parashar, S., & Venkatesh, J. (2010). How did Islamic banks do during global financial crisis? Banks
and Bank Systems, 5(4), 54-62.
Petersen, M. A., & Schoeman, I. (2008). Modeling of Banking Profit via on-Assets and Return-on-Equity. Proceedings of the World Congress on Engineering 2008, 2.