Content uploaded by Yasir Khan Khalil
Author content
All content in this area was uploaded by Yasir Khan Khalil on Aug 14, 2017
Content may be subject to copyright.
317
C 2017 CURJ, CUSIT
City University Research Journal
Volume 07 Number 02 July 2017 PP 317-333
2 Lecturer, Northern University, Nowshera.
3 Ph.D Scholar, Qurtuba University, Peshawar.
318
Arif Hussain et al.
C 2017 CURJ, CUSIT
Risk Taking Behavior of Commercial Banks in Pakistan
319
Independent Variables Dependent Variable
Bank Risk
·
·
·St
N )
·
C 2017 CURJ, CUSIT
320
SDNPL = Standard deviation of ratio of loan loss provision to total loans.
SDROE = Standard deviation of return on equity
SDROA = Standard deviation of return on assets.
Operational Definition and Measurement of Variables
Dependent Variable
Return on Assets (ROA) = Return on asset is the ratio between net profit and total assets
of the firm. It shows operating efficiency of assets to generate profits.
Independent Variables
Standard Deviation of ROE = Standard deviation of return on equity (SDROE) is a
proxy for bank risk. It is measured as standard deviation of each bank ROE based on ten
years annual data. This measure is also used by Tandelilin et al (2007) and Cebenoyan
and Strahan (2004) as a measure for risk management in banks.
Standard Deviation of ROA = Standard deviation of return on assets (SDROA) is used
as a second proxy for bankrisk. It is measured as standard deviation of each bank ROA
based on ten years annual data. This measure is also used by Tandelilin et al (2007) and
Cebenoyan and Strahan (2004) as a measure for risk management in banks.
Standard Deviation of Non-Performing Loans = Standard deviation of non performing
loans (SDNPL) is used as a third proxy bank risk.It is measured asstandard deviation of
loan loss provision for loans to total loan ratio based on ten years annual data. This
measure is also used by Tandelilin et al (2007) and Cebenoyan and Strahan (2004) as a
measure for risk management in banks.
RESEARCH METHODOLOGY
This study empirically tests risk taking behavior on the performance of commercial
banks in Pakistan.Secondary data was collected from the published annual reports of
five large commercial banks and ten small commercial banks for a period of 2005 to
2015. Banks having more than one thousand branches were considered as large
commercial banks and banks having less than one thousand branches were considered
as small commercial banks. There are a total of five public sector commercial banks and
twenty two private sector commercial banks registered in Pakistan. List of these banks
along with their number of branches is given in appendix I. Only five commercial banks
including Habib Bank Limited (HBL), United Bank Limited (UBL), National bank of
Pakistan (NBP), Muslim Commercial bank (MCB) and Allied bank Limited (ABL)
have more than one thousand branches and have 54.4% of the total market share.So total
population based sampling technique applied for the large banks as all these banks have
more than one thousand branches and included all in the data analysis.However, for the
small banks random sampling technique applied as all these banks have less than one
thousands branches across the country.The remaining twenty two commercial banks
have less than one thousand branches. Ten commercial banks were randomly selected
having less than one thousand branches. Descriptive statistics and random effect OLS
regression was used for the analysis of panel data alsoIndependent Sample T-Test was
used for making comparison of risk taking behavior between large commercial banks
and small commercial banks in Pakistan.
Arif Hussain et al.
C 2017 CURJ, CUSIT
321
DATA ANALYSIS
Multicoliniarity Test
Table Multicolinearity Statistics
The above table shows the muilticoleanarity analysis of the independent variables of
this research study. The results of this study show the values tolerance and VIF within
feasible ranges as per the suggestions of (O?Brien& Robert, 2007). The above tolerance
values reflect that the tolerance level is moderate and good and the VIF values also
showing within the feasible ranges
Table1: Multicoliniarity Test
Table 1 shows result of multicolliniarity test. Multicolliniority shows the extreme
correlation with in the independent variables of the study. According to Gujarati (2003)
if the correlation with in independent variables is more than 0.8, then there will be
problem of multicolliniarity in the data. As all the correlations with independent
variables are less than 0.8 so there is no problem of multicolliniarity.
Breusch-Pagan Test
Table 2: Beusch-Pagan test
Table 2 shows result of Breuch-Pagan test to estimate the problem of heteroscedasticity
in the data. The reported values are above than the critical value, so the data is not
showing any problem of non constant variation. Hence there is no problem of
heteroscedasticity in the data.
DESCRIPTIVE STATISTICS
Table 3 Descriptive Statistics for Large Banks and Small Banks
Colinearity Statistics
Tolerance VIF
SDROA .745 1.103
SDSP .567 1.112
SDNPL .657 1.224
Variables ROA SDSP SDROA SDNPL
ROA 1.00
SDSP -0.293
1
SDROA 0.621
0.154 1 1
SDNPL -0.556 0.330 0.343 0.234
Test type Critical/Standard value Reported value
Breusch-Pagan Test 0.05
0.087
Risk Taking Behavior of Commercial Banks in Pakistan
C 2017 CURJ, CUSIT
322
Table 3 shows descriptive statistics for small commercial banks and large commercial
banks. Relative comparisons of the mean values for the risk variables of SDROA,
SDROE and SDNPL show that large banks have lower mean values for all the risk
variables than small banks. This suggests that large banks in Pakistan assume lower risk
as compared to small banks in Pakistan. The return on assets for large banks is higher
than the small banks.
Table 4: Summary of Independent Samples T-Tests
Standard Deviation of ROE (SDROE):
As the probability value 0.023<0.005, we reject the null hypothesis and hence it can be
inferred that based on the analysis there are significant differences between large banks
and small banks SDROE. So it is concluded that large banks have lower risk.
Standard Deviation of Stock Prices (SDNPL):
As the probability value 0.00<0.005, we reject the null hypothesis and hence it can be
inferred that based on the analysis there are significant differences between large bank
and small bank SDNPL. So it is concluded that large banks have lower risk.
Standard Deviation of ROA (SDROA):
As the probability value 0.007<0.005, we reject the null hypothesis and hence it can be
inferred that based on the analysis there are significant differences between large banks
and small banks SDROA. So it is concluded that large banks has lower risk.
Hausman Test
The Hausman test statistic was used in order to choose between random effect and fixed
effectmodel for models. The null hypothesis of the Hausman test was that the random
effects model was preferred to the fixed effects model. Hausman test reported a chi-
square of 2.015 with a probability value of 0.906shows that the chi-square value
obtained was statistically insignificant. The null hypothesis was therefore failed to
reject, so random effect model was recommended.
Arif Hussain et al.
C 2017 CURJ, CUSIT
Variables Hypothesis Decision Comments
SDROA
H3: Large commercial
banks have lower risk
Supported
Differences and
Reject H0
SDROE
H3: Large commercial
banks have lower risk
Supported
If sig <
0.05 then
Differences
and Reject H0
SDNPL
H3: Large commercial
banks have lower risk
Supported
Differences and Reject H0
SDROE 0.259
0.228
0.262
0.279 Large banks have lower risk
SDNPL 36.936 17.991 15.680 38.916 Large banks have lower risk
Variables Small Banks Large Banks Comments
SD
M
SD
M
ROA 0.008
0.012
0.006
0.016 Large banks have higher ROA
SDROA 0.007
0.005
0.002
0.003 Large banks have lower risk
323
Risk Taking Behavior of Commercial Banks in Pakistan
C 2017 CURJ, CUSIT
Variable
s
Large Banks Small Banks
s
Standard Error
T-Values
Standar
d Error
T-Values
SDROE -0.113
0.054
-2.06
(0.042) -0.212
0.097
-2.16
(0.026)
SDROA -0.232
0.109
-2.12
(0.024)
-0.121
0.091
-1.356
(0.076)
SDNPL -0.034
0.015
-2.31
(0.013)
-0.04
0.1611
-0.09
(0.928)
R-Square = 35.7% R-Square = 23.6%
Wald Chi = 41.72 Wald Chi = 46.62
324
Arif Hussain et al.
C 2017 CURJ, CUSIT
325
Source: State Bank of Pakistan (SBP) website.
4 Allied Bank Limited 1048 6.7% Bank of Punjab 405 4%
6 Meezan Bank Limited 551 3.7%
7 Habib Metropolitan
Bank Ltd.
237 3.1%
8
Standard Chartered
Bank (Pakistan) Ltd.
101
3.1%
9
NIB Bank Ltd.
171
2.5%
10
Soneri Bank Ltd.
266
2.3%
11
Summit Bank Ltd.
191
1.6%
12
Silkbank Ltd.
88
1.3%
13
Dubai Islami Bank
Pakistan Ltd. 200
1.2%
14 JS Bank Ltd. 243
1.2%
15 Al Baraka Bank
(Pakistan) Ltd.
121
1%
16
Bank Islami Pakistan
Ltd.
176
0.9%
17
The Bank of Khyber
131
0.9%
18
Sindh bank
242
0.8%
19
Samba Bank Ltd.
34
0.5%
20
First Women bank
43
0.2%
22 Burj Bank Ltd. 74 0.2%
22 MCB Islamic Bank
Ltd.
6 0.01%
5 Muslim Commercial
Bank Limited
1247
6.6% Bank Al Habib Ltd.
420
3.8
S/No Large Commercial
Banks
Branches Market
Share
Small Commercial
Banks
Branches Market
Share
1 NationalBank
of
Pakistan
1406
14.9%
Bank Alfalah Limited
630
6.2%
2 Habib Bank Limited
1663
13.5% Faysal Bank Limited
281
4.2%
3 United Bank Limited
1282
10.5% Askari Bank Ltd.
391
4.1
Risk Taking Behavior of Commercial Banks in Pakistan
C 2017 CURJ, CUSIT