Document Text (Pages 51-60) Back to Document

# Foreign Exchange Risk Management in Commercial Banks in Pakistan

Page 51

Step 1 OS -56.634 7.040 1 .008
TYPE -53.120 .012 1 .911
Size (NA) -73.188 40.147 1 .000
NFXNA -54.109 1.990 1 .158
Current ERV -53.430 .632 1 .427
Step 2 OS -56.706 7.170 1 .007
Size (NA) -73.618 40.996 1 .000
NFXNA -54.139 2.038 1 .153
Current ERV -53.440 .639 1 .424
Step 3 OS -56.801 6.724 1 .010
Size (NA) -73.650 40.420 1 .000
NFXNA -54.964 3.049 1 .081
Table xii: Factors that affect Currency Derivative Usage: Result of Binary Logistic regression

Below is the summary of Logits and Odd Ratios which show that how the binary
logistic regression model is changed from five independent variables to three
independent variables, Step 1 & 2 do not need to be discussed in detail, however,
step 3 needs to be explained in detail as it corresponds to our objective of this
study.

Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Step 1a OS(1) -3.257 1.679 3.763 1 .052 .039
TYPE(1) -.072 .642 .012 1 .911 .931
Size (NA) .086 .021 16.757 1 .000 1.090
NFXNA -.008 .006 1.962 1 .161 .992
Current ERV -.012 .015 .627 1 .428 .989
Constant -.484 .798 .368 1 .544 .616
Step 2a OS(1) -3.262 1.676 3.787 1 .052 .038
Size (NA) .086 .021 16.818 1 .000 1.090
NFXNA -.008 .006 2.013 1 .156 .992
Current ERV -.012 .015 .634 1 .426 .988
Constant -.552 .525 1.104 1 .293 .576

Page | 51

Page 52

Step 3a OS(1) -3.053 1.580 3.736 1 .053 .047
Size (NA) .084 .020 16.779 1 .000 1.087
NFXNA -.010 .006 2.997 1 .083 .990
Constant -.647 .512 1.595 1 .207 .523
a. Variable(s) entered on step 1: OS, TYPE, NA, NFXNA, Current ERV.
Table xiii: Logits & Odds ratio Results of Binary Logistic Regression

The second column, with heading B, shows the values of regression coefficient
during each step. Logits (Log Odds) are the regression coefficients of the model
which show the impact of a unit increase in the independent variable on the
dependent variable i.e. how does a unit increase in independent variables affects
the decision of the commercial bank to select the tools for foreign exchange risk
management. In the last column, Exp (B) i.e. Odds Ratios are mentioned which
shows that how do a unit increase in independent variables changes the odds of
usage of swaps options along with forwards by a commercial bank.
It is important to interpret the odds ratio in step 3 here. If the ownership status of
bank is changed from public sector commercial bank to local private bank, the
odds of using swaps & options are
0.047 times greater than public sector commercial bank. This shows a weak
influence of ownership status on decision of usage of swaps and options. If the
Net Assets of bank are increased by one billion, the odds of usage of swaps and
options are 1.087 times greater. If NFXNA, when expressed as percentage, is
increased by one percent the odds are 0.99 times greater for the usage of swaps
and options.
Cox & Snell R square for the model as in step 3 is 0.321 and Nagelkerke R square
is 0.430. Based on these results, it can be stated that using our model 43% of the
changes in decision to use swaps and option are because of the independent
variables in our model.

Page | 52

Page 53

Findings on Income from Dealing in Foreign Currencies and its
Relationship with other Factors

Income from dealing in foreign currencies is mentioned separately in the income
statement of commercial banks under the head of “Income from dealing in foreign
currencies”. Findings related to this income as per the research objectives of this
study are given below:
Findings on Income from dealing in foreign currencies and Type of
commercial banks (Conventional & Islamic)
Below are the findings regarding the variables income from dealing in foreign
currencies in absolute terms and income from dealing in foreign currencies
relative to its total income.

Descriptive Statistics for Convenntional Banks
N Minimum Maximum Mean Std. Deviation
IFX % of TI 94 -32.01639 33.53655 5.0842763 6.69593911
Income from FX 94 -79327.00 3969057.00 513044.2766 6.98267E5

Descriptive Statistics for Islamic Banks
N Minimum Maximum Mean Std. Deviation
IFX % of TI 16 .58781 21.15430 6.9507450 4.87415155
Income from FX 16 740.00 1019732.00 337289.6875 3.15040E5

Table xiv: Descriptives for IFX and IFXRS by Type of Bank

Comparison of IFX of Conventional & Islamic Banks
Independent sample t-test is used to test the hypothesis that there is no significant
difference between means of conventional and Islamic banks.
The value of F using Levene’s test is insignificant and hence it is assumed that
variances are not different. The value of t is -1.066 which is not significant at
desired level of significance (as it is significant at 0.289) hence Null hypothesis is
accepted that there is no significant difference between the means of IFX of
commercial banks and Islamic banks.

Page | 53

Page 54

Findings on Income from dealing in foreign currencies and Ownership
Status of commercial bank (PSCB and LPB)
Results of the descriptive statistics for commercial banks with respect to
ownership status are below:

Descriptive Statistics for Public Sector Commercial Banks
N Minimum Maximum Mean Std. Deviation
IFX % of TI 14 .24369 7.42242 3.2638994 1.96094421
Income from FX 14 3371.00 3969057.00 835056.3571 1.23401E6

Descriptive Statistics Local Private Banks
N Minimum Maximum Mean Std. Deviation
IFX % of TI 96 -32.01639 33.53655 5.6608261 6.84683360
Income from FX 96 -79327.00 2229809.00 436791.7500 5.18312E5

Table xv: Desxriptive Statistics for IFX and IFXRS by Ownership Status

Comparison of IFX between Public Sector Commercial Bank & Local
Private Banks
To compare the mean IFX of public sector commercial banks & local private
banks, independent sample t-test is used.
The Levene’s test is significant and equality of variances is assumed. The
calculated value of t is -1.297 which is significant at 0.1097 level of significance.
Since level of significance is much higher than the desired one, null hypothesis is
accepted which shows that both groups have not significantly different means.

Table xvi: Descriptive Statistics for IFX and IFXRS by Ownership
Table xvii: Descriptive Statistics of IFX and IFXRS by Ownership Status of Bank

Findings on Income from dealing in foreign currencies and Size of Bank
as measured by Net Assets
Using the simple linear regression, following outputs are produced:
Model 1:

Page | 54

Page 55

Model Summary

Model R R Square

Square
Std. Error of the

Estimate
1 .764a .584 .581 4.26448E5
a. Predictors: (Constant), Net Assets in Billions

ANOVA
b
Model Sum of Squares df Mean Square F Sig.
1 Regression 2.762E13 1 2.762E13 151.849 .000a
Residual 1.964E13 108 1.819E11

Total 4.726E13 109
a. Predictors: (Constant), Net Assets in Billions
b. Dependent Variable: Income from dealing in foreign currencies in ‘000 Rs.

Coefficientsa
Standardized
Unstandardized Coefficients Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 106691.287 51070.033 2.089 .039
Net Assets in Billions 20390.617 1654.717 .764 12.323 .000
a. Dependent Variable: Income from dealing in foreign currencies in ‘000 Rs.
Table xviii: Output of Regression: IFXRS on NA

The first model to study the impact of Size of bank, as measured by Net Assets,
on the Income from dealing in foreign currencies show that there is a significant
relationship. The model R square shows that 58.1% of variation is explained by
the relationship. The value of b i.e. 20390.617 is also significant. Since there is a
relationship between independent and dependent variable, therefore, null
hypothesis is rejected and alternative hypothesis is rejected.

Page | 55

Page 56

Model 2:

Model Summary

Model R R Square

Square
Std. Error of the

Estimate
1 .067a .004 -.005 6.49309152
a. Predictors: (Constant), Net Assets in Billions

ANOVA
b
Model Sum of Squares df Mean Square F Sig.
1 Regression 20.397 1 20.397 .484 .488a
Residual 4553.306 108 42.160

Total 4573.703 109
a. Predictors: (Constant), Net Assets in Billions
b. Dependent Variable: Income from dealing in foreign currencies as a percentage of total income

Coefficientsa

Unstandardized Coefficients
Standardized
Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 5.683 .778 7.308 .000
Net Assets in Billions -.018 .025 -.067 -.696 .488
a. Dependent Variable: Income from dealing in foreign currencies as a percentage of total income
Table xix: Output of Regression: IFX on NA

Results for the model constructed to investigate the relationship between net
assets (size of bank) and income from dealing in foreign currencies as a
percentage of total income indicates that there is no significant relationship
between these two variables. Null hypothesis is accepted in this case.
The difference between results shown by these two models can be interpreted as
increasing the size of bank increases the income from dealing in foreign
currencies in bank. However, if this income from dealing in foreign currencies is
taken as a percentage of total income of bank, net assets does not affect this
income. It can be said on the basis of findings that Size of bank does not help in
earning extra income from dealing in foreign currencies by simply increasing the
size of bank.
Page | 56

Page 57

Findings on Income from dealing in foreign currencies and Currency
Derivatives used by commercial banks
The simple linear regression, as described in the methodology section is used with
the help of SPSS and below are the relevant outputs as produced by SPSS.

Model Summary

Model R R Square

Square
Std. Error of the

Estimate
1 .078a .006 -.003 6.48770060
a. Predictors: (Constant), Tools

Coefficientsa

Unstandardized Coefficients
Standardized
Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 4.920 .817 6.020 .000
Tools 1.019 1.250 .078 .815 .417
a. Dependent Variable: IFX % of TI

ANOVA
b
Model Sum of Squares df Mean Square F Sig.
1 Regression 27.955 1 27.955 .664 .417a
Residual 4545.748 108 42.090
Total 4573.703 109
a. Predictors: (Constant), Tools
b. Dependent Variable: IFX
Table xx: Regression output IFX on Tools

The above mentioned results show that b is not significant. Also the overall
regression is not significant at the 0.10 level of significance. There is a very weak,
almost no relationship between these two variables, as evident from the value of R
square. Since there is no relationship, therefore, null hypothesis substantiates and
it can be said that in Pakistan, income from dealing in foreign currencies is not
affected by using different tools.
Page | 57

Page 58

Findings on Effect of Exchange Rate Volatility on Income from Dealing
in Foreign Currencies
To examine whether Exchange rate volatility have any effect on the income from
dealing in foreign currencies, two separate models are constructed.
Result of Model with Income from dealing in foreign currencies in rupees
in ‘000
First model contains income from dealing in foreign currencies in Rs.‘000 as
dependent variable. Output produced by SPSS is presented below:

Model Summary

Model R R Square

Square
Std. Error of the

Estimate
1 .277a .077 .068 6.35607E5
a. Predictors: (Constant), Current Year Exchange Rate Volatility

ANOVA
b
Model Sum of Squares df Mean Square F Sig.
1 Regression 3.624E12 1 3.624E12 8.971 .003a
Residual 4.363E13 108 4.040E11

Total 4.726E13 109
a. Predictors: (Constant), Current Year Exchange Rate Volatility
b. Dependent Variable: Income from dealing in foreign currencies in Rs. ‘000

Coefficientsa

Unstandardized Coefficients
Standardized
Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 281336.798 91704.883 3.068 .003
Current Year
ERV
1036856.895 346183.103 .277 2.995 .003

a. Dependent Variable: Income from dealing in foreign currencies in Rs. ‘000

Table xxi: Regression Output IFXRS on ERV

Page | 58

Page 59

Using Income from dealing in foreign currencies in Rs. ‘000 in the simple linear
regression model showed a weak but significant relationship between the
independent variable and the dependent variable. R square in this model shows
that less than 7% variation is explained due to linear relationship between the
variables. Therefore, null hypothesis is rejected and alternate hypothesis is
accepted.

Result of Model with Income from dealing in foreign currencies relative to
total income
Results of the simple linear regression to check the impact of the Exchange rate
volatility on the income of the bank using income from dealing in foreign
currencies as a percentage of total income are below:

Model Summary

Model R R Square

Square
Std. Error of the

Estimate
1 .006a .000 -.009 6.50749413
a. Predictors: (Constant), Current Year Exchange Rate Volatility

ANOVA
b
Model Sum of Squares df Mean Square F Sig.
1 Regression .175 1 .175 .004 .949a
Residual 4573.528 108 42.347

Total 4573.703 109
a. Predictors: (Constant), Current Year Exchange Rate Volatility
b. Dependent Variable: IFX % of TI

Coefficientsa

Unstandardized Coefficients
Standardized
Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 5.310 .939 5.656 .000
Current ERV .228 3.544 .006 .064 .949

Page | 59

Page 60

Coefficientsa

Unstandardized Coefficients
Standardized
Coefficients
Model B Std. Error Beta t Sig.
1 (Constant) 5.310 .939 5.656 .000
Current ERV .228 3.544 .006 .064 .949
a. Dependent Variable: IFX % of TI
Table xxii: Regression Output of IFX on ERV

Results of the simple linear regression show that there is no significant
relationship between the independent variable and the dependent variable used in
this model. Therefore, Null hypothesis is substantiated and it can be said that
exchange rate volatility has no impact on the income of the income of the
commercial banks in Pakistan.

Page | 60