# Foreign Exchange Risk Management in Commercial Banks in Pakistan

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 1^{a }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 2^{a }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

Step 3^{a }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

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

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

Model Summary

Model R R Square

Adjusted R

Square

Std. Error of the

Estimate

1 .764^{a }.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 .000^{a}

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.

Coefficients^{a}

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

Model 2:

Model Summary

Model R R Square

Adjusted R

Square

Std. Error of the

Estimate

1 .067^{a }.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 .488^{a}

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

Coefficients^{a}

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

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

Adjusted R

Square

Std. Error of the

Estimate

1 .078^{a }.006 -.003 6.48770060

a. Predictors: (Constant), Tools

Coefficients^{a}

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 .417^{a}

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

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

Adjusted R

Square

Std. Error of the

Estimate

1 .277^{a }.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 .003^{a}

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

Coefficients^{a}

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

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

Adjusted R

Square

Std. Error of the

Estimate

1 .006^{a }.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 .949^{a}

Residual 4573.528 108 42.347

Total 4573.703 109

a. Predictors: (Constant), Current Year Exchange Rate Volatility

b. Dependent Variable: IFX % of TI

Coefficients^{a}

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

Coefficients^{a}

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