Document Text (Pages 21-30) Back to Document

Foreign Exchange Risk Management in Commercial Banks in Pakistan

by Sabri, Maroof Hussain, MS


Page 21

Limitation of Scope of Research

i. Foreign banks are not included in this study; hence, any findings do not
apply to Foreign Banks.

Data

Data used in this study is all of secondary nature. Annual reports of the
commercial banks are the major source of data along with various Statistical
Bulletins & publications by the State bank of Pakistan.

Foreign Currency Exposure of Commercial Banks in Pakistan

Foreign Currency Exposure of Commercial Banks gives an indication of the
foreign exchange risk assumed by bank.
Holding Net Asset or Net Liability in a currency gives rise to foreign currency
exposure into that currency. Banks hold multiple foreign currencies at a time and
sum of all the net positions in all the currencies held by a commercial bank is Net
Foreign Currency Exposure of the bank. This foreign currency exposure varies
from bank to bank and an attempt is made to understand different factors which
influence foreign currency exposure of a commercial bank.
To understand the foreign currency exposure of commercial banks in Pakistan,
below mentioned questions are designed:

1. Whether Commercial Banks in Pakistan take any foreign currency
exposure or not? If YES, does this exposure vary from bank to bank or
there is some fixed rule set to this?
2. Does there exist any relationship between Foreign Currency exposure &

other factors like Ownership Status (Type), Size of Bank and Exchange
Rate Volatility? Which of these factors is most significant?
3. Does Foreign Currency Exposure of Public Sector Commercial Banks

differ from than those of Local Private Banks?
4. Does Foreign Currency Exposure faced by Islamic Commercial banks

vary from those of Conventional commercial banks?
Answers to the above mentioned research questions can be achieved by using
different statistical and econometric techniques. The detailed methodology is
given below:
1. Net Foreign Currency Exposure
The very first research question is to check whether there is any Net Foreign
Currency Exposure assumed by the commercial banks in Pakistan. For this
purpose, Annual Financial Statements of listed commercial banks are studied. As

Page | 21


Page 22

per the statutory requirements, all the banks operating in Pakistan including
commercial banks have to mention in the NOTES to Financial statements “Net
Foreign Currency Exposure” in Pakistani Rupees, the calculated net position by
bank, under the heading of “Foreign Exchange Risk”. Whether this Net Foreign
Currency Exposure varies from bank to bank or there is a set rule for all the
banks? If a bank has zero Net Foreign Currency Exposure, it means it has all of
his assets and liabilities hedged and offset against other currencies or in the same
currency. It can be analyzed either relative to Total Assets or Net Assets of the
bank; however, it is more appropriate to analyze it with its relativeness to Net
Assets. Therefore, a new variable is constructed i.e. “Net Foreign Currency
Exposure relative to Net Assets”, denoted by “NFXNA”. Descriptive are studied
for this variable “NFXNA” i.e. Max, Min, Mean & Standard Deviation.
2. Factors that Affect Foreign Currency Exposure
Does this variable NFXNA depends on different factors like Ownership Status,
Size of Bank & Exchange Rate Volatility and which of these factors is the most
significant one?
Using Linear Regression to Explore Relationship between NFX and Size,
Ownership Status & Exchange Rate Volatility
The below mentioned Regression Model is used to find out the relationship
between Foreign currency exposure and the factors that influence it:

Where

����� = +
�������� +
���� +
������ +

Model 1: NFX depends on Size, Ownership & Exchange Rate Volatility

NFXNA
Size
OS
ERV
α
β

= NFX relative to Net Assets
= Size of the Bank
= Ownership Status of Bank
= Exchange rate volatility
= Population parameter, intercept
= Population parameter, slope, regression coefficient
Hypothesis Testing:
Using this model below mentioned hypothesis will be tested to check the
significance of the model as a whole along with individual βs as well.
Hypothesis 1:

Page | 22


Page 23

H0: There is no relationship between NFXNA & “Size, OS & ERV”
H1: There is a relationship between NFXNA & “Size, OS & ERV”
F-test using Analysis of variance is used to check the overall significance of
regression.

SPSS is used to analyze Regression & Partial Regression coefficients and their
significance is studied from its output. Significance of individual Partial
Regression coefficients will be checked using t-tests & overall significance using
F-test. The vales of statistics t & F should be significant at a level of significance
of less than 0.05.
Regression analysis used in understanding this relationship is backward and the
final model with one independent variable obtained after eliminating all the lesser
significant independent variables out of a total of three tells, if there is any, the
relationship between the most significant variable & the dependent variable.
Independent Variables:
Net Foreign Currency Exposure Relative to Net Assets:
Net Foreign Currency Exposure relative to Net Assets is calculated for the
comparison purpose. Different commercial banks have different Net Assets that
represent the size of bank and obviously different net positions. Net foreign
currency exposure is divided by Net Assets so that a comparison can be done
between different banks. Therefore, it shows Net foreign currency exposure of a
bank relative to its size, otherwise using only NFX in the model will show its
maximum dependence on Net assets of bank and the effect of other factors will be
not taken into account.
Net Foreign Currency Exposure is calculated by using the following formula:
��� = ������
�������� + �����������
��������

��������

± �������������������������������

Equation 1:Calculation of NFX

Net Foreign Currency Exposure is calculated by adding Net Open Position in all

Page | 23


Page 24

currencies held by a bank. All the banks have mentioned this NFX in the Notes to
Financial Statement and hence that is used.
����� = ���/��

Equation 2: NFXNA Calculation

Dependent Variables:

1. Size of Bank: Size of Bank can be measured either using number of
branches or the Net Assets (Net Assets = Equity). Since there are certain
banks in Pakistan having lower number of branches and their Net Assets
(or Equity) are higher than the Banks having higher number of branches.
Therefore, Size of Bank, for the purpose of understanding the relationship
under study, is measured by Net Assets. A further categorization could
have been done into small, medium and large (as done by the KPMG’s
Banking Survey 2009) which has not been done to check the true impact
of bank size on the net foreign currency exposure.
2. Ownership Status: Ownership Status or Nature of Ownership. Usually

there are three categories of banks operating in Pakistan i.e. Public sector
commercial banks, local private banks & foreign banks. As foreign banks
have been excluded, therefore, only two categories are left herewith. As
ownership status is a category variable, therefore, a dummy variable OS is
introduced in this model. Coding for the dummy variable is “0” for Public
Sector Commercial Banks & “1” for Local Private Banks.
3. Exchange Rate Volatility: Exchange Rate Volatility is used as a measure

to understand the fluctuations in the exchange rate. Standard deviations &
percent changes are amongst the several measures used for exchange rate
volatility (Mbutor, 2010). Exchange rate is measured in units of Pak
Rupees Vis-à-vis US Dollars. An increase in exchange rate shows
depreciation or weakening of Pak rupee whereas a decrease shows
appreciation or strengthening of Pak Rupee against US dollar. Exchange
rate volatility in this model is calculated by taking standard deviation of
changes in daily price fluctuations. While constructing this variable, it is
assumed for this particular objective that using the historical price changes
can be used for the next period and not for the current period. Therefore,
volatility of previous year is related to the current foreign currency
exposure of current year i.e. if and only if the volatility has had any
impact on the Banks’s Foreign currency exposure which is unhedged
position in foreign currencies, bank will adjust its exposure in the next
period. For example, if a bank has a policy to adjust its exposure by
Page | 24


Page 25

deciding whether to hedge or take a speculative position depends upon
volatility of exchange rate; it will take Exchange rate volatility of previous
year into account as historical data is used for the future forecasting.
Multicollinearity Problem
Multicollinearity can distort the results in the model; hence it has to be checked
before running the multiple regression on the model. Multicollinearity is the
existence of strong correlation between independent variables. In this model,
independent variables i.e. Size of bank, Ownership Status & Exchange Rate
Volatility may exhibit correlation with each other. A correlation matrix can be
drawn and the one of the independent variables exhibiting a correlation of 0.5 or
higher has to be dropped.
Relationship between Net Foreign Currency Exposure & Net Assets
Relationship between Net Foreign Currency Exposure & Net Assets can be
studied using Simple Linear Regression as below:

Where

��� = +
����

Model 2: Relationship between NFX & Net Assets

NFX = Net Foreign Currency Exposure
NA = Net Assets
α & β = Parameters of Model, intercept and slope respectively
In this model, NFX is dependent variable & NA is independent variable.
Hypothesis 2:
H0: There is no relationship between Net Foreign Currency Exposure and Net
Assets of Commercial Banks in Pakistan
H1: There is a relationship between Net Foreign Currency Exposure and Net
Assets of Commercial Banks in Pakistan
F-Test using ANOVA is used to check the significance of this regression. If the
value of F statistic is significant at a level of significant less than 0.05, it shows
that there is a significant relationship between both of these two variables or NFX
depends on Net Assets of bank.
3. Comparison of Net Foreign Currency Exposure of Public Sector &

Private Commercial Banks

Page | 25


Page 26

Public Sector Commercial Banks & Local private Commercial banks are two
different ownership statuses of banks in Pakistan. It has to be checked whether
there is any difference between these two types as far as their NFXNA is
considered.
For this purpose same variable as constructed previously, Net Foreign Currency
Exposure Relative to Net Assets (NFXNA), is used.
Hypothesis 3:
H0: The mean of NFXNA of Public Sector Commercial Banks is not significantly
different than that of Local Private Banks.
H1: The mean of NFXNA of Public Sector Commercial Banks is significantly
different than that of Local Private Banks
The above written hypothesis is tested using Independent Sample t-Test. For this
purpose, first equality of variances is tested using Levene’s Test. If the value of F-
Test is significant at a significance level less than 0.05, it shows that variances of
both groups are significantly different and if it is not significant, it shows that
variances are not significantly different. Finally t-test is used to check
independence of both groups and in this case if the value of t with the respective
degree of freedom is significant at a significance level of less than 0.05, it shows
that there is a significant difference between Net Foreign Currency Exposure
relative to Net assets of both types of ownerships of banks.

4. Comparison of Net Foreign Currency Exposure of Islamic &

Conventional Commercial Banks
With a continuous growth in the Islamic Banking in Pakistan, it is important to
check whether there is any difference between the Net foreign currency exposures
taken by them is different from than those of conventional commercial banks.
For this purpose same variable as constructed previously, Net Foreign Currency
Exposure Relative to Net Assets (NFXNA), is used.
Hypothesis 4:
H0: The mean of NFXNA of Islamic Banks is not significantly different than that
of Conventional banks.
H1: The mean of NFXNA of Islamic Banks is significantly different than that of
Conventional banks

Page | 26


Page 27

The above written hypothesis is tested using Independent Sample t-Test. For this
purpose, first equality of variances is tested using Levene’s Test. If the value of F-
Test is significant at a significance level less than 0.05, it shows that variances of
both groups are significantly different and if it is not significant, it shows that
variances are not significantly different. Finally t-test is used to check
independence of both groups and in this case if the value of t with the respective
degree of freedom is significant at a significance level of less than 0.05, it shows
that there is a significant difference between Net Foreign Currency Exposure of
both types of banks.

Page | 27


Page 28

Different Tools & Instruments Used by Commercial Banks in
Pakistan to Manage Foreign Exchange Risk

All the commercial banks face Foreign Exchange Risk and they have to use
different tools and instruments to manage this type of risk. Now this study has to
explore in a broader spectrum, by taking into account all the Listed Commercial
Banks, different tools and instruments used by the Commercial Banks in Pakistan
for the management of foreign exchange risk.
To accomplish this objective, answers to the below mentioned questions had to be
found out:
5. Whether commercial banks in Pakistan use any tool or instrument to manage

foreign currency risk? If “Yes”, what are the different tools used by these
commercial banks to manage foreign currency risk?
6. Whether all the commercial banks use the same tools to manage risk or there

exists any difference regarding the usage of these tools? And If usage of these
tools varies from bank to bank, does it depend on Ownership Nature (Private
or Public Sector), Type of Bank (Islamic or Conventional) Size of Bank,
Bank’s Foreign Currency Exposure and Exchange Rate Volatility?
5. Tools & Instruments Used by Commercial Banks in Pakistan
Financial Statements of commercial banks are studied to find out whether the
commercial banks use any tool or instrument to measure this foreign exchange
risk or not. Such instruments and tools are disclosed in the notes to the financial
statements of all the banks. To address this answer a detailed and thorough study
of financial statements of commercial banks is conducted and a list of the tools
and instruments used by the commercial bank is prepared. A detailed report on
such tools and instruments, with reference to their usage by the commercial banks
in Pakistan, is given in the Findings & Analysis Section. A descriptive analysis is
conducted for this purpose.
6. Factors Influencing Usage of Foreign Exchange Risk Management

Tools
Financial statements and their notes provide the answer to the question that
whether all the commercial banks use the same tools or not. What are the tools
that are commonly used by all the banks and what are the tools which are
particularly used by used by certain banks.
Usage of these tools, particularly currency derivatives, certainly depends on
different factors like size of bank, nature of ownership (Private or Public Sector),
Type of Bank (Conventional or Islamic) Exchange Rate Volatility and its Foreign
Currency Exposure.
Page | 28


Page 29

A. Currency Derivatives & Ownership Status of Banks
B. Currency Derivatives & Type of Bank (Conventional or Islamic)
C. Currency Derivatives & Size of Bank
D. Currency Derivatives & Foreign Currency Exposure
E. Currency Derivatives & Exchange Rate Volatility
These different bank specific (A, B, C & D) & macroeconomic (E) factors
influence the usage of currency derivatives by the bank. These different factors
can be studied using different ways.
For A & B: Usage of currency derivatives by banks, as far as ownership status &
type of bank is concerned (A & B), is studied through descriptive analysis and
further their relationship, currency derivatives as dependent variable, is studied
using binary logistic regression.
For C, D & E: Relationship between Size of Bank, as measured by Net Assets (or
Equity, since equity and net assets are equal so net assets are used), and Selection
of currency derivatives is studied using binomial regression. Similarly,
relationship between currency derivatives and foreign currency exposure &
exchange rate volatility is also studied using binomial regression.
Currency Derivative Usage by Type of Banks
Type of banks here refers to conventional or Islamic banks. With a continuous
growth of Islamic banking across the globe and in Pakistan and the Islamic banks
dealing in multiple currencies at the same time, it is important to check whether
the Islamic banks use currency derivatives or not. For this purpose descriptive
study is done.
Currency Derivative Usage by Ownership Status
To explore the usage of currency derivatives by the commercial banks in Pakistan
with respect to its ownership status, a descriptive study is done.
Relationship between currency derivatives and other factors
Relationship between currency derivatives usage and other factors like Size of
Bank, Ownership status of bank, type of bank, net foreign currency exposure
relative to net assets & exchange rate volatility is studied using Binary Logistic
Regression.
Predictors:
Predictors in Binary Logistic Regression Model are:

I. Ownership Status of Bank: This is a categorical variable, having two
categories Public Sector commercial bank and local private bank. Since, it
Page | 29


Page 30

has only two categories; it is entered as a single dummy variable with
values 0 and 1, 0 for Public sector commercial bank and 1 for Local
private bank.
II. Type of Bank: Type of bank is also entered as a dummy variable with two

categories Islamic and conventional, 0 for Conventional and 1 for Islamic.
This variable is entered in to the model to check whether changing the
type of bank results in the change in the usage of currency derivatives.
III. Size of Bank: Size of Bank is measured by the size of equity. Since,
equity is equal to Net Assets; hence, Net assets are used for this purpose.
However, Net Assets as entered in model are measured in billions rather
than in thousands to magnify its effect.
IV. Net Foreign Currency Exposure Relative to Net Assets: This variable, as
used previously, is entered as a continuous variable in this Binary Logistic
Regression. This variable is entered in the model as a percentage.
V. Exchange Rate Volatility: Current Year Exchange rate volatility is

entered as a continuous variable. However, in this model, it is entered after
multiplying by 100.
Dependent Variable
Dependent variable in this case is “Type of Derivatives used”. There are three
types of currency derivatives that are being used by Banks in Pakistan i.e. forward
exchange contracts, currency swaps & foreign currency options. As Forwards are
used by all the banks and other types are not common, hence, a dichotomous
variable is constructed to include in the model. This variable has only two
categories, one is the banks which use forward exchange contracts only and other
one is the banks who use swaps and options also other than forwards. Former is
coded with 0 and later with 1.
Binary Logistic Regression Model
As the dependent variable is a discrete categorical variable with two categories, a
dichotomous variable, therefore the most appropriate statistical procedure for
studying it as Binary Logistic Regression. Whether commercial banks in Pakistan
use only forward exchange contracts or use swaps & options along with these
forwards.
The Logistic Model is as below:
Log[p/1-p]=a+b1X1+b2X2+b3X3+b4X4+b5X5

Model 3: Binary Logistic Model for Currency Derivative Usage

Page | 30

© 2009 OpenThesis.org. All Rights Reserved.