Measuring Interest Rate Risk Considering Key Factors Affecting the Net Interest Margin (NIM) Ratio of Commercial Banks in Bangladesh

This paper attempts to reveal how the commercial banks operating in Bangladesh have measured interest rate risk using Interest sensitivity analysis along with considering the impact of key factors affecting the profitability measured with Net interest margin of banks under Bank specific as well as macro-economic environment. Interest sensitivity (IS) GAP analysis has been deployed to measure the degree of interest rate risk followed by a panel data regression model considering a comparative analysis among fixed effect within group, random effect GLS and Pooled OLS method adopted to measure the causation between Net interest margin and Bank specific along with Macroeconomic factor to accomplish the objective of this paper. In IS GAP along with relative IS GAP analysis, two banks are found to be liability sensitive posing a risk of reducing the net interest margin (NIM) if interest rate has been increased and rest of the banks are found to be asset sensitive again postulating risk of reducing the NIM if market rate interest has been decreased considering the ten years’ data regarding rate sensitive assets (RSA) and rate sensitive liabilities (RSL) of four commercial banks selected using convenience sampling approach. Moreover, Panel Data regression model depicts how several key factors such as degree of risk aversion, credit risk and quality of management, Average operating cost, size of banks, implicit interest payments may significantly affect this NIM ratio measuring the profitability of banks in Bangladesh followed by several diagnostic tests such as model specification test using hausman & LM test, multicollinearity test, heteroscadisticy test and unit root test conducted to check the validity of models.


Introduction
The commercial banks of Bangladesh influence a massive part in financial prosperity of any country. Rate of interest performs a key role in a country so that it will be more important for a country to maintain the fluctuation of rate of interest in an appropriate manner through monetary policy declared by Central Bank of a country giving its full heed in the rate of interest. Conventional banks of Bangladesh perform role of supervising and maintaining the economic sector of a country that leads to the financial prosperity of a country as a whole. Interest Rate Risk (IRR) has been termed as the increase or decrease in a value of the portfolio of the bank owing to fluctuations of the rate of interest. The term is considered as the adverse condition of a financial firm because of the unfavorable reflection in the rate of interest. The term is connected with the influences of alteration in the overall financial condition. If the rate of interest decreases, then the worth of interests on deposits also gets reduced. On the other hand, the worth value of the portfolio of liabilities increases. On the other hand, the worth value of the loans and financial instruments decrease if interest rate increases Interest Rate Risk occurs because of dissimilarities of the time frame of liability & asset. Change can occur in the IRR owing to change in the structure of asset and liabilities, amount of asset and liabilities maturity of portfolios of asset and liabilities, rate of sensitivity of asset & liabilities and the asset quality.

Background of the Study
Interest Rate Risk is one kind of risk that originates for bondholders from the fluctuation of the interest rates. Interest Rate Risk is termed as a mislaying ensuing of an unfavorable change on the cash flow and from an adverse alteration on worth of interest liabilities along with interest asset, for consequence of an increase or a decrease in the rate of interest. Private commercial banks always try to maximize the profitability of financial institutions. They try to confirm adequate amount of liquid assets to pose the firm confidence, that of the deposit makers on the expertise to serve the money provided by the depositors by giving interest in time. In order to obtain the above mentioned things, the real important thing for banks are to supervise, look after and manage the portfolio of liabilities and assets through a methodical system by receiving the consideration that different sorts of risk factors are engaged in the following arenas. The system of maintaining the financial condition has been a systematic decision making of the commercial banks in order to adjust or cope up with the adverse changes in the factors of risk involved. Interest Rate Risk Management of any financial firm mainly looks to reduce the risk generating factors assigned with a financial firm's overall portfolio of liability & asset Interest Rate Risk analysis is executed through Duration Gap Analysis, Interest Sensitive Asset and Liability Analysis and Relative IS Gap Analysis. Interest Rate Risk analysis is coordinated by moving and shifting overall amount and volume of liabilities and assets of banks. The following research attempts to implement various techniques to be used in order to measure the dissimilarities of durations of asset and liability provided a sensitivity ratio and relative IS gap analysis in order to measure the Interest Rate Risk followed by an econometric modeling on the causation between NIM ratio and Bank Specific along with macroeconomic factor.

Literature Review
Many have discussed about the management of Interest Rate Risk. Basically they have followed various aspects in order to mitigate the Interest Rate Risk. Different researchers have provided different definition of Interest Rate Risk in the different aspect of banking. Beutler T. (2015) said that alteration of the rate of interest affects the financial value of portfolio of banks because this change affect the discount factor which is followed to convert the FV into PV. The financial condition of the bank also changes due to change in the net cash inflow and net cash outflow. Dash and Pathak (2011) has used a model in order to assess the liability and asset of any firm. They have observed that the government regulated banks are more successful in maintaining liquidity position than the private commercial banks. Ahmed Anwer S., Beatty Anne and Carolyn Takeda (1997) have said that the Interest Rate Risk is connected to the dissimilarities of duration between the interest sensitive asset and liability which is ISA and ISL.
Carter David (1998) has said that the interest rate derivatives can be used in order to reduce the exposures of Interest Rate Risk in the commercial banks. He has determined the Interest Rate Risk by taking maturity gap of 12 months into consideration. Deng Ran (2011) has provided some advices for the banking sector of China about the precautions regarding the prevention of Interest Rate Risk. Dhanani A. (2007) has found that the volatility of interest rate, the usage of corporate bonds affect the Interest Rate Risk Management of United Kingdom. Galai et al. (1999) has defined that the business environment of any banking organization can put an impact on the Interest Rate Risk. Bank management has been considered as a major factor in order to mitigate the Interest Rate Risk. Haslem et al. (1999) has said that the large companies with a big amount of capital tries to be more efficient while disbursing the loanable fund and put emphasis on the matching strategy of asset and liability. According to P.Allen (2000 Interest Rate Risk is lesser at community banks than in the thrift firms. He has considered the increasing asset duration as well as the excessive usage of more volatile liabilities is a primitive factor which drives to the increasing interest rate. The security portfolios are not used by banks in order to reduce this risk. Rajan and Nallari (2004) has said that more of the banks in India are liability managed. That's mean the banks are willing to manage the liabilities rather than the assets. Reeta (2013) has analyzed that in India, the rate of interest risk is regarded as the most vital risk factor for the banking institutions. It is also found that maximum institutions in India use the various gap analyses.
Shashi Srivastava (2015) measured Interest Rate Risk by using almost all sorts of methods like sensitivity analysis and duration gap analysis and has provided an advice to use derivatives in order that the firms can minimize the risk factors. For measuring the Interest Rate Risk, the banks use various techniques like gap income analysis, duration gap of asset and liability analysis, projection of net interest income, risk and return composition analysis, IS ratio analysis (Williamson, 2008). Interest Rate Risk has two dimensions namely; price & reinvestment rate risk. The risk of price is that the change within the worth of bond owning to alterations in the rate of interest. The effective analysis for price risk is that the level of volatility in the price that is captured in a summary determined by duration, (Bierwag & Fooladi, 2006). (Zhou.Y. and Zheng.X. ,2017) have used two factors in calculating interest rate sensitivity gap. Those are sensitivity gap and sensitivity co-efficient.

Research Gaps
As the literature has been reviewed in order to conduct the research, it has been concluded that none of any researcher has shown comparison of four banks (Basic Bank Limited, Dhaka Bank Limited, Mercantile Bank Limited and Eastern Bank Limited) simultaneously in order to show the interest rate risk considering the factors such as-NIM (Net Interest Margin), IS Gap, Relative IS Gap as well as IS Ratio followed by the formulation of econometric modeling revealing the impact of Bank specific as well as macroeconomic factor on NIM ratio measuring profitability of commercial banks.

Objectives of the Study
This paper imparts at revealing the causation between NIM and Bank specific along with macro-economic factors to demonstrate how the net interest income of banks has been affected with several factors considering micro and macro environment, finding the Interest Rate Risk management of commercial banks, estimating sensitivity of liabilities and assets of commercial banks and finding out the Interest Rate Risk position of commercial banks using IS GAP analysis for 10 consecutive years' data.

Research Design
A research design identifies the overall structure of the research. A research design generally explains about how the information has been collected, what and how the tools will be conducted and analyzed. The proper tools are required to get the actual outcome from the research. 6.1.1 Type of Research The first part of this paper has adopted only the quantitative approaches for this research study using descriptive research formed in order to present the features of measuring interest rate risk using IS GAP, relative IS GAP and IS ratio of commercial banks. The second part of the paper is an explanatory research showing the causation between NIM being dependent variable and several Bank specific factors being explanatory variables along with macro variables to control the impact of exogenous factors.

Data Type and sample selection procedure
We have adopted secondary sources of data since the last 10 years collected from annual reports of 04 (four) commercial banks listed below and selected using non-probabilistic convenience sampling approach depending on the availability of data from Basic Bank Limited, Mercantile Bank Limited, Dhaka Bank Limited and Eastern Bank Limited. So, the sample size is 40.

Data Analysis Tools
In order to determine and observe the numerical data which is related with Interest Rate Risk Management, various sorts of methods and procedures have been followed to get the accurate result from the calculation. For instance, in order to measure the NIM, Interest Sensitivity Ratio, IS Gap, Relative IS Gap and Duration Gap, we have taken the MS Excel as an effective tools and made graphical presentations to represent the information in a much more effective manner. 6.3.1 Analysis of Ratio: We have done two sorts of ratio analysis. These two sorts of analysis include: a. Net Interest Margin Ratio. b. ISA (Interest sensitive assets) to ISL (Interest sensitive liability) Ratio.

6.3.2
Analysis of Gap: We have executed two sorts of gap analysis. These two sorts of analysis include: a. Relative Interest Sensitive Gap Analysis b. Interest sensitivity (IS) Gap Analysis.

6.3.3
Development of Hypothesis and Econometric modeling: Following hypothesis has been developed in order to divulge the impact of bank specific as well as macro factor on Net interest margin (NIM) of commercial banks: Ho: There is no significant relationship between NIM and Bank specific along with macro-economic factor. H1: There is a significant relationship between NIM and Bank specific along with macro-economic factor.
In addition, following econometric modeling has been used to test the hypothesis divulging the causation between NIM and other bank specific and macro factors using the software package STATA 12: Here, NIM= net interest margin measured by net interest income to earning assets ratio X1= AOC standing for average operating cost measured with operating expense to total assets ratio has been used as a proxy for showing the operating efficiency of a bank. X2= DORA is a proxy variable showing the Degree of bank's risk aversion measured with capital ratio calculated by dividing total equity by total assets. X3=CR standing for credit risk measured with the ratio of loan loss provision to total loans. X4= SIZE used as a proxy for measuring volume of operation calculated with natural logarithm of total loans X5= IIP standing for implicit interest payments calculated by dividing net non-interest expenditure by total assets of respective banks X6=OCBR standing for opportunity cost of holding bank's liquid reserve estimated by the ratio of cash due from other banks to total assets European Journal of Business and Management www.iiste.org ISSN 2222-1905 (Paper) ISSN 2222-2839 (Online) Vol.12, No.15, 2020 71 X7= QM stands for quality of management measured by the ratio of cost to income of banks used as a proxy for managerial efficiency of banks. Y1= INF stands for inflation measured with average inflation rate per year since 2009 used as a proxy for macroeconomic factor. We will adopt pooled OLS ( ordinary least square) fixed effect within the groups as well as random effect GLS (generalized least square) regression methods to estimate the coefficients of the models developed earlier.

Data Analysis and Discussion:
Institutions such as banks need to determine the Interest Rate Risk frequently because wrong determination of the risk can bring about a great loss for the firm. That is why they need to be more proactive while assessing the risk. They need to use right sorts of risk measurement tools in order to mitigate the risk. Some banks use only one sort of risk mitigating tool and others try to use more methods so that they can get much more accurate result. The main Interest Rate Risk measurement techniques are as follow:

Management of Interest Sensitive (IS) Gap
The analysis of IS Gap is very common sort of analysis that determine the dissimilarities of the value of RSA and RSL for a specific period of time. A liability sensitive gap happens if the ISL is greater than ISA. If the liability sensitive gap occurs, the firm declines the rate of interest.
An asset-sensitive gap happens if the worth of ISA is greater than ISL for a particular period of time. To handle this sort of scenario, the bank increases the rate of interest because it will enhance spread of interest.  Needs to increase the ISA or to reduce ISL.

Net Interest Margin (NIM) Calculation
The NIM is termed as a profitability parameter which shows prospective analysis regarding a firm's decisions regarding the capital budgeting and investment.

Calculation of IS Gap
The IS Gap is calculated by subtracting ISL from the ISA for a specific time period. A bank is considered asset sensitive if the bank's ISA is greater than that of ISL. Besides it is regarded as a liability sensitive if the bank's ISL is greater than that of ISA or the ISA is less than the ISL. So, it can be said that IS Gap= ISA-ISL   In the year of 2009 to 2018, the status has been liability sensitive which means that the ISL was greater than ISA. When the status is liability sensitive, if the rate of interest rises, the NIM falls and if the rate falls, the NIM increases. The graph is showing that the gap has been in a decreasing trend and it has been falling more than usual. The reason can be that the bank is has not been taking its investment decisions correctly.

Calculation of Relative IS Gap
The relative IS Gap can be determined by dividing the Interest Sensitive (IS) gap with Total Asset of the bank or any financial institution. Thus, if the IS Gap shows a positive result, the Relative Interest Sensitive Gap will show a positive result and vice versa.

Data source: Annual reports of DBL (FY 2012-FY 2018)
The bank or financial firm is termed as asset sensitive provided the Relative Interest Sensitive Gap>0, which means it will provide a positive result. On the other hand, the bank or financial firm is regarded as a liability sensitive provided the Relative Interest Sensitive Gap<0, which means it will provide a negative result. The table shows the result of Relative IS Gap of DBL from the year 2009 to 2018. It shows that in all these years Relative IS Gap has been negative for DBL. On the basis of these information, it can be concluded that DBL is more like an institution with liable sensitivity.

Calculation of Interest Sensitive (IS) Ratio
IS Ratio can be determined by dividing the ISA with the ISL. This indicates the efficiency of Interest Rate Risk Management. If the ISA is greater than the ISL, it will bring a positive result. Besides, if the ISA is lesser than the ISL, it will bring a negative result.

Comparative Discussion
For a comparative discussion we have chosen another three banks to observe the position of DBL. These banks are Basic Bank Limited, Mercantile Bank Limited and Eastern Bank Limited. A comparison has been made among these banks. Thus the competitive positioned will be figured out. Comparison will be made on the basis of NIM (Net Interest Margin), IS Gap, Relative IS Gap and IS Ratio. Organizations. As Basic Bank Limited and Dhaka Bank Limited are Liability Sensitive, they will be in better position if the interest rate decreases and they will be in a spot of bother if the interest rate increases. Mercantile Bank Limited and Eastern Bank Limited on the other hand, will be beneficial if the interest rate increases and they will face problems if the interest rate falls. Among these four banks Mercantile Bank Limited poses the best position as its IS Gap is lesser than the rest of the banks and the results are more consistent compared to other banks. The gap between the interest sensitive assets and interest sensitive liability is huge for Dhaka Bank Limited. It means that it is not efficiently managing its ISA and ISL.

Data source: Annual reports of four Banks (FY 2009-FY 2018)
The graph shows that the relative IS gap of Basic Bank Limited and Dhaka Bank Limited is more than zero. It means that these two banks are liability sensitive. On the other hand, Mercantile Bank Limited and Eastern Bank Limited are Asset Sensitive Organizations as their relative IS gap is less than zero. As Basic Bank Limited and Dhaka Bank Limited are Liability Sensitive, they will be in better position if the interest rate decreases and they will be in a spot of bother if the interest rate increases. Mercantile Bank Limited and Eastern Bank Limited on the other hand, will be beneficial if the interest rate increases and they will face problems if the interest rate falls. Mercantile Bank Limited has the best position among these banks as the relative IS gap is more close to zero and the relative IS gap is more consistent than that of other banks. The Relative IS Gap of Dhaka Bank Limited is inconsistent and not that much great as compared to other banks. 7.6.4 Comparative IS Ratio of Banks Graph 08: Interest Sensitivity (IS) Ratio of Banks

Data source: Annual reports of four Banks (FY 2009-FY 2018)
As the IS Ratio of Basic Bank Limited and Dhaka Bank Limited is less than 1, so these banks can be termed as Liability Sensitive banks. On the other hand, Mercantile Bank Limited and Eastern Bank Limited has the IS ratio of more than 1. That means these banks are Asset Sensitive. Among these banks, Mercantile Bank Limited has the best position as the IS Ratio of this bank is more close to 1 compared to other banks. The IS Ratio of Dhaka Bank Limited is not that much great.

Econometric modeling with Random Effects, Fixed Effects and Pooled OLS
According to the coefficients estimated by Random Effect GLS (generalized least square method) method as per following table, only credit risk and quality of management followed by cr and qm variables respectively are found statistically significant in explaining the variation of net interest margin ratio followed by nim being dependent variable to reveal the changes in net interest income of banks. Credit risk measured with the ratio of loan loss provision to total loan of bank is found positively related with net interest margin ratio as higher credit risk pushes the bank to increase the net interest margin in order to cover the excess provision kept against the high probability of default loan. Moreover, quality of management measured with the ratio of Cost to Income of bank is found inversely related with net interest margin ratio because high ratio of cost to income reveals operating inefficiency of banks propelling into high NIM to cover the operating cost of banks. All other variables are found statistically insignificant in explaining the variation of net interest margin of bank. The overall R square value of 0.6935 divulges that 69.35% variability in net interest margin has been explained by the fitted regression model estimated by Random Effect GLS method. The Chi-square value of 70.15 is also found statistically significant considering the assumption that all explanatory variables such as average operating cost, degree of risk aversion, credit risk, size of operation, implicit interest payment, opportunity cost of holding bank's liquid reserve, quality of management and inflation are jointly significant in explaining the variation of net interest margin of bank as depicted from the following table: Apart from the credit risk and quality of management, degree of risk aversion and opportunity cost of holding liquid reserve of banks followed by dora and ocbr respectively are also found statistically significant in explaining the variation of net interest margin ratio followed by nim being dependent variable to show the changes in net interest income of banks as per the output revealed by Fixed effect regression model. The degree of risk aversion measured by capital ratio calculated with dividing equity capital by total assets of bank is found positively related with net interest margin as banks increase the NIM to cover the higher cost of equity financing compared to debt financing. In addition, Opportunity cost of holding bank's reserve measured by dividing cash due from other banks by total assets is found inversely related with net interest margin ratio as higher liquid reserve increases the opportunity cost for banks and thus reduces the profitability of banks measured by NIM. On the contrary, all other variables are found statistically insignificant in explaining the variation of net interest margin of bank. The overall R square value of 0.6047 divulges that 60.47% variability in net interest margin has been explained by the fitted regression model estimated by Fixed effect method. The F -value of 7.84 is also found statistically significant considering the assumption that all explanatory variables such as average operating cost, degree of risk aversion, credit risk, size of operation, implicit interest payment, opportunity cost of holding bank's liquid reserve, quality of management and inflation are jointly significant in explaining the variation of net interest margin of banks. The rho value also known as intra-class correlation value of 0.6287 reveals that 62.87% variability in NIM is explained by the differences across panels. Moreover, there is no strong evidence that the model suffers from endogenity problem as the correlation value between residual within groups and the regressors (explanatory variables) is found 0.0941.
According to the output estimated by Pooled OLS method, only credit risk and quality of management variable are found statistically significant in explaining the variation in net interest margin of banks which resemble the output estimated by Random Effect method as described earlier.

) Using Hausman test (Random effect vs Fixed effect)
According to the output of Hausman test mentioned below to determine between fixed or random effects method, the null hypothesis is that the preferred model is random effect vs the alternative is fixed effect (Green 2008). In fact, it tests whether the unique errors followed by ui are correlated with regressors, the null hypothesis is they are not. As the Chi-square value of 31.73 is statistically significant at 1% level of significance, we can reject the null hypothesis and conclude that Fixed effect model is preferable to Random effect model.

) Breusch and Pagan Lagrangian multiplier test (Random Effect vs Pooled OLS):
This LM test suggesting decide between a Random effect and Pooled OLS regression has assumed the null hypothesis is that variance across estimates is zero which means there is no significant difference across units (i.e. no panel effect). According to the Chi-square value of 14.17 being statistically significant, we can reject the null hypothesis and deduce that there is a significant difference across the panels suggesting Random effect is better estimates than Pooled OLS as per the following output: None of the correlation is superior to 0.80 suggesting that any model doesn't suffer from the problem of multicollinearity as depicted from the following table:   2222-1905(Paper) ISSN 2222-2839(Online) Vol.12, No.15, 2020   We have adopted LLC test standing for Levin-Lin-Chu unit root test to know whether the mean, variance and covariance of series are stationary assuming the following hypothesis: -H0: The series is non-stationary or it has a stochastic trend -H1: The series is stationary or has a non-stochastic trend Decision Rule: Reject H0 if the p-value of unit root tests less than significance level. Otherwise, do not reject H0. So, the adjusted t-value of -2.5051 is statistically significant suggesting that we can reject the null hypothesis and conclude that the dependent variable NIM is stationary as per the following table:  2222-1905(Paper) ISSN 2222-2839(Online) Vol.12, No.15, 2020 sorts of banks will be benefited if the rate of interest rises.  The gap between ISA and ISL is huge for Dhaka Bank Limited compared to other banks. The gap between ISA and ISL is increasing gradually which indicates that there is a lack management of interest sensitive assets and liabilities.  As Dhaka Bank Limited and Basic Bank Limited are Liability sensitive organizations, their relative IS Gap is less than zero. The relative IS Gap of Mercantile Bank Limited and Eastern Bank Limited are greater than zero as they are Asset Sensitive organizations. The relative IS Gap of Dhaka Bank Limited is very far from zero which is not a good sign compared to bank like Mercantile Bank Limited and Eastern Bank Limited.  The IS Ratio of Dhaka Bank Limited and Basic Bank Limited are less than 1 which means that they are Liability Sensitive Organizations. The banks like Mercantile Bank Limited and Eastern Bank Limited are Asset Sensitive Organizations as there is Ratio is more than 1. The IS Ratio of Dhaka Bank Limited is also not that much satisfactory compared to bank like Mercantile Bank Limited and Eastern Bank Limited because its IS Ratio is far less from 1 which also indicates that it is not managing its ISA and ISL properly.  Moreover, Degree of risk aversion, credit risk and quality of management are found individually statistically significant in explaining the variation of Net interest margin of these banks although all variables mentioned earlier are found statistically jointly significant in explaining the dependent variable measured with Net Interest Margin ratio of banks under Fixed effect, Random effect and Pooled OLS modeling revealing the causation between NIM and Bank specific along with macro-economic factors.

Conclusion
This paper has accomplished the objectives set at earlier stage by revealing how the changes of interest rate affect the Net interest margin (NIM) ratio which is one of the key profitability ratios of commercial banks due to the position of IS GAP followed by asset sensitive when RSA exceeds RSL and liability sensitive when RSA deficits RSL of balance sheet of four local commercial banks operating in Bangladesh. Although two banks consisting of Dhaka Bank and BASIC Bank out of four banks are found liability sensitive meaning liability will be affected more compared to assets when interest rate moves, the IS Gap of Dhaka Bank Limited is very far from zero which is not consistent compared to Mercantile Bank Limited and Eastern Bank Limited due to the managerial inefficiency of rate sensitive assets & liability of Dhaka Bank Ltd. In addition, the variation in NIM ratio of these banks is also explained either individually or jointly by several bank specific as well as macro-economic factors as described earlier depending on the estimation of coefficients using fixed effect, random effect and pooled OLS model in order to divulge the causation between NIM and these factors.