Impact of Financial Risks on the Pro tability of Commercial Banks in India

The Indian banking sector is exposed to various types of risks which arise from both the external and internal environments. Banks long-term sustainability and  nancial feasibility are vulnerable  nancial risk. Credit risk, operationalrisk, marketrisk, and liquidity risk stances a major challenge, despite growth in the banking system. This study examines the relationship between pro tability and  nancial risks of 43 Indian commercial banks for the period of 11 years, (2008 to 2018). The quantitative research design was adopted in this study and the pro tability measures that have been used in this study are the Return on Assets (ROA) and Return on Equity (ROE) while the  nancial risks are Interest Rate Risk (IRR) and Foreign Exchange Risk (FER). In this study, TimeSeries Cross-Sectional secondary balanced panel data regression analysis of  xed effect and random effect model have been implemented. The  ndings of the study indicated that the relationship between ROE and IRR were found to be weakly signi cant, and on ROA the effect of IRR is signi cant for all the commercial banks. On both pro tability measures, the FER was found to have an insigni cant impact. The study concludes that there exists an inverse relationship between banks pro tability and  nancial risk. Hence, the commercial banks in India together with the bank supervisors should make a trade-off between pro tability and  nancial risk.


Introduction
Profi tability is the ultimate test1for the effectiveness1of risk management.1It is the bottom-line of any fi nancial1institutions. After knowing the fi nancial risk impact on the bank's profi tability, it would be the most crucial aspect for all the banks as it would give heads-up to the bank to mitigate those risk effectively. Likewise, a profi table and healthy banking systempromote comprehensive fi nancial fi rmness and perceive to raise the economy's pliability to adverse macroeconomic surprises. Between risk and return the tradeoff is well recognized -the higher return comes with higher risk and viz versa. Therefore, in order to expand business and to increase profi tability, fi nancial institutions should be aware of the risk factors which have a major impact on profi tability measures. Moreover, it's a known fact that the amount of risk faced by fi nancial institutions is a great concern and is of a signifi cant nature to the policymakers. The Basel committee report also highlights the importance of studying bank risks (BCBS-BIS 2001) 1 and the Central bank's ongoing and consistent effort to record it in the capital adequacy guide lines (Shukla 2013).

International Journal of Management
The present study focuses primarily on fi nancial risks such as IRR and FER related to Indian commercial banks. Despite the fact that banks face various types of risks, these risks stand out and are often related to one another. "The interest rate isoften the trigger for other forms of risk" (Narayana and Mahadeva 2016).

Review of Literature
This study describes the external and internal factors that affect the commercial banks' profi tability. It forms thebasis for the development of themodels in the present study by the impact of riskson banks profi tability measures. The relationship between the bank and net interest margins (NIM), IRR, default risk,andoff-balance sheet (OBS) banking activities of US banks between 1989 and 2003 were sampled by (Angbazo 1997). The pooled sample result in the documents like management effi ciency, non-interestbearing reserves, default risk, and leverage are associated positively with bank's interest spread and the European bank's profi tability during the 1990s was explored by (Goddard, Molyneux, and Wilson 2004). (Muriithi 2016)examined the relationship between fi nancial risk and banks profi tability and the impact of the fi nancial risks on the commercial banks' profi tability in Kenya. The fi ndings of this study exhibited that, the operational, liquidity, market and credit risks have a signifi cant negative impact on ROE. A cost to income ratioof the component of fi nancial risk that had the most impact on fi nancial performance and she concludes that there exists an inverse relationship between performance and fi nancial risk of Kenyan commercial1banks. (Narayana and Mahadeva 2016) made an attempt to identify the various types of risks handled by the banks and the risk management process. They also examined the different tools adopted by the banks for mitigating the risk. (Shukla 2013) explored the various indicators to evaluate the changes in the solvency position and capital structure of banks for highlighting risk profi le of Indian banking system and in detail the risk profi le of top ten private and public sector banks. (Tafri et al. 2009)examined the Islamic and Malaysians conventional banks' relationship between fi nancial risks and profi tability measures for 10 years between 1996 to 2005. They employed a PDR analysis of GLS of FE and RE models and conclude that the relationship between ROE and IRR were found to be weakly 1signifi cant for the conventional1and1insignifi cant for the1Islamic banks. The impact of IRR on1ROA is signifi cant for1the conventional1banks. Liquidity1risk (LR) was found to have an1insignifi cant1impact on both1profi tability1measures. (Driga 2012) 2 focuses on measuring the performance of Romanian banking systems of a commercial bank to fi nancial risks. ) examined that, the OBS activities includes contingent indentures which produce income to a bank but are considered neither as sources of fund nor application of funds as per conventional accounting method. Contingent items may be considered as notes to balance sheet, invisible banking, contingent commitment banking or even asset less banking in banking records. (Hegde and Subramanian 2016)This work studies the current risk management practices of Indian banks and their adherence to Basel norms. (Aktan, Chan, and Evrim-Mandaci 2013) examined the effect of OBS activities on the bank'sprofi tability, listed on the Istanbul Stock Exchange. In this study, four performance measures were used i.e. bank's liquidity position, profi tability, risk exposures, and leverage. The OBS activity results indicate that banks stock returns have been improved due to its hedging perception, but have a negative impact on ROE. Furthermore, they conclude that the OBS activities of the banks do not have a statistically signifi cant infl uence on banks liquidity position.
Though, there are few studies which examined the relationship between IRR & the NIM and also the IRR and effectiveness relationship of the banks. (Tafri et al. 2009)examined that interest rate unpredictability has a positive impact on NIM. (Angbazo 1997) found a mixed result for the IRR and NIM relationship. (Muriithi 2016) as per this study, there is also a mixed result between operating effi ciency of the bank and IRR. Hence, from this literature, it is not clear that whether it will be a positive or negative impact on banks profi tability measures. The gap of the study is there is no specifi c 2 Financial Risks Analysis for a Commercial Bank in the Romanian Banking System. Annales Universitatis Apulensis : Series Oeconomica S International Journal of Management literature to discuss about the impact of IRR & FER on the profi tability of the bank. This study describes how these two major risks are going to make an impact on profi tability of the commercial banks in India.

Research Objective
• To study the relationship between Financial Risks and Profi tability of the commercial banks • To examine the impact of fi nancial risk on the profi tability of the commercial banks in India

Data Analysis and Empirical Framework Sources of data & Methodology
The secondary data for this study was collected through the audited fi nancial reports and annual reports of the Indian commercial banks from banks website and RBI Time series publications (Statistical Tables Relating to Banks in India). The study period contains 11 years data between 2008 to 2018, because all banks complete data were available during these periods. The above data set comprises of 43 Indian commercial banks which includes both private and public sector banks. Hence, this pool aggregated data comprises a total of 473 (43*11) observations. For the present study Panel Data Regression analysis technique is considered because of its many advantages over either cross-section or time series data (Paul 2012). Firstly, by combining time series and cross-sectionobservations, more informative data can be collected through panel data with more variability but less collinearity among the variables. Furthermore, "it provides an augmented number of data points and hence produces additional degrees of freedom as well as more effi ciency" (Paul 2012). Thus, for the present study it is appropriate asit increases the quality andquantity of data whereby the timeseries is short (11 years) and also the number of banks are fewer. Secondly, by integrating the information relating to variables cross-section and time series, heterogeneity is explicitly taken into account by consent for specifi c individual variables(Regression 1991) 3 , 3 Regression, P. D. (1991). Panel Data Regression Models, 591-613 According to (Gujarati 2004) 4 , "Panel Data analysis suggest that individuals, fi rms or countries are heterogeneous, if heterogeneity is not controlled,there is the possibilityof running into the risk ofobtaining infl uenced results". Thirdly, "by integrating data pertaining toboth crosssection and timeseries variables, it can signifi cantly reduce the problems that may arise from omitted variables"(Baltagi 2014) 5 .PDR is chosen over the Ordinary Least Square method (OLS) because under certain assumptions, PDR will turn out to be more competitive compared to OLS (Gujarati 2004 (Baltagi 2014).

Fixed Effect Model
The FEM is also termed as a Least1 Square1 Dummy1 Variable model (LSDV). In this model, it is assumed that the "coeffi cients are constant and time-invariant" (Gujarati, n.d.).

The basic equation for this model is as follows:
Yit=α i+ β iXit+μit (1)1 Here, Yit= a dependent variable for banks measures of profi tability.

Dependent Variables
In this study, ROA & ROE are measures of profi tability, while a measure of spread is the NIM, and the dependent variable is selected as profi tability. These measures are preferred based on the literature (Tafri et al. 2009), (Muriithi 2016), (Angbazo 1997). ROE measures profi tability from the shareholder's viewpoints while ROA measures the bank management's ability to make a profi t from the bank's assets and it is defi ned as the ratioof net income to an average of total assets and it measures banks profi tability per rupee of assets and another dependent variable ROE measures banks accounting profi t per rupee of equity capital and hence, ROE is defi ned as net income divided by average equity.

Independent Variables
The independent variables namely IRR, FER and OBS activities are considered on the basis of its latent relevance to this model as well as for this study, and also because of its importance in representing a bank's real fi nancial situation.
IRR: The maturity gap is proxy for IRR, which is derived by the ratioof the difference between the rupee value of assets and liabilities which is repriced within a year to total capital (Driga 2012).
The following items like money market deposits accounts, loans maturing within a year, variable rate deposits, marketable securities maturing within a year and fl oating rate loans are all considered as rate sensitive (Fleeson et al. 2017) while cash, cash equivalent, liquidity reserves, assets and liabilities physical in nature such as owners' equity and longterm loans are the non-RSA and non-RSL (Tafri et al. 2009). As we have not come across any prior prospect studies conducted on the effect of IRR on profi tability.
FER: The proxy for FER is Net Foreign Currency exposure between assets & liabilities to Total Assets.
OBS: It can be divided into a lending product or credit-related products and risk management derivative product. lending products such as loan commitments and letters of credit.

International Journal of Management
Risk management derivative products such as forwards, futures, options and swaps (Baxter et al. 2008) 6 (Angbazo 1997). The OBS activities are embodied by the ratioof OBS to total assets ). However, in this study, a testable implication is that the independent variable OBS activities should improve the profi tability of the banks, because they authorize banks to investments in risky projects that would be passed up if restricted to equity or deposit fi nancing. However, it would lead to greater exposure to risks if the OBS activities are increased (Chaudhry, n.d.2009).

Controlled Variables
In order to segregate the impact of risk factors on the performance, it is very important to control the other factors which have a marginal infl uence on profi tability. Some controlled variables are included in this study, based on the literature, where it was stated that they have a signifi cant association with profi tability. The following are some important controlled variables which are likely to infl uence the bank's profi tability.

Bank Size
Ayear-end log of total assets are being used to measure the size of the bank. (Tafri et al. 2009)study also supports that credit risk exposure is size related and large banks always have the advantage of lower credit risk. In this study, with relation to profi tability measures, bank size is expected to have a positive relation.

Bank Capital
As per the study by (Shukla 2013), this variable is represented by the bank's ratioof equity to total assets. Well capitalized banks have higher exposure to NIM and with that benefi t makes more profi t.
(2008). Prudential Norms for Off-balance Sheet Exposures of Banks the bank's profi tability.

Sub Hypotheses
H1a: IRR has a signifi cant impact on the bank's profi tability.

Limitation of the Study
• Short span of the study period (2008 -2018).
• Foreign banks operating in India are not considered for the study.   (Gujarati 2004), "a normally distributed data is an unbiased, effi cient, and consistent estimator and a normally distributed data are refl ected in its descriptive statistics". Above table 1 summarizes the Mean & SD of the selected dependent and independent variables of the study. The above analysis shows the value of Jarque -Bera test is signifi cant. Hence, we can conclude that the selected data is not normally distributed. Therefore, OLS estimation is not suggested to be used compared to Panel Data Regression method.   Table 2 shows, the correlation matrix of all the selected variables in this study. Between bank size and both the dependent variables there is a negative correlation. Furthermore, between IRR & dependent variables, and also between FER and both the dependent variables there is a positive correlation. Therefore, we can conclude that the above-selected variables for the study is not highly correlated with each other.

Multivariate Result
The Looking at these models, we can say that the model seems satisfactory for judging the relationship between banks profi tability and fi nancial risks. Furthermore, the F test results generated show the signifi cance of the models.

Speci cation Test
There dundant fi xed effect test has been used to1select the model between Pooled regression and FE model. The RE estimator1is the asymptotically effi cient estimator1while the FE is unbiased*and consistent estimator but not*effi cient. In order to*specify the model, in the static1panel data analysis,a model specifi cation test was1performed. In choosing the model between the2FE model and the RE model, this study employs the*specifi cation test developed by (Levin, Lin, and Chu 2002). The Hausman specifi cation (HS)1test compares the FE and RE under the nullhypothesis that the individual effects are uncorrelated1with other regressors in the model. The test statistics has a symptotic χ2 distribution. If the null hypothesis1is rejected,1it means that the effects are correlated, thus an RE model produces biased results, violating one of the theLGauss-Markov assumptions; the conclusion is that RE model is not appropriate and it is suggested to use FE modell (Levin, Lin, and Chu 2002  Source: Secondary data It can be observed that the chi-squared statistic of redundant effects test has high statistical signifi cance (p-value zero till the fourth decimal). Thus, it can be concluded that pooled sample regression is not suitable for this data. Hence, the FE model and the RE model are fi tted to the data and the outcomes are shown.

Correlated Random Effects -Hausman Test
Test cross -section random effects Table 7 Source: Secondary data The output of the Hausman test shows that the p-value of 0.000 and this is less than 0.05. Hence the null hypothesis is rejected and the fi xed effects model is considered appropriate.

Multivariate Result with ROA as the Dependent Variable
The effect of IRR on ROA is positive & signifi cant. This indicates that a signifi cant impact of IRR on profi tability measures. (Angbazo 1997) S International Journal of Management examined the relationship between IRR and NIM and found that they have an inverse relationship. The impact of FER on ROA is insignifi cant. Looking at the effect of the OBS1 credit-related activities, the calculated coeffi cients are positively related to ROA for all the selected banks. As for OBS2, the relationship is found to be signifi cantly negative for all the banks, the impact of bank size variable on ROA is signifi cant and negative for all the banks. This fi nding is in line with the fi ndings of (Azam and Siddiqui 2012) . Bank size is usually used to take the potential advantage of economies of scale in the banking sector. The positive relationship between profi tability and size means that the banks benefi t from the scale and there is risk divergence according to the size of the bank (Goddard, Molyneux, and Wilson 2004 Table 8 Source: Secondary data

Redundant Fixed Effects Tests
Test cross-section fi xed effects % Table 9 Source: Secondary data It could be observed that the chi-square statistic of redundant effects test has high statistical signifi cance (p-value zero till the fourth decimal). Thus, it can be concluded that pooled sample regression is not appropriate for the above data. Hence, the FE model and the RE model are fi tted to the above data and the results are shown  Table 11 Source: Secondary data In order to discover the most appropriate model between RE and FE, the HS test of Correlated Random Effects is implemented and the result is proven.

Hausman Test
Test cross-section random effects Table 12 Source: Secondary data The outputof the Hausman test shows that the p-value of 0.000 and this is less than 0.05. Hence the null hypothesis is rejected and the fi xed effects model is considered suitable.
As per the above analysis, Table 12 describes that, IRR is signifi cant for all the banks, this indicates a signifi cantimpact of IRR on profi tability measures and FER is insignifi cant for all banks. In the case of controlled variables, OBS1 which is related to credit activities is negatively signifi cant with ROE, but Derivative related activity i.e., OBS2 is insignifi cant with ROE for all the banks. Furthermore, Bank capital & Bank size are signifi cant with ROE for all the banks. Bank Size is commonly used to get the advantage of the potential economies of scale in the banking sector. Excessive profi tability tends to be associated with banks that keep a notably high amount of capital. consequently, a positive relationship between profi tability and size means that the banks benefi t from the scope of economics and there is risk divergence according to the size of the bank (Goddard, Molyneux, and Wilson 2004). The result exhibits that the size of the bank impact is insignifi cant. The viable motive will be that the size isn't the fi nest one that might contribute to higher profi tability. A positive relationship indicates that higherowner's capital offers the banks the opportunity to maximize their ROE and hence their profi tability.

Conclusion
The previous literature shed some light on the relationship between profi tability measures and various fi nancial risks of the commercial banks. Based on the empirical analysis, it cannot be concluded that fi nancial risks have an impact on the selected profi tability measures of the banks ( Table  5 & 10). Based on the above empirical evidence it is clear that FER does not have any major impact on the profi tability of the banks in India, however, another independent variable IRR has a major impact on banks profi tability measures, and it is statistically signifi cant indicating the fact that higher risk results in lower return.
As per the above result, it can be concluded that, Interest Rate Risk has a positive signifi cant impact on ROA & ROE of the banks. It means if IRR increase by 1%, then ROE & ROA also increase to that extant and viz versa. Similarly, FER is insignifi cant with ROA & ROE, i.e., If any percentage increase in FER will not make any impact on ROE & ROA.
As for the measures of profi tability study, several extensions would be very useful. In the current S International Journal of Management scenario, the Indian new generation private sector banks are still in their infancy stage, hence in this study dynamic models could not be employed effectively, however, it is possible to extend the study period in the future. It would also be suggested to use quarterly data so that a clearer understanding of the dynamic responses of bank profi tability movements can be obtained. It is therefore suggested that future research may consider a wider cross-section data, a different and longer time period and can also add diverse and a wider range of variables. In this study, the main limitation is the short span of the study period, as most of the new generation private banks in India are still new. Therefore, it is not suggested to apply the dynamic PDR analysis in the present study. There are still a lot of avenues and opportunities to explore further in this area. As a matter of fact, further studies should not be limited to the banking industry only but should also be extended toother industries as well.