Private Sector Bank Modeling- A Comparative Modeling Approach Involving Panel Data Regression

 

Ms. Nisha Kumari

Christ University, Bengaluru-560029, India.

 

ABSTRACT:

The banks are the important component of the economy and play a vital role in activating and sustaining economic growth but there is downturn in the economy which led to a rise in bad loans or non-performing assets (NPAs), poor capital adequacy ratio, and unhedged forex exposure. ‘Earnings per Share’ (EPS) is a pivotal parameter which will impact the performance of the bank’s share in the stock exchange.so in this study out of 11 banks of SandP BSE Sensex, the private banks have taken into consideration that is HDFC,ICICI and AXIS bank. EPS is accurate measureof banks profitability and by determining it will help the bank to attract FDI and FPI which is two important paths for investors that will increase the fund, capital inflow in the economy to maintain liquidity which will enhance the growth of economy. To determine the output, PDR (panel data regression) method which is an econometric method (Fixed effect) has been used to prove that EPS is corrective and predictive measure to construct a robust and accurate model. Apart from that 11 other significant variables have been taken to compose the predictive model.

 

 

 

INTRODUCTION:

The Indian banking industry is enormously affected by external factors like ineffective recovery tribunal, willful defaults, capital policy and regulations, digitization and industrial sickness. The internal effects like poor credit appraisal system, management inefficiency and imperfect technology have impacted the recovery process of loan and eventually increase NPA, which reduces the capital inflow and results declining of stockvalue. Basically NPA is a default asset which is not repaid by the borrower and thus impacts the earnings of the bank. To protect the earning and growth of bank BASEL III norms have been implemented.

Several studies have been done in past that is Bansal and mohanty (2013) in his study he elaborated the CAMEL model which is for capital adequacy ,asset quality, management earning, and liquidity to measure the performance  of the bank. The financial health of the bank is measured by its net income, Earning per share is nothing but net income divided by no of shares.EPS is very important parameter which directly impacts the performance of bank stocks in the market.it also showcase the position and value of the bank in the market, its performance not only impact the market share but also the PE ratio of a bank. EPS is important investment tools to assess the performance of the company either in the short or long term. Thus it measure the financial status and prospect of a company.it is important tool for investor to estimate the position of bank for their investments. By various study it has been found that private bank has less EPS as compared to public banks so private bank should increase their PAT to increase the shareholder value, they are also required to utilize their long term fund efficiency to maximize the return.This depicts in this table given below

 

Table 1

 

Demonstrating the earning per share of private and public bank of SandP BSE SENSEX for the period 2013 to 2017, public bank has high EPS which showcase that it is giving high return on investment to its shareholder.

 

 

Figure 1 EPS OF PRIVATE AND PUBLIC BANK

 

Determining the EPS is important work for bank. The major task of bank is to cater the needs of the customer, to understand their needs. Various technology have been introduced which will provide clear picture of behavior of customer about their loans and savings. So that bank will bring out the product which will satisfy the customer need, because retaining the customer is very important task of bank. Data analytics helps the bank to achieve this objective. Hence providing good customer service according to their needs with technology advancement, retaining them will increase theprofitability and earnings which is an essential aspect of banking industry for their business. Without this, the earnings of the banks will suffer losses.

 

To estimate the EPS and impact of other variables on EPS we have used the panel data regression which is econometric tool which help to analyses the data to come up with the proper result and understanding.

 

Panel Data is a mixture of time series and cross section. It is also known as cross sectional time series data. It is the data that is derived from a minimum number of observations with time on a large and maximum number of cross-sectional units like firms, households, individuals, or governments.

 

Panel data consists of observations made byresearchers shown through panel data with numerous eventsand terms that were collected by them over several time periods for the same set of group of entities or data.

The three types of Panel data are:

1. Fixed panel data:

A fixed effects model is a statistical model in which the model parameters are fixed or non-random quantities.

 

2. Dynamic panel data:

In a dynamic effects model, some parameters are not random and some are fixed.

 

3. Random panel data:

A random effects model is a statistical model in which the model parameters are not fixed or random quantities.

 

In this study FIXED EFFECT have used for data analysis. The equation for the fixed effects model becomes:

 

                 [Equation 1]

 

Where –is the dependent variable (DV)

Where i = entity and t = time.

 

LITERATURE REVIEW:

Prediction of the market price is very much important for investors for their investment and trading purpose. This study is basically to forecast the market price of the stock of private sector bank. The main objective of this study is –

·        Which all is the best statistical or econometrical tool to forecast the results accurately?

·        What all are the variables that impact the market price of the stock of especially private sector bank.

 

EPS is cardinal parameter which showcase the financial health and performance of the company, EPS gets impacted by many factors.To find out the explanation of above statements, Will begin with some of the literature review the current researcher are predicting and forecasting the various future outcomes such as Debasish, S. S., and Das, B. (2009) in his study he explained the importance of spread between the E/P (earning to price) ratio and interest rate for determining the stock return and various statistical and econometrics tools has been used likeRegression Analysis ,Granger’s Causality Test,Correlation Analysis, to Forecast the  Performance  of Nifty - NSE Index(India), Abdullah, L. (2013) in his test he used Fuzzy time series, Fuzzy rules which is known as Distance based Fuzzy Series Model to estimate the exchange rates between different currencies. A sample had taken into consideration in which comparison was done between MYR against USD and NTD against USD .it got successfully to predict the exchange rates between Malaysian Ringgit and US Dollars more accurate than the exchange rates between New Taiwan Dollar and US Dollar. other researcher like Saramat, O. et al (2013) he uses  European and united states geographical data with the help of GMM methodological framework to determine the financial health of company by evaluating stock price with some financial ratios. By using the multiple regression analysis study has been done by Mohamed, P. (2013) in their study they elaborated the correlation between the financial performance Iranian private bank and the returns that have given by bank. A study by Vardar, G. (2013) in which he demonstrated the importance of cost and revenue management for business productivity which in turn increase the profit and maximize the return of stock, this is done with the help of relevant data of Eastern European bank and central bank by using regression analysis model and Stochastic Frontier Analysis Model.Interbank transaction is nothing but dealing of one bank with another bank for transaction for importers and exporters. It involves lots of risk, Risk contagion, Li, J. et al (2013) in his work, he developed a model which emphasis the interbank exposure and its impact on whole banking system. Data of 16 banks has taken into consideration for this study. Vaisla, K. S., and Bhatt, A. K. (2010) in their study they have done the comparative study between various statistical tool like Multiple Regression, Time Series analysis, and Neural Network  with Artificial neural network  to determine the accuracy levels of various tools for determining the impact of EPS in stock return.

 

The reviews of the above researchwork contribute logical affirmation that panel data and multiple regression models would be able help us in forecasting with a greater degree of accuracy, the EPS of private Sector Banks in India.

 

OBJECTIVE OF THE STUDY:

To estimate the EPS and variables that has been taken into consideration.

To deduce the cardinal variable that very much impacts the performance of EPS.

To maximize the predictability and sustainability chances of variable which have taken into account?

 

RESEARCH METHODOLOGY:

This study has been done with panel data regression model (Fixed Effect) to generate accurate predictive model for listed private banks of SandP BSE Sensex. EPS is dependent variable in this study. However there are eleven independent variablesalso present to impact the performance of EPS. Time period is chosen carefully from 2008 to 2017. Other important variables were Total Assets (TA), Average Assets (AA), Return on Assets (ROA), Return on Investments (ROI), Non-performing assets (GNPA), Price to Book ratio (PB), Book Value (BV), Graham’s Number for the measurement of bankruptcy (GRHM), Profit after tax (PAT), Price earnings ratio (PE), Interest Income (II), Interest Expense (IE), and Net Interest Margin (NIM).The data set has both cross sectional and time series; it is balanced panel data set. Here instead of random effect the fixed effect has taken into consideration because NPA are going down for all these three banks, since rate cut or hike are time dependent effects.

 

VARIABLE SUMMARY:

Total Assets (TA):

Cash, interest-earning loans, government securities and bond, collateral such as mortgages, interbank loans are total asset of bank which increase the capital inflow.

 

Average Assets (AA):

The sum of total assets for the present year to the total assets for the relevant previous year, and divide by two which gives average asset of bank.

 

Return on Assets (ROA):

It is a pivotal parameter gives an indication that how efficiently company is using its assets to generate revenue. It indicates the profitability of the company through its assets.

 

Return on Investment (ROI):

It measures the amount of return on investment relativeto Investment costs.

 

Non-Performing Asset (GNPA):

It is considered as default asset because when repayment of loan has not been made within or after 90 days of period.  

 

Price to Book ratio (PB):

This ratio showcase market value of stock price to its book value.

 

Book Value (BV):

It is an actual value of asset when it is purchased. Actual value of an asset which is carried in balance sheet is called as Book value.

 

Graham’s Number for the measurement of bankruptcy (GRHM):

This is cardinal parameter to measure the stock value by considering earning per share and book value per share.

 

Profit after tax (PAT):

Ratio tells that how much income is kept in the company as compared to the total revenue.it is measure of profitability of the company.

 

Price earnings ratio (PE):

It indicates the company current price to its earning per share which is an indicator for the investor to buy the shares and market value of the shares is increasing.

 

Interest Income (II):

It is the difference between the revenue that is generated from a bank's assets and the expenses associated with paying out its liabilities.

 

Interest Expense (IE):

This expense is related to the repaymentof interest on any borrowings such as loans, convertible debt, bonds or lines of credit etc.

 

Net Interest Margin (NIM):

It shows that more reserves for the investors and the company can exp and further or can reinvest also.

 

STUDY OUTPUT:

TABLE 2

Dependent variable	EPS
Method	Panel data regressions fixed effect
Sample	2008 TO 2017
Pereiod included	10
Cross sectional included	3

 

TABLE 3

 

TABLE 4

 

MODEL BASED ON PDREQUATION:

EPS--3.70197395752-1.34086229118e-0.7*AA-0.172937034472*BV-1.01168337923e-0.6*GNPA+0.17449475636*GHRM+4.14614588212e-0.5*IE-2.36515238448e-0.6*II-14.166599006*NIM-8.2845985-0.5*PAT-0.406435290961*PB+0.175462773935*PE+258.284206412*ROA+0.190368813168*ROI-4.94500973417e-0.7*TA+[CX=F,PER=F]

 

 

Figure 2 ANALYSIS RESULT

 

Figure 3ANAYSIS RESULT

 

CONCLUSION:

·        As per the fixed panel test done above, the following independent variables have high level of

·        occurrence

·        Book Value has a very high occurrence level of 100% with a p-value of 0.0002

·        Graham’s number has a high occurrence level of 100% with a p-value of 0.0000

·        The above independent variables have high occurrence level indicating that they have asignificant impact on the dependent variable, which is ‘Earning per share’.

·        Book Value and Graham’s numberis two major indicators which determine the probability of bankruptcy with EPS as 100% occurrence. BV is inversely proportion to EPS but graham number is directly related to EPS they have strong correlation.

·        PAT is 93% and PE is 86% occurrence which will not consider as the confidence of 95% in PDR.

·        The other independent variables, which are Total Assets (TA), Average Assets (AA), Return on Investments (ROI), Price to Book ratio (PB), Book Value (BV), Return on asset (ROA), Interest Income (II), Interest Expense (IE), and Net Interest Margin (NIM), do not influence the EPS to a significant extent as per the test resultsabove has fewer occurrences which will not be taken into consideration.

·        The R-squared is at 0.99, which means that the prediction success rate is 99.99%. For instance, if the model is run 100 times, then the probability of getting successful predictions is 99.99%   which  makes it credible that the ‘Earnings per Share’ (EPS) is a pivotal parameter in BSE Sensex which will impact the performance of the bank’s share (HDFC, ICICI and AXIS)

·        F-statistics is very high indicating the average error is very less and which means that the average deviation is high  and  the error tend to be normally distributed, which is very good sign for the model. F-statistic iscalculated as average deviation divided by average error. It shares an inverse relationship with the average error. Hence, higher the F-statistic, lower is the error and vice-versa. This meansthat the model is sustainable in future skewed errors is not present, which evidence about the robustness of the model.

·        AIC, SC and HQ should be within the range of 0-10.coefficients are very small (near to zero) indicating the accuracy as very high, which expressed the model is robust and sustainable for long run.

·        Durbin-Watson is 1.9 which indicates there is feeble negative autocorrelation and in the data greater than R Squared means the PDR (Panel Data Regression) is valid and it is authentic in nature.

·        The value of root mean square error is 0.1478 which is less than its upper limit (which is 1), the value of mean absolute error is 0.1235 which is low than its upper limit (which is 1), the value of mean absolute percentage error is 0.296 which is very lower than its upper limit (which is 5), the value of Theil Inequality Coefficient is 0.001205 which is substantially low than its upper limit (which is 1). Since all the values of errorsare substantially less than their upper limits, it is a good sign for the data which signifies its credibility. So it’s holistic view emphasis that accurate predictive model and it will sustainable for long run.

·        In the second graph, we can see that the actual and fitted lines are strongly bounded   to one another which is the strength of the graph indicating that the model is robust, sustainable and there is no structural breaks no man made data.

·        Hence panel data regression proofs the result.

 

FUTURE SCOPE OF STUDY:

Further scope of study is possible on the foreign banking sectors in India; apart from public and private sector bank inside SandP BSE Sensex, there are foreign banks. And, neural network, Fuzzy logic and ANFIS-ICA methods could make more accurate and robust models in future.

 

REFERENCES:

1.       Abdullah, L. Performance of Exchange Rate Forecast Using Distance-Based Fuzzy Time Series. Al-Shubiri, F. N. (2010). Capital structure and value firm: an empirical analysis of abnormal returns. Economia. Seria Management, 13(2), 240-253.

2.       Arnerić, J., Poklepović, T., and Aljinović, Z. (2014). GARCH based artificial neural networks in forecasting conditional variance of stock returns. Croatian Operational Research Review, 5(2), 329-343. Bansal Mohanty (2013) A Study on Financial Performance of Commercial Banks in India: Application of Camel Model. Al-Barkaat Journal of Finance and Management, Volume 5, Number 2, July 2013, pp. 60-79.

3.       Debasish, S. S., and Das, B. (2009). Forecasting Movement of Stock Index–Use of Spread between E/P Ratio and Interest Rate. Asian Social Science,4(11), 68.

4.       Eichengreen. B, Gupta. P, (2013). The financial crisis and Indian banks: Survival of the fittest?, Journal of International Money and Finance, Volume 39, December 2013, Pages 138-152. Ghodrati, H., and Taghizad, G. (2014). Credit risk assessment: Evidence from banking industry. Management Science Letters, 4(8), 1765-1772.

5.       Ghosh and Mondal (2012) "Intellectual capital and financial performance of Indian banks", Journal of Intellectual Capital, Vol. 13 Issue: 4, pp.515 – 530.

6.       Li, J., Liang, C., Zhu, X., Sun, X., and Wu, D. (2013). Risk contagion in Chinese banking industry: A Transfer Entropy-based analysis. Entropy, 15(12), 5549-5564.

7.       L. Abdullah. (2013). Performance of Exchange Rate Forecast Using Distance-Based Fuzzy Time Series.

8.       Mahapatra and Mahapatra (2015), Service Quality of Indian Banks: A Fuzzy inference system approach. Asian Academy of Management Journal, Vol. 20, No. 2, 59–80.

9.       Motamedi, P. (2013). Investigating different factors influencing on return of private banks.Management Science Letters, 3(9), 2467-2472.

10.     Şărămăt, O., Dima, B., Angyal, C., and Ştefana Maria, D. (2013). Financial ratios and stock prices on developed capital markets. Studia Universitatis Vasile Goldiş, Arad-Seria Ştiinţe Economice, (1), 1-12. Šterba, K. H. M. L. J. The impact of financial crisis on the predictability of the stock markets of Pigs countries–comparative study of prediction accuracy of technical analysis and neural networks.

11.     Taghva, M., Bamakan, S., and Toufani, S. (2011). A data mining method for service marketing: A case study of banking industry. Management Science Letters, 1(3), 253-262.

12.     Vardar, G. (2013). Efficiency and Stock Performance of Banks in Transition Countries: Is There A Relationship?.International Journal of Economics and Financial Issues, 3(2), 355.

13.     Vaisla, K. S., and Bhatt, A. K. (2010). An analysis of the performance of artificial neural network technique for stock market forecasting.International Journal on Computer Science and Engineering, 2(6), 2104-2109.

14.     Wuerges, A. F. E., and Borba, J. A. (2010). Neural networks, fuzzy logic and genetic algorithms: applications and possibilities in finance and accounting. JISTEM-Journal of Information Systems and Technology Management, 7(1), 163-182.

15.     Ghosh, bikramaditya, M.C Krishna, T.S Ramachandran. “PSU Bank Modeling- A comparative modeling approach involving Artificial Neural Network and Panel Data Regression”. Asian Journal of Researchin Business economics and management, 6.6(2016):27-36.

 

 

Received on 20.09.2017                Modified on 01.11.2017

Accepted on 05.12.2017         © A&V Publications All right reserved