A Quantitative Analysis of Indian Banks’ Performance and Efficiency-A Panel Regression Approach using EVIEWS 7.0.

A Quantitative Analysis of Indian Banks’ Performance and Efficiency-A Panel

Regression Approach

Submitted to:

Dr. VIGHNESWARA SWAMY

Page | 1

Abstract The importance of banks which constitutes the main structure of the financial system

increased with globalization on the both sides of country and firm. Factors that affect banks‟

performance can be macroeconomic or internal factors relevant with banks. In this study,

different internal factors that can have an impact on banks‟ profitability are examined with

panel data analysis. The dataset includes 18 public sector banks and 14 private sector banks.

The time frame considered for the analysis is from March‟02 to March‟11. Return on Net

Worth is taken as the dependent variable as a proxy for the profitability of a bank.

Table of Contents Abstract ...................................................................................................................................... 1

Table of Figures ......................................................................................................................... 2

Table of Tables .......................................................................................................................... 3

1. Introduction ........................................................................................................................ 4

2. Literature Review............................................................................................................... 4

3. Indian Banking Sector........................................................................................................ 5

3.1 Performance and Trends of Indian Banks ................................................................... 6

4. Empirical Analysis ............................................................................................................. 9

4.1 Data Analysis .............................................................................................................. 9

4.1.1 Dependent Variables ............................................................................................ 9

4.1.2 Control or Independent Variables ...................................................................... 10

4.1.3 Data Sample ....................................................................................................... 11

4.2 Model Estimation (Methodology) ............................................................................. 12

4.2.1 Panel Unit Root Test .......................................................................................... 13

4.2.2 Results for the whole sample (Public and Private Banks) ................................. 14

4.2.3 Results for only Public Sector Bank Sample ..................................................... 18

4.2.4 Results for only Private Sector Banks Sample .................................................. 19

4.3 Summary of the Results and Conclusion .................................................................. 21

5. Limitations ....................................................................................................................... 22

6. Annexures ........................................................................................................................ 23

7. References ........................................................................................................................ 25

Page | 2

Table of Figures Figure 3-1 Growth in Aggregate Deposits of SCBs .................................................................. 6

Figure 3-2 Demand Deposits Figure 3-3 Term Deposits ................................................. 7

Figure 3-4 Share in Deposits-Ownership Wise ......................................................................... 7

Figure 3-5 Growth in Gross Bank Credit ................................................................................... 8

Figure 3-6 Share in Bank Credit-Group Wise ........................................................................... 8

Figure 3-7 NIM Trend-Group Wise ........................................................................................... 9

Page | 3

Table of Tables Table 3-1 Major Achievements since Nationalisation ............................................................... 6

Table 3-2 Movement in Bank BPLR-Group Wise .................................................................... 8

Table 4-1: List of Public Sector Banks (PSB) ......................................................................... 11

Table 4-2: List of Private Sector Banks ................................................................................... 11

Table 4-3: Model 1-Fixed One Way (Banks) Effect ............................................................... 14

Table 4-4: Model 2-Fixed 2-Way Effect ................................................................................. 15

Table 4-5: Model 3-Random 2 Way Effect Model .................................................................. 15

Table 4-6: Model 4-Fixed 2-Way Effect for Efficiency .......................................................... 16

Table 4-7: Model 5-Random 2 Way Effect for Efficiency ...................................................... 17

Table 4-8: Model 6- Random 2-Way Effect Regression on RONW of PSB ........................... 18

Table 4-9: Model 7-Random 2-Way Effect Regression on NIM of PSB ................................ 18

Table 4-10: Model 8- Random 2-Way Effect Regression on RONW of PrSB ....................... 19

Table 4-11: Model 9-Random 2-Way Effect Regression on NIM of PrSB ............................. 20

Table 4-12: Summary of Results for Regression on RONW ................................................... 21

Table 4-13: Summary of Results for Regression on NIM ....................................................... 22

Table 6-1: LLC Panel Unit Root Test ...................................................................................... 24

Page | 4

1. Introduction The Indian financial sector underwent a radical change during the nineties. From the

relatively closed and regulated environment in which agents had to operate earlier, the sector

was opened up as part of the efficiency enhancing structural policies to bring about high

sustainable long-term growth of the economy. The banking sector was also not an exception

to this rule. New measures were undertaken to induce efficiency and competition into the

system. Accounting and provisioning norms, capital adequacy rules, proper risk management

measures, etc. were brought in place and entry regulations were also relaxed. The

environment was made friendlier for domestic private sector and foreign banks. So, as a

result, many new private players entered the banking sector giving rise to the heightened

competitive pressure.

In this report, an attempt has been made to analyze the effects of various internal factors and

the effect of ownership structure on the profitability of a bank. The methodology used for the

analysis is that of Panel Regression which becomes relevant when there are data for a period

of time for each of the units being considered and thus, becomes readily applicable to the

present case because for the banks that have been considered in this paper, the data on the

relevant variables are available for several years. In order to make full use of the available

data, this technique assumes relevance. Besides making full use of the data, this technique

also has some very important specific advantages. In an analysis of this kind, there might be

several bank-specific and time specific influences that are unobservable and hence not

captured by the variables used in the regression. For checking the effect of ownership, pooled

estimation technique has been used in place of Fixed and Random effect estimation

techniques.

2. Literature Review Berger (1995) found a strong positive relationship between return on equity (ROE) and

Equity/Total Assets, using banks‟ performance variables as ROE, return on assets (ROA) and

net interest margin. Naceur and Goaied (2001) determined that good performer banks‟ labor

and capital efficiencies are higher, deposit accounts‟ volume is higher relative to income

earning assets and they increase their equity or enhanced equity with undistributed profit.

Türker (2002) researched Turkish banking sector profitability determinants using panel data

including the periods 1997-2000. Two-step approaches are applied to measure the relative

importance of the micro and the macro elements to determine the profitability. Within the

micro determinants: capital, liquidity, personnel expenditures, deposits and market share are

found to have significant influences on net interest margins. Among the macro variables,

inflation and budget deficits have significant effect on net interest margins. At the end of the

analysis results concluded that capital, liquidity, personnel expenditures, loans, non-

performing loans and deposits are the micro determinants of return on assets (ROA). The

findings of her study also revealed that, the most important contributors of return on equity

(ROE) are capital, securities portfolio, liquidity, personnel expenditures, loans, deposits and

market share.

Page | 5

The most extensive research about this subject belongs to Demirgüç and Huizingha (1998).

These two researchers tried to explain the interest spread and profitability components using

80 countries data during the period of 1988-1995. In their study, macro-economic variables,

taxing, banking regulations, financial structure and legal criterion are used as variables. They

found that when bank assets/GDP ratio increases and bank concentration rate decreases,

banks interest spread and profitability declined. In developing countries foreign banks are

more profitable than national banks whereas in developed countries that relationship reverses.

The relationship between bank ownership and performance has not been analyzed extensively

and only a few references in this regard include the papers by Sabi (1991), Davies and

Brucato (1987) and Sarkar, et. al. (1998). La Porta, et. al. (2000) identifies the

“development” view and “political” view of the government ownership of banks.

In the literature, there has been extensive analysis on the issue of ownership and performance

of firms. The broad lines of thought in this regard are the property rights approach (as

exemplified by the writings of Alchian, 1965 and de Alessi, 1980) and the public choice

approach (as represented by the writings of Nickskamen, 1971 and Levy, 1987).

It must be remembered that most of the evidence on the ownership-performance relationship

is centered on developed countries and a similar line of reasoning might not work for

developing countries because of the absence of a well-defined market for corporate control.

This is so because in many developing countries, there is a lack of free flow of information,

lack of transparency and the presence of incomplete markets, which are prerequisites for

defining property rights. The ownership-performance effects noticed in the context of

developed countries might not be working in these cases. India, thus, provides an interesting

example in this regard because the country has come out from the regulated environment and

is moving to a more market-oriented scenario. In this sense of the term, India is an emerging

economy and the period chosen for the analysis is the one when one can comfortably say that

both-public and the private sector banks have equal opportunities in terms of growth and

competition.

3. Indian Banking Sector The Indian banking sector consists of the Reserve Bank of India (RBI), which is the central

bank, commercial banks and co-operative banks. Commercial banks are of two types-

scheduled, which are subject to statutory requirements and non-scheduled, which are not.

Scheduled banks can be further classified into public sector banks [comprising of the State

bank of India, its seven associates, other Nationalized banks and the Regional Rural Banks

(RRBs)] and private sector banks, which can be either domestic or foreign.

The primary objective of bank nationalization in 1969 was to provide assistance at

concessional rates of interest to relatively backward areas. Pursuant to the nationalisation, the

banking sector became dominated by a plethora of rules and regulations. Nationalisation

increased the scale of banking operations substantially (as depicted in Table 3-1, which

illustrates the major achievements since nationalisation) but, at the cost of profitability and

efficiency of the banking system; in many instances, this led to a piling of Non-Performing

Assets (NPAs) with the banks, causing major concern. As part of the reform process initiated

after the balance of payments crisis in 1991, large- scale reforms were brought about in the

Page | 6

financial sector in general and the banking sector in particular. As the architect of these

reforms, M. Narasimham (1998) had pointed out, the reforms in the banking sector can be

classified into two phases: The first phase consisted of the curative measures, which were

brought about for making the banking sector more oriented to the market and impart

competition to the environment. The second phase consisted of the preventive measures,

which were brought about to ensure smooth functioning of the banking sector in the long run.

Business Indicators June 1969 March 1991 March 2000

Total Number Of

Offices

8,262 60,220 67,339

Population Per Office

(000’s)

65 14 15

Total Deposits (Rs.

billion)

137.8 1101.2 8452

Deposits Per Office

(Rs. lakhs)

56 334 1255

Total Credit (Rs.

Billion)

106.8 667 4822

Credit Per Office (Rs.

lakhs)

44 202 716

Table 3-1 Major Achievements since Nationalisation1

3.1 Performance and Trends of Indian Banks2

a) Deposits:

Figure 3-1 Growth in Aggregate Deposits of SCBs3

1 Source: Sen and Vaidya (1997) and Statistical Tables Relating to Banks In India: 1999-2000

2 Source: RBI and Dun & Bradstreet Reports on Indian BFSI Sector

Page | 7

Figure 3-2 Demand Deposits Figure 3-3 Term Deposits

Figure 3-4 Share in Deposits-Ownership Wise

3 Scheduled Commercial Banks

Page | 8

b) Credit4:

Figure 3-5 Growth in Gross Bank Credit

Figure 3-6 Share in Bank Credit-Group Wise

Table 3-2 Movement in Bank BPLR-Group Wise

4 Source: Annual Report, RBI, Various Issues

Page | 9

c) Net Interest Margin:

Figure 3-7 NIM Trend-Group Wise

4. Empirical Analysis The effect of various control variables on banking performance can be analysed by estimating

an empirical model that would test the hypothesis of any significant effect of these variables

on performance. In this section, a model to test this hypothesis is estimated.

4.1 Data Analysis Performance of a bank can be judged through various angles but in this report, the

profitability and the efficiency are being considered as the proxy for the performance

measurement. The proxy for the profitability that has been used in this report is Return on

Net worth (RONW) and that for the efficiency, Net Interest Margin (NIM) has been

considered. All the data that has been considered in this project is Secondary in nature.

4.1.1 Dependent Variables

Return on Net Worth or Return on Equity is the amount of income as a percentage of

shareholder‟s equity. Return on equity measures a bank's profitability by revealing how

much profit a bank generates with the money shareholders have invested.

Return on Equity = (Net Income/Stockholder Equity)

The measure of efficiency used here, the Net Interest Margin (NIM) is defined as the

difference between interest earned and interest expended as a proportion of average total

assets. Here, a bit modified version of this ratio has been taken into consideration. It is a ratio

between Net Interest Ratio and Total Fund of a bank instead of average total assets.

Net Interest Margin = Net Interest income/ Average Earning Assets

Page | 10

4.1.2 Control or Independent Variables

a) Non-Interest Income Ratio (NIIR): This factor has been brought into the picture to

capture the effect of diversification of a bank. Non-interest income tells about the income

earned by the bank through commission, brokerage, service charge, exchange and other

fee-based services which can play a pivotal role in the profits of any bank. The variable

taken into the consideration is the ratio between the Non-interest income and total fund of

a bank.

b) Interest Expended/Interest Earned (IntExpIntEar): This ratio represents the expenses

over the earnings. The interest expenditure represents the cost of fund of a bank. It

consists of interest paid on deposits and borrowings.

c) Investment/Deposit Ratio (InvesR): Investments include total investment including

investment in non-approved government securities. A major percentage of a bank‟s

investment is on Government approved securities. But now the banks are also shifting

their investments to other fields because of limited returns on Government securities (but

these are safer).

d) Total Loans/Total Assets (LINTNSTY): This ratio is also known as the Loan Intensity

of a bank. The loan to assets ratio measures the total loans outstanding as a percentage of

total assets. The higher this ratio indicates a bank is loaned up and its liquidity is low. The

higher the ratio, the more risky a bank may be to higher defaults.

e) C-D ratio (CDRATIO): Credit-Deposit ratio is one of the main indicators of banks

investment activities and also states the credit deployment for the resources raised in the

form of the deposits. It also a measure of liquidity of any bank. CD ratio is an index of the

health of banking system in terms of demand for credit in proportion to total deposit

growth in the banking sector. A declining CD ratio implies that banking sector was flush

with funds without any corresponding demand for credit affecting the bank's profitability

in the long run as they have to pay interest to depositors without corresponding income

from the credit outflow. RBI has indicated the banks that their average C-D ratio should

not exceed 70% mark.

f) Operating Expenses/Total Income (OperExpIncomeR): The operating expense ratio

shows the percentage of a bank's income that is being used to pay maintenance and

operational expenses. Operating Expenses equals the non-interest expenses. It is useful to

measure how costs are changing compared to income - for example, if a bank's interest

income is rising but costs are rising at a higher rate looking at changes in this ratio will

highlight the fact. It is also an efficiency measure.

g) Relative Deposit Market Share (RDS): It is defined as the amount of deposit at a

particular bank divided by the total amount on deposit at all banks. The deposit market

share is a way of measuring the size and performance of a bank.

h) Ownership Structure (Ownership): Here, a dummy variable OwS has been taken to

analyse the effect of type of ownership on the bank‟s performance. Only the Public (PSB)

and the Private (PrSB) category of ownership have been considered in this paper. Foreign

banks have been excluded from the analysis. The Ownership dummy will take the value

„1‟ if it is a public sector bank and „0‟ if it is a private sector bank.

Page | 11

Note: For the estimation purpose, all the variables have been transformed into their

natural logarithm.

All the measures of profitability and efficiency used in this paper are based on accounting

information and as such, are accounting measures. As such, they do not capture the

underlying determinants of shareholder value [See for example, Padhye and Sharma (2002)].

As has been pointed out by Padhye and Sharma (2002) and Mor and Sharma (2002) in this

context, Economic Value Added (EVA) or Shareholder Value Added (SVA) might be better

measures of performance of banks. The main reason for using the accounting measures in

spite of their inherent imperfections is that the balance sheets of banks are highly opaque and

the cash flow statements required for the calculation of these measures are extremely difficult

to obtain. For the calculation of SVA for example, one needs to have a forecast of future cash

flows for the bank in question. This job would have been next to impossible for our sample of

banks belonging to different categories and concentrated in different regions. The need was

thus felt to continue using the accounting measures and as has been pointed out earlier, these

measures had also been used in earlier studies.

4.1.3 Data Sample

The data sample consists of 18 public sector banks (PSB) and 14 private sector banks (PrSB).

The time frame considered for the panel regression is of 10 years from March, 2002 to

March, 2011. In total the sample size is of 32 banks and time frame of 10 years. Thus, the

panel dimension is 32*10=320 (excluding the time-series and cross-sectional ids). The list of

the banks taken into the sample is shown as below:

i. Allahabad Bank

ii. Andhra BANK

iii. Bank of Baroda

iv. Bank of India

v. Bank of Maharashtra

vi. Canara Bank

vii. Central Bank of India

viii. Corporation Bank

ix. Dena Bank

x. IDBI Bank

xi. Indian Bank

xii. Indian Overseas Bank

xiii. Oriental Bank of Commerce

xiv. Punjab National Bank

xv. State Bank of Bikaner & Jaipur

xvi. State Bank of India

xvii. Syndicate Bank

xviii. Union Bank of India

Table 4-1: List of Public Sector Banks (PSB)

i. Axis Bank

ii. City Union Bank

iii. Dhanlakshmi Bank

iv. Federal Bank

v. HDFC Bank

vi. ICICI Bank

vii. IndusInd Bank

viii. ING Vysya Bank Ltd

ix. Jammu and Kashmir Bank

x. Karnataka Bank

xi. Kotak Mahindra Bank

xii. Lakshmi Vilas Bank

xiii. South Indian Bank

xiv. Karur Vysya Bank

Table 4-2: List of Private Sector Banks

The data for all the variables have been collected and calculated from the Capitaline Plus

Database, RBI and Indian Bank‟s Association.

Page | 12

The panel data that will be used for the modeling is a balanced panel data. A balanced

panel is one which has same number of time-series observations for each cross-sectional

unit.

For some of the variables, natural log transformation has been done just to normalize the

data for all the variables. It helps in increasing the normality of data if there is lot of non-

normality amongst the variables.

All the data used in this project is of secondary in nature and taken from public domain

sources mentioned above.

4.2 Model Estimation (Methodology) The technique that will be used for modeling is a bit complicated regression technique known as

panel regression technique. A panel of data will embody information across both time and space.

For generating and the testing of the model, advanced econometric software EVIEWS v7.0 has

been used extensively.

The model that is being used to check the performance of the bank is shown as below:

Here, (Performance)it is the performance measure for the ith

bank during the tth

period, D is a

vector of dummy variables that characterize ownership, Xit is a vector of other control

variables that might affect performance and vit is a random error term. δ and β are the column

vectors of the coefficients to be estimated. The elements of β characterize the effect of

various control or independent variables on the performance of a bank.

In the analysis, both the estimation techniques of Panel Regression have been considered. But

the main focus will be on the Random effect model5. The reason for choosing a random

effects model over a fixed effect one is primarily driven by data. In a fixed effects model, in

this case, the presence of the ownership dummy which takes the same value for the same

bank across all time-periods gives rise to a matrix of explanatory variables which is singular,

that is, the value of the determinant of that matrix becomes zero and as such, it cannot be

inverted. This happens because a linear combination of the vectors of ownership dummies

gives rise to the intercept vector. As the explanatory variable matrix cannot be inverted

because of the collinearity of the regressors, the coefficients cannot be estimated. It is also

termed as the dummy variable trap.

In random effects models, the estimation framework considers that the constant term or the

intercepts for each cross-sectional unit (i.e., individual stocks) are assumed to occur from

common intercept term i.e., α plus a random variable εi that varies cross-sectionally but is

constant over time in case of one-way random effect estimation. εi measures the random

5 For Details please see Baltagi (2005), Chris Brook (2008, pp.498) and Kennedy (2003, pp.315)

Page | 13

fluctuations of each entity„s intercept term from the global intercept term α. random effects

panel model can be written as:

Yit = α + βXit + ωit , ωit = εi + vit

The main required assumption of this framework is that the error term (here only cross-

sectional error term) εi has zero mean and is independent of the cross-sectional error term

(vit). Also, the variance σ2

ε is constant and error-term is independent of the explanatory

variables (Xit).

In random-effect models, a generalized least squares (GLS) method is usually used for the

estimation. The transformation used in this GLS estimation procedure is to subtract a

weighted mean of the Yit over time (i.e. part of the mean rather than the whole mean). Then,

define the „quasi-demeaned‟ data as Yit∗ = Yit – θYi

’ and Xit

∗ = Xit – θXi’, where Yi

’ and Xi

’ are

the means over time of the observations on Yit and Xit respectively. θ will be the function of

the variance of the entity-specific error term, σ2v , and of the variance of the entity-specific

error term, σ2

ε

This transformation ensures that there are no cross-correlations in the error terms, but

fortunately it will be automatically implemented by the Eviews.

4.2.1 Panel Unit Root Test

It is very important for any time-series data to pass this test. A unit root test tests whether a

time-series variable is non-stationary. Recent literature suggests that the panel-based unit root

tests have higher power than those based on individual time series. For the testing of the unit

root in a panel data, Levin, Lin and Chu (LLC6) Test will be used using the EVIEWS v7.0.

The null hypothesis is that each individual time series contains a unit root against the

alternative that each time series is stationary.

LLC consider the following basic Augmented Dickey-Fuller (ADF) specification:

Δyit = αyit-1+ΣPj=1 βijΔyit-j+Xit’δ+εit

Where the assumption is that α=ρ-1 (ρ are the autoregressive coefficients). So, the null hypothesis

of the test is written as:

H0: α = 0

H1: α < 0

6 For details, please see Baltagi, Econometric Analysis of Panel Data 3e,Willey & Sons, 2005, pp. 240

Page | 14

The LLC test is performed for each of the 9 variables (both dependent and independent). The

results of LLC Panel Unit Root Test are shown in the Annexures. The Table for the unit root

test tells that No Unit Root is present in any of the variable. It means that the data for all the 9

variables are stationary and hence we can proceed for our estimation procedure.

4.2.2 Results for the whole sample (Public and Private Banks)

4.2.2.1 Profitability Measure (RONW) for Whole Sample

4.2.2.1.1 Model 1-Fixed One Way Effect for Profitability (Cross-Sectional/Banks

Effects Only)

Table 4-3: Model 1-Fixed One Way (Banks) Effect

On examining the above model, it is evident that the Non-interest Income Ratio (NIIR),

Interest Expended/Interest Earned Ratio and the Operating Expense to Income Ratio are the

only three variables that are affecting the profitability i.e. Return on Net-Worth (RONW). All

these are showing very high significant effect on the profitability of the banks. NII ratio is

showing a strong positive correlation with RONW while the other 2 are showing the strong

negative correlation with RONW. The direction of these effects is in line with the previous

researches.

Variables Coefficients Standard

Error

t-Statistic Probability

C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OPEREXPINCOMER

-2.355922

0.063802

-0.263787

0.374752

-0.181924

-0.218084

-1.131066

-0.908618

0.435621

0.053667

0.235028

0.048915

0.119903

0.123696

0.176452

0.097522

-5.408191

1.188843

-1.122363

7.661236

-1.517260

-1.763066

-6.410048

-9.317051

0.0000

0.2355

0.2627

0.0000

0.1303

0.0790

0.0000

0.0000

R-squared 0.715247

Adjusted R-square 0.676739

Probability (F-Statistic) {Chow-Test} 0.000000

Durbin-Watson stat 1.834630

Page | 15

4.2.2.1.2 Model 2-Fixed Two Way Effect for Profitability

Table 4-4: Model 2-Fixed 2-Way Effect

The results and the interpretation are pretty much similar to the previous model. But by

introducing the time effect, Investment/Total Deposit ratio is also showing significant effect

on RONW. The correlation between this and RONW is negative.

4.2.2.1.3 Model 3- Random 2-Way Effect Model for Profitability

Table 4-5: Model 3-Random 2 Way Effect Model


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OPEREXPINCOMER

-2.316522

-0.013453

-0.033949

0.373783

-0.056786

-0.494459

-1.151982

-0.646782

0.773557

0.064431

0.351306

0.147598

0.212188

0.243019

0.373691

0.232007

-2.994637

-0.208795

-0.096638

2.532429

-0.267622

-2.034651

-3.082710

-2.787773

0.0030

0.8348

0.9231

0.0119

0.7892

0.0429

0.0023

0.0057

R-squared 0.432848





Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OPEREXPINCOMER

OWNERSHIP

-1.776610

0.047185

-0.473256

0.374793

0.085547

-0.534993

-0.871891

-0.332721

0.160887

0.708863

0.045497

0.532804

0.098045

0.247103

0.259092

0.284856

0.089351

0.111641

-2.506281

1.037098

-0.888237

3.822656

0.346198

-2.064875

-3.060819

-3.723748

1.441108

0.0127

0.3005

0.3751

0.0002

0.7294

0.0398

0.0024

0.0002

0.1506

R-squared 0.153394




Page | 16

The Random 2 way effect model estimation shows that the factors NIIR, INVESR,

INTEXPINTEAR and OPEREXPINCOMER are significantly affecting the RONW of the

banks. Ownership doesn‟t seem to have any significant effect on the banks‟ profitability i.e.

RONW. NIIR is positively correlated to the RONW while the rest are negatively affecting

RONW or the profitability of the banks.

4.2.2.2 Efficiency (NIM) Measure for Whole Sample

For efficiency factor, the variable Operating Cost to Income ratio (OPEREXPINCOMER)

has been excluded because that is itself a direct measure of the efficiency of any bank.

4.2.2.2.1 Model 4: Fixed Two Way Effect for Efficiency

Table 4-6: Model 4-Fixed 2-Way Effect for Efficiency

As shown in the above table, only 2 factors are significantly affecting the efficiency of the

banks. These are loan intensity of the banks and the Interest Expended to Interest Earned

Ratio. Both of these are negatively correlated. It shows that if banks are giving more loans

relative to the size of their assets then it has a negative effect on the NIM or the efficiency of

the banks. Similar is the case with the Interest Expended to Interest Earned ratio. Only

problem in this model is that Durbin-Watson Statistic is more than 2 which shows that there

might be some negative serial correlation amongst the error terms. But it is not that higher so

we can ignore it.


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

-3.943912

0.028500

-0.864065

0.096115

0.337336

0.134659

-1.338791

0.620390

0.088879

0.392746

0.083271

0.186957

0.211506

0.270163

-6.357147

0.320656

-2.200062

1.154236

1.804351

0.636664

-4.955498

0.0000

0.7487

0.0286

0.2494

0.0723

0.5249

0.0000

R-squared 0.504946




Page | 17

4.2.2.2.2 Model 5: Random 2-Way Effect for Efficiency

Table 4-7: Model 5-Random 2 Way Effect for Efficiency

If we compare our results from the previous model (model 4), then we can see that two more

variables are showing significant effect on NIM. These are Relative Deposit Share (RDS) and

Ownership. The result shows that the RDS is having negative correlation with NIM. It means

that the banks which have large deposits (relatively) are less efficient than the banks with

lower deposits. Ownership dummy is positively correlated with the efficiency of Banks. It

means that the Public sector banks (code is 1) are more efficient than the private sector banks

(code is 0). This is in contrast to the view that the privatization or the private banks are more

efficient than the public sector banks.


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OWNERSHIP

-5.218617

-0.067844

-0.853572

0.039816

0.241950

-0.267077

-1.763927

0.111146

0.401450

0.019504

0.324347

0.054359

0.147073

0.150831

0.134503

0.046350

-12.99942

-3.478526

-2.631659

0.732471

1.645098

-1.770700

-13.11443

2.397984

0.0000

0.0006

0.0089

0.4644

0.1010

0.0776

0.0000

0.0171

R-squared 0.404907




Page | 18

4.2.3 Results for only Public Sector Bank Sample

4.2.3.1 Profitability Measure for Public Sector Banks (PSB)

Table 4-8: Model 6- Random 2-Way Effect Regression on RONW of PSB

The results shown in the above table are same as shown in table 4.5. Only problem in this

model is that Durbin-Watson Statistic is less than 1 which means that there is some positive

serial correlation between the error terms.

4.2.3.2 Efficiency Measure for Public Sector Banks

Table 4-9: Model 7-Random 2-Way Effect Regression on NIM of PSB


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OPEREXPINCOMER

-1.566017

0.063465

-1.082809

0.480145

0.570280

-1.007887

-0.981144

-0.213477

1.030672

0.069314

0.738198

0.149099

0.331177

0.351455

0.360794

0.085386

-1.519414

0.915618

-1.466827

3.220316

1.721981

-2.867758

-2.719398

-2.500127

0.1305

0.3611

0.1442

0.0015

0.0869

0.0047

0.0072

0.0134

R-squared 0.199450





Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

-6.241409

-0.069081

-2.102312

0.019099

0.697052

-0.686699

-1.916391

0.828704

0.041548

0.642118

0.118402

0.284719

0.296975

0.287264

-7.531533

-1.662685

-3.274028

0.161308

2.448211

-2.312316

-6.671191

0.0000

0.0982

0.0013

0.8720

0.0154

0.0219

0.0000

R-squared 0.310228




Page | 19

If we compare the results shown by the above model with the model 5 shown in Table 4.7,

then we can say that the results are bit different in this case. Here, the significant variables are

Loan-Intensity, CDRATIO, Investment Ratio and Interest Expended to Interest Earned Ratio.

Except from CDRATIO all the remaining 3 are negatively correlated with the NIM. This

shows that in case of Public Sector banks CDRATIO plays a significant role in their

efficiency. Also, RDS has no significant effect on NIM in the case of Public Sector Banks.

4.2.4 Results for only Private Sector Banks Sample

4.2.4.1 Profitability Measure for Private Sector Banks (PrSB)

Table 4-10: Model 8- Random 2-Way Effect Regression on RONW of PrSB

All the results are consistent with the previous model 3 shown in Table 4.6. Only difference

is that in the case of private sector banks Investment Ratio does not have any significant

effect on RONW.


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

OPEREXPINCOMER

-2.618381

0.022764

-0.296664

0.389714

-0.428801

-0.138149

-1.412150

-0.862477

0.978910

0.047114

0.745894

0.128399

0.385505

0.356173

0.410849

0.201979

-2.674793

0.483163

-0.397729

3.035173

-1.112310

-0.387871

-3.437150

-4.270129

0.0084

0.6298

0.6915

0.0029

0.2680

0.6987

0.0008

0.0000

R-squared 0.239410




Page | 20

4.2.4.2 Efficiency Measure of PrSB

Table 4-11: Model 9-Random 2-Way Effect Regression on NIM of PrSB

It is only for this model that NII ratio is also significantly affecting the NIM. Loan Intensity

does not have any significant effect on the efficiency of Private Sector Banks. Rests of the

results are similar to the Model 5.


Error


C

RDS

LINTNSTY

NIIR

CDRATIO

INVESR

INTEXPINTEAR

-4.281945

-0.046824

0.144608

0.097330

-0.091292

-0.121039

-1.821875

0.224716

0.012434

0.188298

0.032947

0.098050

0.091071

0.092036

-19.05491

-3.765717

0.767974

2.954137

-0.931077

-1.329070

-19.79516

0.0000

0.0002

0.4439

0.0037

0.3535

0.1861

0.0000

R-squared 0.774331




Page | 21

4.3 Summary of the Results and Conclusion

a. Summary for Profitability Measure:

Most of the results shown by all the models are more or less consistent with each other

which is a positive sign. Following table shows the consolidated results of all the

regression models related to the RONW:

Sample Model RDS LINTNSTY NIIR CDRATIO INVESR IntExpIntEar OperExpIncomeR OWNERSHIP

All

Banks

Fixed

1-Way NE

NE

+ve NE NE NE -ve NE

Fixed

2-Way NE NE +ve NE NE NE -ve NE

Random

2-way NE NE +ve NE NE -ve -ve NE

PSB Random

2-Way NE NE +ve NE -ve -ve -ve NE

PrSB Random

2-Way NE NE +ve NE NE -ve -ve NE

Table 4-12: Summary of Results for Regression on RONW

NE: No Significant Effect on RONW

+ve: Positive Significant Correlation with RONW

-ve: Negative Significant Correlation with RONW

From the above table, it is pretty much clear that Non-interest income ratio which has been

taken as the proxy for the diversification is having a very high significant and positive effect

on the Return on net worth of the banks. The time frame that has been considered in this

paper is one when the banks started to focus on non-fund based activities or fee-based

activities. This is when the banks tried to diverse their portfolio and come out with different

and innovative services like advisory services, investment services etc. From the results, we

can confidently say that this diversification from the traditional fund based activities has

significantly affected the profitability of the banks in the positive manner.

Operating Expense ratio and Interest Expended to Interest Earned Ratio show a significant

negative effect on the profitability of the banks which is quite true in practice also.

Ownership structure does not show any significant effect on the profitability of the banks and

thus the analysis done in this paper poses a question on the myth that the Private Sector banks

are more profitable than the Public Sector banks.

One interesting result shown by the regression model run for the PSB only is that Investment

to deposit ratio is negatively correlated with the RONW. The possible interpretation for this

result is that the Public Sector banks are more conservative in nature and the major

percentage (30-40%) of their investment is on the Government approved securities which are

less profitable (though more safer). On the other hand private sector banks invest in different

avenues which are bit riskier but provide better returns.

Page | 22

b. Summary for Efficiency (NIM) Measure:

The consolidated results for all the regression model run on the NIM is shown in the

following table:

Sample Model RDS LINTNSTY NIIR CDRATIO INVESR IntExpIntEar OWNERSHIP

All

Banks

Fixed 2-

Way NE

-ve

NE NE NE -ve NE

Random

2-way -ve -ve NE NE NE -ve +ve

PSB Random

2-Way NE -ve NE +ve -ve -ve NE

PrSB Random

2-Way -ve NE +ve NE NE -ve NE

Table 4-13: Summary of Results for Regression on NIM

The results are not that consistent in this case. One important result shown by the model is

that Public Sector Banks are more efficient than the Private Sector banks. Interest Expended

to Interest Earned is significantly affecting the banks‟ efficiency in a negative manner. The

results for PSB and PrSB are bit different like RDS is negatively correlated with the NIM in

case of PrSB which shows that the private banks which are having larger deposits relatively

are less than those which are smaller in this aspect. Also in case of PrSB, NII is having a

positive significant effect on the NIM. This shows that the private sector banks try to improve

their efficiency by focusing more on fee-based activities. In case of PSBs, Investment ratio is

negatively correlated to the NIM which tells that the Investment strategies used in the Public

Sector banks are affecting the NIM in a negative manner. CDRATIO is positively correlated

with the NIM for the PSB.

5. Limitations In this project only the listed banks have been considered. There are a considerable number of

other banks which are not listed. Also, foreign banks have not been included in the

considered sample.

In this project only the accounting variables have been considered which are internal to the

banks. There are many macroeconomic factors like GDP, Inflation, Interest Rates, CRR, and

SLR etc. which directly or indirectly affect the performance and efficiency of the banks.

Also, some other important factors like Capital Adequacy Ratio have not been taken into the

account.

Page | 23

6. Annexures Panel Unit Root Test (LLC Test)

Null Hypothesis: Unit root (common unit root process)

Date: 10/02/11 Time: 10:30

Sample: 2002 2011

Exogenous variables: Individual effects

User-specified lags: 1

Newey-West automatic bandwidth selection and Bartlett kernel

Total (balanced) observations: 256

Cross-sections included: 32 Series: RONW Method Statistic Prob.**

Levin, Lin & Chu t* -6.55633 0.0000 Series: RDS Method Statistic Prob.**

Levin, Lin & Chu t* -12.6470 0.0000 Series: LINTNSTY Method Statistic Prob.**

Levin, Lin & Chu t* -12.6367 0.0000 Series: NIIR Method Statistic Prob.**

Levin, Lin & Chu t* -11.1109 0.0000 Series: CDRATIO Method Statistic Prob.**

Levin, Lin & Chu t* -87.7283 0.0000 Series: INVESR Method Statistic Prob.**

Levin, Lin & Chu t* -21.8001 0.0000 Series: INTEXPINTEAR Method Statistic Prob.**

Levin, Lin & Chu t* -13.6823 0.0000

Page | 24

Series: OPEREXPINCOMER Method Statistic Prob.**

Levin, Lin & Chu t* -10.6429 0.0000 Series: NIM Method Statistic Prob.**

Levin, Lin & Chu t* -3.30728 0.0005 ** Probabilities are computed assuming asymptotic normality

Table 6-1: LLC Panel Unit Root Test

Page | 25

7. References Ownership Effects On Bank Performance: A Panel Study Of Indian Banks, Bikram De,

ICICI Research Center Jan‟2003, Paper presented at the Fifth Annual Conference on

Money and Finance in the Indian Economy

The Performance of Indian Banks During Financial Liberalisation, Petya Koeva, IMF

Working Paper, July‟2003

Bank-specific, Industry-specific and Macroeconomic Determinants of Bank Efficiency :

Empirical Evidence from the Thai Banking Sector, Fadzlan Sufian and Muzafar Shah

Habibullah, The Journal of Applied Economic Research 2010 4: 427

Examining Internal Factors That Affect Banks‟ Performance Through Panel Regression

Analysis, Journal of Modern Accounting and Auditing, ISSN1548-658, March 2011, Vol.

7, No. 3, 310-315

Ownership Structure, Performance and Risk in Indian Commercial Banks, Siva Reddy

Kalluru, IUP journal for Applied Finance, ICFAI University Press

The Relative Efficiency of Commercial Banks in Thailand: DEA Approach, International

Research Journal of Finance and Economics, ISSN 1450-2887 Issue 18 (2008)

Baltagi H. Badi, 3rd

Edition, 2005.Econometric Analysis of Panel Data: John Willey and

Sons

Wooldridge Jeffrey M., Econometric Analysis of Cross Section and Panel Data: London:

MIT Press, Cambridge, Massachusetts

Brooks C., 2nd

Edition, 2008. Introductory Econometrics for Finance. New York:

Cambridge University Press

Gujarati Damodar N., Sangeetha., 4th

Edition, 2007. Basic Econometrics: Tata McGraw-

Hill Publishing Co.Ltd.

www.capitaline.com

http://www.iba.org.in/

http://www.rbi.org.in

www.in.yahoo.finance.com

http://papers.ssrn.com

http://www.bseindia.com

http://www.nseindia.com

http://www.capitaline.com/

http://www.iba.org.in/

http://www.rbi.org.in/

http://www.in.yahoo.finance.com/

A Quantitative Analysis of Indian Banks’ Performance and Efficiency-A Panel Regression Approach using EVIEWS 7.0.

Economy & Finance

government

net interest

net interest

operating

variables

indian banking

private sector

public sector