Non-Performing Assets of Banks in India: Efficiency in Management

Swati Raju
Mumbai School of Economics & Public Policy (Autonomous)
University of Mumbai

1
 CONTENTS

Section Title

I Introduction

II Non-Performing Assets: Definition and Classification

III Focus of the Study

IV Theoretical Underpinnings

V Review of Efficiency Studies Relating to Banks in India

VI Methodology

VII Non-Performing Loans: Some Stylized Facts

VIII GNPAs: Bank Level Analysis

IX Data Envelopment Analysis: Results

IX.1 Super Efficiency

IX.2 Input Slacks

X Determinants of Efficiency

XI Asset Quality and Solvency

XII Prompt Corrective Action (PCA) Framework

XIII Conclusions

References

Appendix I Bank wise GNPAs as a percent of Gross Advances: 2011 -2017

Appendix II Efficiency Scores of Individual Banks

Appendix III Changes in GNPA as a percent of Gross Advances and CRAR for
all banks from 2011-2017

Appendix IV Prompt Corrective Action (PCA) Framework

2
I. Introduction

India is a bank based system and commercial banks play a vital role in financial

intermediation in the country and can be observed from the distribution of gross financial

savings of the household sector. Gross savings of the household sector accounted for 54.2

percent of the total savings in the country in 2016-17 and the bulk of household savings is in

the form of bank deposits. Bank deposits as a percentage of total savings increased from 32

percent in 1991 to a high of 57 percent in 2008-09 and has since declined to 41 percent in

2015-16. A marginal decline, however, has been observed in the growth rate of bank

deposits in 2017-18.

Savings held as shares and debentures over this same period have been quite small.

Shares and debentures constituted a mere 8 percent of gross financial savings of the

household sector in 1990-91 and have since then shown mixed performance. It peaked to 9.62

percent in 2007-08 and has since declined to 2.73 percent in 2015-16. Bank credit to the

commercial sector increased by 47 times from Rs. 1654.27 billion in 1990-91 to Rs. 77829.86

billion in 2015-16. Bank credit as a percentage of GDP has almost doubled over this period

from 29 percent in 1990-91 to 57 percent in 2015-16. Further, public sector banks dominate

India’s banking sector and the share of public sector banks and private sector banks can be

seen from Figure 1.

Figure 1: Share of Bank Group in Total Credit of the Banking

System
100

80

60
Percent

40

20

0
2016 2015 2010 2005
PSBs Private Banks Foreign Banks

3
Given this pre-dominance of the banking sector in India’s financial sector, it is imperative

that this sector is financially sound. A sound and healthy banking sector is essential for the

efficient working of the financial sector. Financial soundness is captured through a series of

indicators/ ratios which cover different aspects such as capital adequacy, asset quality,

earnings and profitability, liquidity and sensitivity to market risk. The evolution of these

indicators should indicate potential vulnerabilities of the financial/banking sector point out

possible weaknesses, thereby functioning as tools of macroeconomic and macro prudential

policy analysis (Navajas and Thegeya, 2013). The global financial crisis (GFC) of 2008 has

once again brought into focus the importance of detecting early warning signals of potential

financial or banking sector crises through the use of financial soundness indicators.

Asset quality is one of the most critical facet in determining the overall condition of a bank.

It concerns with the quality of a bank’s loan portfolio and its credit administration program.

A substantial portion of a bank’s assets comprises of loans & carry the greatest amount of

risk to their capital. Other items which can impact asset quality are real estate assets, other

assets, off-balance sheet items and, to a lesser extent, dues from accounts, and premises and

fixed assets. Banks are concerned with the quality of their loans since they provide earnings

to a bank. Asset quality concerns with the balance sheet of a bank and more so with the left-

hand side of a bank balance sheet.

The problem of stressed assets or non-performing assets (NPAs) is not new for India’s

banking sector. Since the reforms of 1991, India has had two episodes of NPA problems, one

during 1997-2002 and the current one after the global financial crisis of 2008. The problem

of NPAs in the current episode started around 2010 and aggravated after 2013. The NPA

problem this time over was broad based and coincided with slowing economic growth,

extended to the priority sector and more acute in the infrastructure sector like iron and steel,

4
textiles, mining and aviation. Although the share of NPAs to gross advances was higher in

the period 1997-2003, its growth was much higher in the current period (Singh, 2016;

Sengupta and Vardhan, 2017). Section II presents the definition and classification of non-

performing assets while Section III highlights the focus of the study. Sections IV and V

contain the theoretical background and a review of efficiency studies pertaining to banks in

India respectively. Section VI discusses the methodology adopted in this study. Sections VII

to X discuss the empirical evidence on non-performing assets for banks in India to include

bank group level and individual bank level analysis; the results of the efficiency estimates

and the determinants of efficiency. Section IX presents the analysis for asset quality and

solvency. Section XII examines the relationship between asset quality indicators, solvency

and efficiency of banks under the Prompt Corrective Action (PCA) framework and Section

XIII contains the conclusions.

II. Non-Performing Assets: Definition and Classification

The terms non-performing assets (NPAs) and non-performingloans (NPLs), are often used

interchangeably and, refer to loans that do not meet their stated principal and interest

repayments. In other words, these are assets which do not generate income. Non-performing

loans, thus, create a problem for asset quality. According to the Reserve Bank of India

Prudential Norms, ‘an asset, including a leased asset, becomes non-performing when it

ceases to generate income for the bank’ (Reserve Bank of India, Master-Circular, 2005).

Prior to 1991, banks in India did not have follow uniform accounting practices for income

recognition, classification of assets, provisioning for non-performing assets, and valuation of

securities held in a bank’s portfolio. Consequently, there was no uniform framework for

capital adequacy requirements and very few banks had a capital adequacy ratio of 8 percent.

5
Prudential norms reflect the regulatory side of banking sector reforms and were introduced

following the recommendations of the Committee on Financial System, 1991 (Narasimham

Committee I) The Committee recommended the implementation of uniform prudential norms

and standards that conformed with the Basel Committee on Banking Supervision. Norms for

income recognition, asset classification and provisioning for the loan portfolio in banks were,

thus, introduced by the Reserve Bank of India in a gradual manner and became fully

operational from March 31, 1996. Assets/ loans were classified into standard, sub-standard

and doubtful. Standard assets (loans) required no provisioning. Loans were classified as sub-

standard when payments become overdue for a period exceeding two quarters (the

international norm was one quarter), sub-standard assets were downgraded to doubtful if they

remained in the sub-standard category for two years (international norm was one year).

Besides, loans with government guarantees were treated as zero-risk assets and not

categorised as non-performing loans despite a payment default. The implementation of the

Narasimham Committee I recommendations resulted in strengthening the banking system in

the country and the capital adequacy ratio of most banks was higher than the mandated 8

percent by March 1998. Although much of this recapitalisation was through government

capital infusion, banks also contributed from their internal reserves and raised capital from

the market. The impact was also vindicated by a decline in the ratio of net NPA to total

advances from 16.3 percent in 1991-92 to 8.2 percent in 1997-98.

With the objective of reviewing the progress achieved in banking sector reform and closely

aligning domestic prudential norms with international best practices, the Committee on

Banking Sector Reforms (Narasimham Committee II) was appointed in December 1997. The

committee made specific recommendations to improve standards and the capital adequacy

ratio for banks was increased to 9 percent (the Committee recommendation was 10 percent)

to be attained by March 31, 2000; a risk weight of 2.5 percent was assigned to investments in

6
government securities along with specifying criteria for classifying non-performing assets

(Ahluwalia, 2000).

The definition of NPAs was tightened in phases andthe RBI Master Circular (2005) defines A

Non-performing asset (NPA) is a loan or an advance where;

(i) interest and/ or instalment of principal remain overdue for a period of more than 90 days

in respect of a term loan,

(ii) the account remains ‘out of order’, in respect of an Overdraft/Cash Credit. (According to

the RBI Master circular, an account should be treated as 'out of order' if the outstanding

balance remains continuously in excess of the sanctioned limit/drawing power but there are

no credits continuously for 90 days as on the date of Balance Sheet or credits are not enough

to cover the interest debited during the same period).

(iii) the bill remains overdue for a period of more than 90 days in the case of bills purchased

and discounted,

(iv) a loan granted for short duration crops will be treated as NPA, if the instalment of

principal or interest thereon remains overdue for two crop seasons.

(v) a loan granted for long duration crops will be treated as NPA, if the instalment of

principal or interest thereon remains overdue for one crop season.

The RBI Master Circular, 2014 added the following to aspects to the definition of a NPA

(vi) the amount of liquidity facility remains outstanding for more than 90 days,in respect of a

securitisation transaction undertaken in terms of guidelineson securitisation dated February

1, 2006.

(vii) in respect of derivative transactions, the overdue receivables representingpositive mark-

to-market value of a derivative contract, if these remainunpaid for a period of 90 days from

the specified due date for payment.

7
Besides, according to the Reserve Bank of India guidelines an account may be classified as

NPA only if the interest charged during any quarter is not serviced fully within 90 days from

the end of the quarter. Banks are required to classify NPAs since March 31, 2005 into three

categories, on the basis of the period for which the asset has been non-performing and the

extent to which dues can be realised into sub-standard assets, doubtful assets and loss assets.

A requirement being that all these ‘assets possess well defined credit weaknesses that

jeopardise the liquidation of debt and are characterised by the distinct possibility that the

banks will sustain some loss, if deficiencies are Not corrected’ (Reserve Bank of India, 2005).

Sub Standard Assets would be those, which have remained NPAs for a period less than or

equal to 12 months.

Doubtful Assets are those which have remained in the sub-standard category for a period of

12 months.

Loss Assets are considered uncollectible and of little value that its continuance as a bankable

asset is not warranted and even though there may be some salvage or recovery value.

III. Focus of the Study

Given the problem of NPAs started in 2010 and has aggravated since 2013, this study will

focus on

(i) Compare the magnitude of the NPA problem for three time periods 2008-10,

2011-13 and 2014-17 at the level of bank groups.

(ii) Quartile analysis of NPAs and comparison at the individual bank level for the

periods 2011-13 and 2014-17.

(iii) Analyze and compare the efficiency of performing loans at the individual bank

level for two time periods 2011-13 and 2014-17. The ability of a bank to manage

its performing or interest earning assets or in other words, a higher efficiency of a

8
 bank in managing its performing assets for a bank would implicitly mean lower

Non-performing assets for the bank.

(iv) Identify the determinants of efficiency (univariate approach using rank correlation

as well as the Tobit analysis)

(v) Analyze whether there is a relationship between asset quality (GNPAs) and

financial soundness (CRAR).

IV. Theoretical Underpinnings

Banks can be viewed as financial intermediaries and /or providers of services and these

different views form the basis for different approaches to study the efficiency of banks. Each

of these views uses a different set of inputs and outputs as the choice of inputs and outputs in

DEA measurement has been a subject of debate and can have an impact on the results. Das

and Ghosh (2006) elucidate the commonly discussed approaches in literature to study and

measure bank efficiency are the production approach, the intermediation approach, the

modern approach and the income based approach. The production and intermediation

approach extend the traditional theory of the firm to banks and differ from each other in terms

of the specification of banking activities.

The production approach was first discussed by Benston (1965) who pointed that banks are

entities that provide services to customers. This approach highlights that banks produce loans

and other financial services. The focus of the approach is on studying the number of type of

transactions undertaken by a bank or services provided over a given period of time. This

approach focuses includes physical variables and/or their costs as only physical inputs are

necessary to perform and process transactions and provide advisory services to customers.

Hence the input set under this approach often includes physical variables like labour (staff),

material or space. Interest costs are excluded from inputs. Outputs are the services provided

9
to customers and is measured by the number and type of transactions, documents processed

or specialized services provided. Given that data on such details is difficult to obtain it is

often proxied by the number of deposits and loan accounts.

The intermediation approach views banks as financial intermediaries who transfer funds from

one set of agents (depositors) to another set of agents (creditors). Assets created by the bank

through loans and advances, securities and investments generate income while deposits of the

bank are liabilities where interest has to be paid by the bank. This approach takes into

accounting interest costs as well as operating costs. However, there are differing views on

whether deposits of a bank should be used as an input or output. This has led to three sub-

approaches under the intermediation approach, namely, the asset approach, the user cost

approach and the value added approach. The asset approach (Sealey and Lindley, 1977)

considers as inputs deposits and other liabilities along with real resources such as labour and

capital while loans alone are considered as output. The user cost approach (Hancock, 1985)

considers a financial product to be an input or output based on its net contribution to bank

revenue. If the financial returns on an asset is greater than the opportunity cost of funds, then

it is treated as an output and if the financial returns are lower than the opportunity cost of

funds then they are treated as inputs. The value added approach is based on the balance sheet

classification of bank activity and those activities that contribute to bank value are treated as

outputs (Berger and Humphrey, 1977; Berger, Hanweck and Humphrey, 1987). Under this

approach, deposits, loans and investments, contribute to bank value and are hence treated as

outputs while labour expenses, capital related operating expenses and interest expenses are

treated as inputs.

Berger and Humphrey (1977) have pointed out that the intermediation approach is better

suited to examine bank level efficiency while the production approach is more apt for

10
studying branch level efficiency as at the bank level the focus is often on reducing costs so as

to be cost efficient. Processing of transactions takes place at the branch level and hence

efficiency in processing the services provided by a bank, the focus of the production

approach, is suited for branch level efficiency.

The modern approach (Frexias and Rochet, 1997) takes into account the quality of bank

assets and the probability of bank failure in the estimation of costs. This approach seeks to

integrate/combine agency costs, some measure of risk and quality of bank services. This

approach is best analysed using the CAMELS framework.

The income based or operating approach (Leightner and Lovell, 1998) considers banks to be

business entities that generate revenue from costs incurred. Hence, under this approach total

revenue (interest and non interest) is considered as output and total expenses (interest and

operating) as input.

Most studies in banking efficiency focus on the intermediation approach as data is easily

available from the Profit and Loss Account and Balance Sheet of the Bank.

V. Review of Efficiency Studies Relating to Banks in India

An important objective of this study is the analysis and comparison of the efficiency of

performing loans of banks in India and examine whether there is any link between asset

quality and solvency for a bank. Hence the review of literature has discussed only those

studies that have used frontier approaches of Data Envelopment Analysis (DEA) and

Stochastic Frontier Analysis (SFA) to measure bank efficiency in India. This section also

contains studies which have analysed bank performance with capital adequacy.

Bhattacharya et al. (1997) examined the productive efficiency of 70 Indian commercial banks

in the pre-deregulation era for the period 1986-91. The paper used DEA and SFA models to

11
measure the efficiency in service provision and found public sector banks to be the most

efficient followed by foreign banks and private banks. The paper further observed that

although foreign banks were least efficient at the beginning of the period by the end of the

period these banks witnessed an improvement in efficiency and that they were as efficient as

the public sector banks.

Rajaraman, Bhaumik and Bhatia (1999) discussed the inter-bank variations in NPAs for

1996-97 using regional clustering. The paper studied the impact of region of operation on

domestically owned banks. The paper regressed bank-specific net NPAs on bank ownership

and regional presence, controlling for bank-specific efficiency and prudential indicators.

Regional presence was measured by taking the total number of branches in each state as a

percentage of the national total. Under ownership structure, the paper classified banks into

four types, namely old private sector banks, new private sector banks, public sector banks

and, foreign banks. Foreign banks were further classified into the group of Asian and West

Asian origin, banks and another group of other than Asian and West-Asian origin. The paper

found significant positive correlation for the cluster comprising three eastern and seven north

eastern states while the cluster for southern and northern states showed a significant negative

correlation. These findings support those of Demirguc-Kunt and Huizinga on the importance

of operating environment for bank efficiency. The paper observed that no major improvement

in performing efficiency of domestic banks is possible without prior improvement in the

operating environment in difficult regions of the country.

Barr (2000) studied the productive efficiency of the US commercial banks for a 15 year

period from 1984-1998 using an input-oriented DEA model. The model was a multiple input-

output model and the output variables were earning assets, interest income, and noninterest

income and the five input variables were expenditure incurred on salary, premises and fixed

12
assets, other noninterest expenses, interest expenses, and purchased funds (large dollar

deposits). The paper undertook a quartile analysis of efficiency and examined using a t test

whether the efficiency scores of best performing banks significantly differed from the worst

banks and found that the efficiency scores across quartiles differed significantly. The paper

also studied efficiency of banks with respect to size and found that smaller banks were more

efficient than larger banks. Further, banks rated as strong on the CAMEL indicators were

found to be more efficient than those rated as weak on the CAMEL indicators. The paper

concluded that efficient banks had lower expenditure on salary, lower fixed assets levels,

lower interest and Non-interest expenses and lesser purchased funds. The paper did not find a

consistent relationship between efficiency and interest income. Also, efficiency was

positively related to the return on average assets.

Rajaraman and Vashishtha (2002) studied the non-performing loans of 27 public sector banks

in India using a fixed effects panel regression framework for a five year period ending 1999-

2000. The objective was to examine the variations in non-performing loans among the public

sector banks which are a homogeneous group on the criterion of ownership. Two models

were estimated with the gross non-performing loans as a percent of gross advances as the

dependent variable, one, capital adequacy as an indicator for solvency and the second with

operation profits as a percentage of working capital which would reflect operating efficiency

as explanatory variables. The paper finds that banks with higher than average NPA’s can be

grouped into those which have poor operating efficiencies and those which have high NPAs

which cannot be explained by operating efficiency and hence with an unexplained intercept

shift. The paper observed that Indian Bank and United Bank of India fall under the second

category and that recapitalization would not solve the problems and were a case for

incentivizing closure whereas closure was not required for UCO Bank.

13
Kumbhakar and Sarkar (2003) employed a stochastic cost frontier model to analyse the

efficiency of public and private sector banks for the period 1986-2000. The paper attested to

the prevalence of cost inefficiency among Indian banks and that inefficiencies have declined

over time. Further cost efficiency among private banks was higher than the public sector

[Link] also found that the decline in inefficiency was slower in the post deregulation period

as compared to the pre-deregulation period. Analysis at the individual bank level showed

substantial intra-group volatility in efficiency changes among private banks as compared to

public banks between the pre and post deregulation period.

Sathye (2003) used data envelopment analysis to measure the productive efficiency of a total

of 94 banks comprising of public sector, private sector and foreign banks for 1997-98 in

India. The paper estimated two models with different measures of inputs and outputs to

measure efficiency. The results obtained indicated that the efficiency score of Indian banks

are comparable with those of international banks and that public sector banks were more

efficient than private sector banks.

Shanmugam and Das (2004) analysed the efficiency of banks in India for 1992–1999 using

stochastic frontier analysis. The data set was an unbalanced panel of 94 banks belonging to

four different ownership groups - State Bank group, nationalized banks, private banks and

foreign banks. The paper considered four outputs, viz., interest margin, non-interest income,

investment and credit and the results indicated that deposits were an important determinant of

output and that the observed outputs are less than their potential levels. The paper also found

that the technical efficiency of raising interest margin varied widely across bank groups and

that reform measures introduced since 1992 had not helped raise interest margins in banks.

The paper found efficiency improvement as regards non-interest income, investments and

14
credits for all bank groups and particularly in the case of private banks along with wide

variations among the sample banks.

Ram Mohan and Ray (2004) analysed the revenue and cost efficiency of public, private and

foreign banks in India for the year 1999-2000 using data envelopment analysis. The paper

found that the variability in revenue efficiency measures (interest spreads and interest spread

net of provisions as a ratio of total assets) was higher than the variability in cost efficiency

(ratio of intermediation cost to total assets). The output variables considered were loans,

investments, and other income and deposits and operating costs were treated as inputs. The

paper, further, decomposed revenue efficiency into allocative efficiency and technical

efficiency. The paper concluded that PSBs were more efficient than private banks and were at

par in efficiency with foreign banks. Further, PSBs outperform private banks in technical

efficiency rather than allocative efficiency. Further, to determine the robustness of their

findings, the paper compared results of efficiency by removing the State Bank of India and

found no major change in the results.

Das et al. (2005) analysed the efficiency of Indian banks using data envelopment analysis for

the time period 1997-2003 and observed that not much differentiation existed between input

oriented or output oriented technical efficiency and cost efficiency. Indian banks, however,

were largely different in terms of revenue and profit efficiency. The paper found that bank

size, ownership and listing on the stock exchange had a positive impact on profit and revenue

efficiency. The paper also observed that the median efficiency scores of Indian banks and

more so of bigger banks had improved during the post reform period.

Sensarma (2005) employed stochastic frontier analysis and estimated bank specific cost and

profit efficiency of all scheduled commercial banks in India for 1986-2003. The paper noted

15
an increase in cost efficiency and a decline in profit efficiency of the banking industry over

this period. The paper also found domestic banks to be more efficient than foreign banks.

Das and Ghosh (2006) used data envelopment analysis to study the performance of Indian

commercial banks in the post reform period (1992-2002). The paper examined three different

approaches – intermediation approach, value added approach and operating approach to

determine if efficiency scores varied with changes in inputs and outputs. The paper indicated

medium sized public sector banks to operate at higher levels of technical efficiency and

pointed that more efficient banks had on an average lower non-performing assets.

Galagedera and Edirisuriya (2005) analysed the technical efficiency of 17 PSBs and 23

private banks using data envelopment analysis and productivity growth using the Malmquist

Index from 1995 to 2002. Efficiency was examined using the intermediation approach with

two inputs (total deposits and total operating expenses) and for two outputs (loans and other

earning assets). Under the CRS – DEA model, the average efficiency of all banks under the

CRS-DEA model showed a mixed trend, an increase from 0.84 in 1995 to 0.95 in 1997 and

followed by a decline until 1999 (to 0.927). A slight improvement in average efficiency was

seen till 2002 (0.94.3). Analysis of technical efficiency for each bank group separately

showed both bank groups had similar efficiency scores of 0.924 for this period whereas the

managerial efficiency of PSBs (0.98) was higher than that of private banks (0.95). The

analysis of productivity change for all banks indicated that there has been no significant

productivity growth during the sample period. In terms of bank groups, the paper observed

little difference in the performance of public sector banks and private banks which can be

attributed to the no TPF growth in private banks and a modest growth in public sector banks.

Sensarma (2006) employed a stochastic frontier model to estimate the cost efficiency of

Indian banks using the value added approach to model bank behaviour. The paper compared

16
the performance of foreign and the new private sector banks with public sector banks for two

types of models, Model 1 - without bank group dummies (which provided a common frontier

for the entire banking sector relative to which the inefficiencies are estimated) and Model 2 -

with the bank group dummies (which created a frontier specific to each bank group). The

output vector consisted of value of fixed deposits, saving deposits, current deposits,

investments, loans and advances and the number of branches (as a proxy for quality of

services or size of bank transactions) while the dependent variable was the total operating

cost. The coefficients for fixed deposits as well as branches was positive as was the cost

elasticity of output. The paper, further, carried out the generalized likelihood ratio test to test

for bank level inefficiency effects and observed that there were significant inefficiencies for

both models. The analysis of cost efficiency concluded that new private sector banks

performed better as compared to the public sector banks and foreign banks. The paper

estimated the total factor productivity for all bank groups and the composite TFP measure

showed a sustained rise in the overall performance in the banking industry for both Model 1

and Model 2 suggesting thereby that the ownership effect is Not very strong. The TFP of

public sector banks increased the highest followed by new private sector banks and foreign

banks.

Ray (2007) evaluated the size and scale efficiency of Indian banks for [Link] paper

employed a modified version of the intermediation approach, with borrowed funds, labour

and equity as inputs and credit, investments and other income as the outputs. The paper used

the Most Productive Scale Size (MPSS) as a benchmark for largeness of a bank which can be

applied to the input bundle of any bank with diminishing returns to scale. If a bank’s input

mix needed to be scaled down in order to achieve MPSS, then it was termed “too large”. In

suchcircumstances, it would be considered optimal to break up the firm into smaller input

bundles. However if the combined output of the smaller input bundles is lower than the

17
combined input bundle, then the firm cannot be broken up. Such a firm would notbe scale

efficient but is size efficient. The results of the paper indicated that the State Bank of India

could be broken down into more than 25 smaller banks to achieve size efficiency; Canara

Bank was also found to be too large and consequently could be broken into at least 10 banks;

Punjab National Bank was not size inefficient for the first two years of study but was later

found to be too large and hence estimated to be broken into 5-11 banks over the years. The

paper, likewise, has highlighted the large size of various banks in India and found 25 percent

of the Indian banks to be too large and hence suggested that banks be broken into smaller

banks in order to achieve higher technical efficiency.

Sensarma (2008) studied the effects of deregulation on the banking sector using profit-based

measures of performance for a panel comprising of 83 Indian banks for the period 1986 to

2005 using stochastic frontier analysis. The paper differentiated between the private, public,

new private and foreign banks in order to get a clearer picture of the impact of the

deregulations and increased competition on each bank group. The paper observed that profit

efficiency had declined over time and de-regulation had no impact on this decline. However,

there are significant ownership effects in deregulation and that post deregulation profit

efficiency of public banks declined, while that of private banks increased. Further, the impact

of bank size on efficiency revealed that profit efficiency of banks increased with size. The

paper also measured profit productivity and observed a fall in the average profit productivity

which is associated with a downward shift in the profit frontier. The decline in profit

productivity was the highest for public sector banks (100 in 1986 to 24 in 2005) followed by

foreign (100 to 29.1) and new private banks (100 to 89). The paper attributed the fall in profit

efficiency and profit productivity could be on account of increased competition in the

industry.

18
Mahesh and Bhide (2008) examined the impact of financial sector reforms on the efficiency

of commercial banks and used stochastic frontier analysis to estimate bank specific cost,

profit and advance efficiencies for the period 1985-2004. The results indicated that

competition had significantly impacted all three measures of efficiency measures. While loan

efficiency had not improved much, cost and profit efficiencies displayed variations for

different bank groups’ post- deregulation.

Zhao (2008) examined the impact of banking reforms on the performance of Indian

commercial banks using a balanced panel from 1992 to 2004. The paper estimated the

Malmquist index of total factor productivity change to find the impact of the reforms on the

risk taking capacities of banks in the framework of the intermediation approach. Earning

assets, performing loans (a proxy for quality of risk management) and fee-based income were

considered as output and total operating cost as input. Given a small sample size, as a

robustness check, the paper further considered two additional input specifications viz. the

monetary stock of fixed assets and the number of employees while the other was interest

payment, other operating costs and personnel costs. The paper investigated the risk taking

behaviour of the market participants by considering another model that used total loans

instead of performing loans keeping all other outputs same. The DEA model was estimated

for the overall period as well as for the following two sub-periods, viz. 1992-97 and 1998-

2004 and it was observed that the average technical (CRS) efficiency of foreign banks

declined from 0.82 to 0.80 while that of private banks (0.71 to 0.78) and public sector banks

(0.74 to 0.80) increased between the periods. Further, the pure technical (VRS) efficiency of

foreign banks remained approx. the same (0.86-0.87), of private banks rose from 0.76 to 0.83

and of public banks rose from 0.85 to 0.90. Finally, scale efficiency of foreign banks declined

from 0.96 to 0.91, of the new private banks rose from 0.95to 0.94 and of the public banks

rose from 0.90 to 0.93. An analysis of returns to scale showed that in 10 of the 12 years

19
under consideration, more than 95 percent public sector banks and more than 50 percent

private banks exhibited decreasing returns to scale which could be attributed to overstaffing

and over-branching whereas in most years, more than 50 percent foreign banks operated at

increasing returns to scale. The results of the Malmquist index indicated that the banking

sector experienced an average annual total factor productivity growth of 5.1 percent over the

entire sample periodwas due to technological progress (4.3 percent on average) as change in

technical efficiency was relatively low.

Kumar and Gulati (2010) analysed trends in cost efficiency for public sector banks in India

for the period 1992-93 to 2007-08 and also examined whether convergence was obtained in

cost, technical and allocative efficiency of banks. The results pointed to the positive impact of

deregulation on cost efficiency of banks. Besides, the technical efficiency of the industry

showed an upward trend whereas the allocative efficiency indicated a declining trend. The

paper also pinpointed that the cost inefficiency of banks in India was more on account of

technical inefficiency rather than allocative inefficiency. The paper observed strong alpha and

beta convergence in cost efficiency for public sector banks in India.

Bhattacharya and Pal (2011) analysed the effects of financial sector reforms on the technical

efficiency of Indian commercial banks from 1989-2009. The results highlighted that banks

were operating at an average efficiency of 64 percent during the sample period and an

efficiency decline was observed in both public and private sector banks for large part of the

post reform period. Further, public sector banks displayed higher efficiency levels as

compared to private sector and foreign banks. The paper also found a negative impact of

capital adequacy ratios and no significant impact of the number of bank branches on

efficiency.

20
Ray (2014) studied the overall cost efficiency of the network of a single large public sector

bank within Calcutta, India using data of 193 branches in 2012. The objective was to find the

optimal number of branches in an area that could provide maximum service at minimum cost.

Credit, deposits and other non-interest income were treated as outputs while labour and

capital were used as inputs. The paper found a wide variation in the credit-deposit ratio

between branches i.e. some branches had deposits of Rs. 2000 per Rs. 100 credit while some

branches had deposits of Rs. 100 per Rs. 700 credit. The paper divided branches into 4

groups to ensure homogeneity based on credit-deposit ratios as group 1 where the ratio was

below 13.75 percent; group 2 (between 13.75 and 18.5 percent); group 3 (between 18.5 and

25 percent) and group 4 had a ratio greater than above 25 percent. There was no major

variation in the average cost efficiency for groups 1 to 3 (around 0.70) while the cost

efficiency of group 4 was found to be low (0.554). Further the paper also measured the

difference between the actual cost and the potential minimum cost and found that the costs of

group 1 banks can be reduced by 13 percent while that of group 4 banks can be more than

50percent, if cost efficiency was to be achieved. Further, if all bank branches were to be cost

efficient than the overall cost efficiency would reduce by 35 percent. Another important

finding of the paper was that the optimal number of branches across groups 1, 2 and 4 be

smaller than the actual number of existing branches indicating significant over branching.

The number of branches in group 3 is near to the optimal number of branches. The paper also

decomposed the overall efficiency into cost efficiency and branching efficiency and pointed

out that saving could come from restructuring of the network primarily through consolidation

of branches.

Tabak and Tecles (2014) compared the economic efficiency of Indian banks for the period

1998-2012 using both the frontier methods – Data Envelopment Analysis (DEA) and

Stochastic Frontier Analysis (SFA). The analysis of efficiency revealed lower average

21
efficiency scores and higher volatility in the DEA model as compared to the SFA model and

notes that this could be due to the fact that DEA model does not separate inefficiency from

the random error. Further, no increasing trend of economic efficiency of banks in the period

of study was seen in both models and that the positive impact of financial reforms on the

banking system had dissipated by 2000. Also, there was a statistically significant correlation

between the efficiency scores from the DEA and SFA models.

Bhatia and Mahendru (2015) compared the efficiency of PSBs for two periods 1990-91 to

2000-01 (reform period) and 2001-02 to 2011-12 (post-reform period) using a two stage DEA

model. The paper used the intermediation approach taking four inputs namely deposits,

borrowings, interest expenses, and operating expenses and three outputs namely investments,

advances, and total income. The paper observed an increase in the mean technical efficiency

of banks from 0.977 in the reform period to 0.986 in the post-reform era for the CRS-DEA

model, from 0.988to 0.994 for the VRS-DEA model and in scale efficiency from 0.989 to

0.992. The paper attributes the lower efficiency in the reform period to the lack of

preparedness of PSBs to reforms, slow technological up-gradation and management

inefficiencies. The analysis, further, observed that PSBs had high average efficiency

accompanied by small average variation in efficiency. The paper estimated a Tobit model to

identify the determinants of efficiency in stage two of the DEA model (CRS as well as VRS)

using the CAMEL framework and found that the debt to equity ratio, non-performing assets

to net advances, the ratio of total investments to total advances and operating expenses to

total expenses, ratio of liquid assets to total assets had a negative and statistically significant

impact on the technical efficiency Further, although capital adequacy ratio, ratio of

government securities to total investments and return on assets had a positive impact on

efficiency it was not statistically significant. The size of a bank, captured by the log of its

total assets, had a negative and significant impact on bank efficiency. The paper concluded

22
that PSBs adapted to reforms which led to a substantial improvement in the CAMEL

indicators.

VI. Methodology

The analysis of bank performance, traditionally, employed the method of financial ratio

analysis. Sherman and Gold (1985) and Yeh (1986), however, pointed out that the ratio

approach however, may not be the best method to assess performance as financial ratios tend

to aggregate several aspects of performance and do not necessarily reflect long term

performance as also the financial ratios approach can be misleading as it depends on an

arbitrary benchmark ratio. Sherman and Gold (1985) were the first to apply frontier methods

to banking efficiency analysis. Recent studies on performance/efficiency measurement have

applied frontier methods and the extensively used methods are the non-parametric Data

Envelopment Analysis (DEA) and the parametric Stochastic Frontier Analysis (Sathye,

2003).The parametric approach requires the specification of a functional form whereas the

Non parametric DEA uses mathematical programming. Seiford and Thrall (1990) write ‘that

the kind of mathematical programming procedure used by DEA for efficient frontier

estimation is comparatively robust’. The mathematical formulations of the DEA models can

be converted to simpler formulations which are easier to estimate using the linear

programming (LP) procedure to estimate relative efficiency in decision making units (DMUs)

and can be used to calculate the best practice production frontier for firms (Ali and Seiford,

1993). A main property of DEA is that it does not require any a priori assumptions about the

functional form and the distribution of the error term (Coelli [Link], 2005).

Efficiency of a DMU, in this case a bank, can be defined as the ability with which it can

convert inputs to outputs and is calculated as the ratio of outputs to inputs. Efficiency can

also be defined as the ratio of minimum inputs that would have to be required to the actual

23
inputs employed to produce a given level of output. Efficiency measures always lie between

0 and 1.

Data Envelopment Analysis

The DEA method allows for the generalization of the single output/input technical efficiency

measure to multiple outputs/inputs by constructing a relative efficiency measure. DEA

models can be estimated either under constant returns to scale or variable returns to scale.

The Constant Returns to Scale (CRS) efficiency is obtained by solving the Charnes, Cooper,

and Rhodes (CCR) model. The CRS model estimates gross efficiency of a DMU which is a

composite of technical and scale efficiency. The efficiency of transforming inputs into output

denotes technical efficiency while scale efficiency estimates that most productive scale size

where the scale at which efficiency is 100 percent. The Banker, Charnes and Cooper (BCC)

model measures the Variable Returns to Scale (VRS) efficiency. This model takes into

consideration the variation in efficiency with respect to the scale of operation and therefore

measures pure technical efficiency. The use of the VRS DEA model allows technical

efficiency to be decomposed into pure technical efficiency and scale efficiency. The scale

efficiency of a DMU can be computed as the ratio of its CRS to VRS efficiency. The CRS

efficiency of a firm is always less than or equal to its VRS efficiency. Thus, other things

being equal, the VRS model gives the highest efficiency score while the CRS model gives the

lowest score. The focus in DEA models on variable returns is mainly to ascertain whether a

DMU exhibits decreasing, increasing or constant returns to scale rather than to quantify the

degree of returns to scale (Fukuyama, 2000).

DEA models can be expressed either as output-oriented (maximization) or input-oriented

(minimisation) models. The output-oriented DEA model seeks to maximize output

production for a given level of resources whereas the input-oriented envelopment model aims

24
to produce the observed output with minimum inputs. Thus, the dual of the output

maximizing DEA model is the input oriented envelopment model and vice-versa

(Ramanathan, 2003). In other words, an input oriented DEA model is one where the output

of the firm is treated as a given or as an assigned task and the efficiency of the firm is

determined by the maximal reduction (equi-proportionate) in all inputs without

compromising the feasibility of the target output.

Consider an industry producing ‘m’ outputs from ‘n’ inputs. An input-output bundle (x, y) is

considered feasible when the output bundle ‘y’ can be produced from the input bundle ‘x’.

The technology faced by the firms in the industry can be described by the production

possibility set.

T = {(x, y): y can be produced from x} (1)

In the single output case, the production function will take the following form:

f(x) = max y: (x, y) ∈T (2)

In the multiple output case, frontier of the production possibility set is the production

correspondence F (x, y) = 1.

Suppose that (xj, yj) is a feasible input-output bundle observed for firm j (j = 1, 2,…, N). The

production possibility set satisfying the assumptions of convexity and free disposability for

such bundles is

N
S = {( x, y ) : x ≥ ∑ λ j x j
j =1
N
y ≤ ∑λj y j
j =1 (3)
N

∑λ
j =1
j =1

λ j ≥ 0( j = 1,2...N )

25
The set S is also known as the free disposal convex hull of the observed input-output bundles.

Different measures of efficiency of a firm can be obtained from using set S as the reference

technology such as output oriented technical efficiency, input oriented technical efficiency,

cost efficiency, and profit efficiency.

An input oriented CRS –DEA model of efficiency for a firm (here a bank) can be expressed

as (used in the study):

(4)

≥
≤
≥0
where, xj is the vector of inputs and yjis the output of firm (here a bank) j. Solving the above

equation helps obtain the minimum inputs and a comparison of minimum to actual inputs

which would provide an estimate of input efficiency.

The measure of technical inefficiency obtained from the input oriented DEA models

corresponds to Farell’s input based measure of technical inefficiency. The estimates of

efficiency obtained from the input oriented and output oriented measures would be similar

under the CRS model but are unequal for the VRS model as the CRS DEA model is only

appropriate when all firms are operating at optimal scale. However, due to imperfect

competition or constraints on finance, firms may not operate at the optimal scale and hence

an input-oriented variable return to scale may be used to calculate technical efficiency.

Further, the choice of orientation (input or output) is dependent on which quantities (inputs or

outputs) the firm/DMU has most control over and are the unit’s primary decision variables

(Coelli, [Link], 2005). Coelli and Perelman (1999) point out that the choice of orientation has

only a minor influence on the scores obtained. Coelli [Link] (2005, p.181) also note that ‘the

output and input oriented DEA models will estimate exactly the same frontier and therefore

26
by definition, identify the same set of firms as being efficient. It is only the inefficiency

measures associated with the inefficient firms that may differ between the two methods’.

VII. Non-Performing Assets: Some Stylized Facts

The asset quality of banks, in general, and of PSBs in particular has seen a decline in recent

years. This decline can be attributed partly to what is now known as the Twin Balance Sheet

Challenge, where the difficult financial position of some large corporate houses had an

impact on the balance sheets of PSBs (Economic Survey, 2016). Another related reason is the

aggressive lending strategy adopted by banks in the period 2008-12. This period saw a

substantial increase in bank credit to the commercial sector. Figure 2 highlights this increase

and the subsequent decline. The percentage of bank credit to the commercial as a percentage

of total non-food credit increased from 67.06 percent in 2009-10 to 69.70 percent in 2013-14

and has shown a decline since then and stands at 61.78 percent as of end March 2018. Other

reasons for rising NPAs could be the poor monitoring of loan accounts and/or the exposure of

a bank to risky/sensitive sector lending.

The aggressive lending strategy adopted by banks coupled with the global financial crises

(GFC) and the consequent twin balance sheet problem provided the rationale for the analysis

to be divided into three sub-periods: 2008-10 when bank credit was high and NPAs were low;

2011-13 – the start of the second episode of the NPA crises after 2010; and the period 2014-

17 when the asset quality review was implemented.

27
 Figure 2: Bank Credit to Commercial Sector as a percent of Total
Non food Credit
72
69.70
70

68 67.06

66
percent

64
61.78
62

60

58

56
2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18

The NPA crises facing the banking sector in India can be inferred by looking at the

magnitude and its increase over the period [Link] total size of the banking sector

GNPAs as of end March 2017 was Rs. 7.91 million (an increase of 14.10 times) from Rs.0.56

million as of end March 2008. A decomposition of the increase in GNPAs of SCBs indicates

that GNPAs increased by 1.50 times during 2008-10, 3.30 times during 2011-15 and 2.40

times during 2015-17. Figure 3 plots the GNPAs of Scheduled Commercial Banks (SCBs) for

the period since 2008 (after the GFC) and it can be observed that the increase has been higher

after 2010 while a significant increase is seen after 2015. This inflection observed in 2015

may be attributed to the Asset Quality Review that banks were asked to undertake by the

Reserve Bank of India. Further, as of end March 2017, GNPAs of the PSBs constituted nearly

86 percent of all GNPAs (Table 1).

28
 Table 1: Gross Non-Performing Assets: Bank Groups (Rs. Million)
Bank Group 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Nationalised
249743 265431 363948 442711 696245 1022272 1484572 2049595 4179878 5069213
Banks
State Bank
154780 184139 235325 303928 482144 627785 798165 735085 1219686 1778106
Group
Public
Sector
Banks 404523 449570 599273 746639 1178389 1650057 2282737 2784679 5399564 6847320
Private
129974 169266 176400 182406 187678 210705 245424 341062 561857 932092
Banks
Foreign
28594 64445 71336 50687 62966 79771 115650 107610 158052 136291
Banks
ALL SCBs 563092 683282 847008 979732 1429033 1940533 2643811 3233352 6119473 7917907

Figure 3: GNPAs: Scheduled Commercial Banks

9000000
8000000 7917907.16
7000000
Rs. Million End March

6000000
5000000
4000000
3000000 2643811.38
2000000
979731.51
1000000 563091.50
0
2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Since GNPAs represent the non-performing assets of a bank, it would be of interest to

examine the proportion of bank loans that are non-performing as a ratio of the gross

loans/advances. Table 2 and Figure 2 present a bank group wise profile of GNPAs as a

percentage of gross advances.

Table 2: GNPAs as a percent of Gross Advances

2008-10 2011-13 2014-17
Nationalised Banks 2.27 2.59 8.62
State Bank Group 2.38 3.93 6.40
Public Sector Banks 2.45 3.00 7.88
Private Banks 3.05 2.02 2.77
Foreign Banks 3.35 2.79 3.93
ALL SCBs 2.32 2.80 6.45

29
It can be observed that GNPAs as a percentage of gross advances of SCBs has increased by

2.78 times over the period 2008-10 to 2014-17. The increase has been much higher for PSBs

(3.22 times) as where a decline is observed for the privates sector banks from 3.05 to 2.77

percent over the same period. It is also of interest to note that during 2008-10 private banks

and foreign banks had a higher percentage of GNPAs compared to the PSBs. The GNPAs as

a percentage of gross loans of PSBs have shown an increase whereas that of private sector

and foreign banks showed a decline in 2011-13 followed by an increase during 2014-17.

Within the PSBs, the increase was higher for nationalised banks (3.80 times) as compared to

the State Bank group (2.68 times) for the period 2008-10 to 2014-17.

Figure 4: GNPA as a percent of Gross Advances: Bank

Groups
10
8.62
9 7.88
8
7 6.40 6.45
Percent

6
5 3.93 3.93
4 3.00 3.05 2.77 3.35
2.59 2.45 2.79 2.80
3 2.27 2.38 2.02 2.32
2
1
0
Nationalised State Bank PSBs PVT Banks Foreign ALL SCBs
Group Banks

2008-10 2011-13 2014-17

A deeper analysis of GNPAs for the years 2014-17 (Table 3) for the bank groups reveals that

GNPAs as a percent of gross advances were much higher in 2016 and 2017 and can be

attributed to the Asset Quality Review (AQR) implemented by the Reserve Bank of India

during August- November 2015-16.

30
 Table 3: GNPAs as a percent of Gross Advances: 2014 to 2017
Year SCBs PSBs Private Banks
Mar 2014 3.93 4.47 1.83
Mar 2015 4.38 5.09 2.15
Mar 2016 7.75 9.65 2.90
Mar 2017 9.76 12.32 4.10

The AQR was basically an exercise to ensure that banks recognize and identify NPAs. The

AQR was conducted by the Reserve Bank of India with a focus to clean bank balance sheets.

The AQR was a more stringent inspection of bank books wherein most large borrower

accounts were inspected to determine if banks had classified their assets according to

prudential norms. Such an exercise became necessary as the Reserve Bank of India realised

that banks were resorting to ever greening of accounts and deliberately deferring bad loan

classification and the consequent provisioning by asking for forbearance. The AQR severely

impacted the performance of banks as several large banks reported losses and it has been

reported that about 200 accounts were identified which banks were asked to consider as non-

performing. Further, banks were given two quarters October-December 2015 and January-

March 2016 to finish asset classification. As a consequence, banks reported losses in both

quarters and some banks for the full financial year. Saha (2016) writes that almost all public

sector banks were affected and that large banks such as Bank of Baroda (Rs.3230crore),

Punjab National Bank (Rs.5367 crore), IDBI Bank (Rs.1376 crore) reported huge losses.

Further, Saha (2016) also pointed out that among the new private banks ICICI Bank and Axis

Bank were severely affected while the impact on HDFC Bank was small as it had no large

exposure to infrastructure projects. Besides, a report in The Times of India, July 31, 2018

pointed that of the 200 companies identified for defaulting on loan payments, most were from

the power sector.

31
A glimpse into the sectoral composition (priority and non-priority sector) of GNPAs for PSBs

and private banks shows that PSBs had a substantially higher level of priority sector GNPAs

as compared to the private banks in 2013. Since 2013 there has, however, been a steady

decline in priority sector GNPAs for both public and private sector banks accompanied by an

increasing profile of non priority sector GNPAs (Table 4).

Table 4: Sectoral GNPAs: Priority &Non Priority Sector

Year PSBs PSBs Private Private

Priority Non Banks Banks
Priority Priority Non
Priority
2013 42.90 57.10 26.00 74.00
2014 36.54 63.46 26.62 73.38
2015 35.66 64.34 22.84 77.16
2016 25.50 74.50 20.96 79.04
2017 24.07 75.93 18.00 82.00

Table 5: Priority Sector GNPAs - Details

Sector March 13 March 14 March 16 March 17
Agriculture 4.7 4.4 8.6 8.3
Medium & Small 5.1 5.2 12.5 11.3
Enterprises
Other Priority 3.0 3.0 3.7 3.7
Sector
As a percent of Total GNPAs
Source: RBI, Annual Report (various issues).

A further analysis of priority sector GNPAs since March 2013 (Table 5) showed that within

the priority sector GNPAs of the Medium and Small Enterprises (MSME) sector has been

high and rest of the priority sector. Besides, the GNPAs of the agricultural sector increased

by 1.76 times while that of the MSME sector increased by 2.21 times.

32
Non-Performing Assets: Write Offs and Recovery

Figure 5 presents the write offs of loans by banks at the level of bank groups and it can be

observed the write offs as a percent of GNPAs have been increasing and have been higher for

private banks compared to public sector banks and within the public sector banks write offs

have been higher for the State Bank group than for nationalised banks.

Figure 5: Write offs as a percent of GNPAs: Bank Group Wise

35 32.27
30

25 21.89
20.47
20 16.58 16.68
15 11.86 14.12 12.92

10 8.42 8.50
6.08
4.74
5 0.94 2.29 3.04 6.88
1.77 0.75
0
Nationalised State Bank PSBs Private Banks Foreign Banks ALL SCBs
Banks Group

2008-10 2011-13 2014-17

The recovery of NPAs has usually been undertaken by the following three channels, namely,

Lok Adalats, Debt Recovery Tribunals and the SARFAESI Act (Table 6). Table 6 highlights

the decline in recovery of NPAs from 47.5 percent in 2012-13 to 30.1 percent in 2015-16 and

a marginal increase to 35 percent in 2016-17. Although, the extent of recovery is quite low,

the SARFAESI Act channel has shown higher recovery as compared to the DRT and Lok

Adalats.

33
 Table 6: Recovery of NPAs

Year LokAdalats DRTs SARFAESI Act

2012-13
Amount involved 66 310 681
Amount recovered 4 (6.1) 44 (14.2) 185 (27.2)
2013-14
Amount involved 232 553 953
Amount recovered 14 (6.0) 53 (0.6) 253 (26.6)
2014-15
Amount involved 310 604 1568
Amount recovered 10 (3.2) 42 (7.0) 256 (16.3)
2015-16
Amount involved 720 693 801
Amount recovered 32 (4.4) 64 (9.2) 132 (16.5)
2016-17
Amount involved 1058 671 1131
Amount recovered 38 (4.0) 164 (24.0) 76 (7.0)
Note: Amount in INR Billion. Figures in parentheses are percentage
of recovery.
Source: Reserve Bank of India, Annual Reports (various issues).

VIII. GNPAs: Bank Level Analysis

The analysis of GNPA at the individual bank level has focused 27 banks and includes all the

public sector banks (PSBs) and the six major new private sector banks viz. Axis Bank, HDFC

Bank, ICICI Bank, IndusInd Bank, Kotak Mahindra Bank and Yes Bank. Figures 6 and 7

graph the GNPAs as a percent of gross advances for individual banks for 2011-13 and 2014-

17 respectively and it is obvious that GNPAs as a ratio of gross advances were much higher

in 2014-17 than in 2011-13. Appendix I contains the year wise and bank wise details of

GNPAs as a percent of gross advances.

34
 Figure 6: GNPAs as a percent of Gross Advances: 2011-13
16.0
14.0
12.0

Percent
10.0
8.0
6.0
4.0 2.39
2.0

Oriental Bank of…

0.0
State Bank Group
Allahabad Bank
Andhra Bank
Bank of Baroda
Bank of India
Bank of Maharashtra
Canara Bank
Central Bank of India
Corporation Bank
Dena Bank
IDBI Bank
Indian Bank
Indian Overseas Bank

Punjab & Sind Bank

Punjab National Bank

UCO Bank

United Bank of India

Vijaya Bank
Axis Bank

ICICI Bank
IndusInd Bank
Kotak Mahindra Bank
YES Bank
Average
Syndicate Bank

Union Bank of India

HDFC Bank
Figure 7: GNPAs as a percent of Gross Advances: 2014-17
16.0
14.0
12.0
10.0
7.23
Percent

8.0
6.0
4.0
2.0
0.0
Oriental Bank of…
Bank of India

Average
State Bank Group

Andhra Bank

Dena Bank
IDBI Bank
Allahabad Bank

Corporation Bank

Indian Bank
Indian Overseas Bank

Syndicate Bank
UCO Bank

Vijaya Bank
Axis Bank
HDFC Bank
ICICI Bank

Kotak Mahindra Bank

Punjab & Sind Bank
Punjab National Bank

IndusInd Bank

YES Bank
Bank of Baroda

Bank of Maharashtra

Central Bank of India

Union Bank of India

United Bank of India
Canara Bank

A quartile analysis of the ratio of GNPA to gross advances at the individual bank level was

undertaken which helped identify the best and worst performing banks during 2011-13 and

2014-17 and the results are in Table 7.

The results of Table 7 show that the range (difference between minimum and maximum

GNPAs) for GNPAs as a percent of Gross Advances of banks during 2011-13 was 3.87

which increased by 3.52 times to 13.62 during 2014-17. Banks in the first quartile were the

best performers as they had the smallest ratio while banks in the fourth quartile were the

worst performers as they had the highest percent of GNPA to gross advances. It can be seen

35
that most of the new private sector banks considered in the study excluding ICICI Bank are

among the best performing banks on the asset quality indicator while ICICI Bank featured

among the worst performers in 2011-13. The country’s largest bank the State Bank group was

also among the worst performers during 2011-13. Interestingly, most of the banks with the

highest percentage of GNPA to gross loans are PSBs and several are common to both periods

under study viz. Punjab National Bank, Indian Overseas Bank, United Bank of India and

UCO Bank.

A quartile rank difference was computed to understand if there has been any movement in

bank performance (measured by GNPAs as a percent of gross loans) measured as a

movement from one quartile to another during the two periods under consideration, namely,

2011-13 and 2014-17. Table 8 displays the quartile ranks of the various PSBs and the six

major new private sector banks in 2011-13 and 2014-17 as also the change in ranks over the

period. A negative rank difference would imply that the bank has moved to a lower quartile

rank while a positive rank difference would imply that the bank has moved up to a higher

quartile. A zero rank would indicate no change in the quartile position and a bank has

continued to be in the same quartile in both periods 2001-13 and 2014-17.

Table 7: GNPA as a percent of Gross Advances: Banks with the Highest and Lowest
Ratios
2011-13 2014-17
Minimum 0.22 (Yes Bank) 0.75 (Yes Bank)
Maximum 4.09 (UCO Bank ) 14.37 (Indian Overseas Bank)
Yes Bank , IndusInd Bank, Yes Bank, IndusInd Bank,
HDFC Bank, Axis Bank, HDFC Bank, Axis Bank,
Best Performers Corporation Bank, Kotak Kotak Mahindra Bank,
(First Quartile) Mahindra Bank, Bank of Vijaya Bank, Syndicate Bank
Baroda
Punjab National Bank, Indian
Punjab National Bank, Indian Overseas Bank, United Bank
Worst Performers Overseas Bank, United Bank of India, UCO, Central Bank
(Fourth Quartile) of India, ICICI Bank, State of India, IDBI Bank, Bank of
Bank Group, UCO Maharashtra

36
It can be observed from Tables 7 and 8 that GNPAs as a percentage of gross loans were

higher in 2014-17 compared to 2011-13 for all banks. The quartile rank difference gives us a

glimpse into individual bank performance on this important parameter on financial health of

[Link] banks – State Bank group (2 quartiles), Syndicate Bank, UCO Bank, Union Bank

of India, Vijaya Bank (2 quartiles) and ICICI Bank (2 quartiles) witnessed an improvement in

terms of quartile ranks. Likewise, six banks – Andhra Bank, Bank of Baroda, Bank of

Maharashtra (2 quartiles), Corporation Bank (2 quartiles), Dena Bank, and IDBI Bank saw

worsened quartile performance between 2011-13 and [Link] remaining 15 of the 27

banks (Allahabad Bank, Bank of India, Canara Bank, Central Bank of India, Indian Bank,

Indian Overseas Bank, Oriental Bank of Commerce, Punjab & Sind Bank, Punjab National

Bank, United Bank of India and Axis Bank have continued to remain in the same quartile.

Further as can be observed from Tables 7 and 8, the average level of GNPAs as well as

GNPAs of individual banks as a percentage of gross advances has increased in 2014-17 as

compared to 2011-13.

The quartile rank movements or the inter quartile shifts of banks from their positions in 2011-

13 to their new quartile positions in 2014-16 indicates that five banks (State Bank group,

Syndicate Bank, Union Bank of India, Vijaya Bank, ICICI Bank)have shown an

improvement; six banks (Andhra Bank, Bank of Baroda, Bank of Maharashtra, Corporation

Bank, Dena Bank, IDBI Bank) have shown a worsening and the remaining sixteen banks

have continued to be static and in the same quartile indicating no shift in quartile ranks (Table

8).

37
 Table 8: Quartile Positions of Banks: 2011-13 and 2014-17
GNPA/ Quartile GNPA/ Quartile
Gross Rank Gross Rank Rank Times
Bank Name Advances 2011-13 Advances 2014-17 Difference Increase
State Bank Group 3.93 4 6.40 2 2 1.63
Allahabad Bank 2.53 3 8.81 3 0 3.49
Andhra Bank 2.44 2 8.14 3 -1 3.33
Bank of Baroda 1.79 1 7.12 2 -1 3.98
Bank of India 2.69 3 9.27 3 0 3.44
Bank of Maharashtra 2.11 2 9.34 4 -2 4.43
Canara Bank 1.93 2 6.55 2 0 3.39
Central Bank of
India 3.90 4 11.24 4 0 2.88
Corporation Bank 1.30 1 7.72 3 -2 5.93
Dena Bank 1.92 2 9.18 3 -1 4.77
IDBI 2.53 3 11.52 4 -1 4.56
Indian Bank 2.14 2 5.70 2 0 2.67
Indian Overseas
Bank 3.22 4 14.37 4 0 4.46
Oriental Bank of
Commerce 2.81 3 8.42 3 0 2.99
Punjab & Sind Bank 1.88 2 6.67 2 0 3.55
Punjab National
Bank 3.05 4 9.72 4 0 3.19
Syndicate Bank 2.34 2 5.38 1 1 2.30
UCO Bank 4.09 4 11.71 4 0 2.86
Union Bank of India 2.83 3 7.52 2 1 2.66
United Bank of India 3.43 4 12.78 4 0 3.73
Vijaya Bank 2.58 3 4.69 1 2 1.82
Axis Bank 1.13 1 2.58 1 0 2.28
HDFC Bank 1.02 1 0.98 1 0 0.97
ICICI Bank 3.89 4 5.53 2 2 1.42
IndusInd Bank 1.01 1 0.94 1 0 0.93
Kotak Mahindra
Bank 1.73 1 2.22 1 0 1.28
YES Bank 0.22 1 0.75 1 0 3.45

38
IX. Data Envelopment Analysis: Results

The DEA model estimated in this section has employed the value based approach and has

used variables from a bank’s balance sheet of a bank (assets & liabilities). According to this

approach, banks create value by acquiring deposits from customers and giving loans. The

model measures the efficiency with which banks are able to manage their performing assets

(in other words, assets which earn interest) which in turn may provide insights into their

ability in managing their non-performing assets. The argument being that if a bank exhibits

high efficiency in managing its performing or interest earning assets then implicitly it would

mean lower non-performing assets for the bank. This reasoning is that NPAs are created ex-

post variable reflect the failure to judge to credit risk. Hence a single output and three inputs,

input oriented CRS-DEA model was estimated to measure bank level efficiency for two time

periods 2011-13 and 2014-17 to measure the efficiency of banks in managing their portfolio

of performing assets. The output variable, thus, was the performing loans of a bank while the

input variables were: expenditure on salary, operating expenses of a bank and expenditure

incurred on fixed [Link] efficiency of performing loans was analyzed and compared for

all the PSBs and the six major new private sector banks for two time periods 2011-13 and

2014-17. It can be seen that average/ mean efficiency of banks in managing their portfolio of

performing assets has remained constant at 0.96 in both time periods. Further, the number of

banks with the maximum efficiency score of 1 is only lower by one bank in the period

[Link] is worth noting that Corporation Bank, Punjab & Sind Bank and Vijaya Bank were

among the most efficient banks in both the time periods. IDBI Bank had an efficiency score

of 1 in 2011-13 but not in 2014-17. (Table 9 and Appendix I). The mean efficiency scores

obtained compare well with the efficiency scores observed in literature and to mention a few

-Das, Nag and Ray (2005), Das and Ghosh (2006), Bhatia and Mahendru (2015), Zhao (2008)

for different time spans.

39
 Table 9: Summary of Efficiency Estimates: DEA Model
2011-13 2014-17
Mean Efficiency 0.9627 0.9605
Minimum Efficiency 0.8747 0.8854
Maximum Efficiency 1 1
No. of Banks with E=1 5 4
Banks with efficiency Corporation Bank Corporation Bank
score 1 Punjab & Sind Bank Punjab & Sind Bank
Vijaya Bank Vijaya Bank,
IDBI Bank YES Bank
YES Bank
No. of Banks with 15 13
efficiency lower than
mean
Source: Author Calculations

Likewise, the number of banks with an efficiency score lower than the mean efficiency score

were marginally fewer in 2014-17 (13 banks) compared to 2011-13 (15 banks). The twelve

banks which had an efficiency score lower than the mean efficiency in both time periods

(2011-13 and 2014-17) are State Bank group, Allahabad Bank, Bank of Baroda, Bank of

India, Bank of Maharashtra, Canara Bank, Central Bank of India, Indian Bank, Indian

Overseas Bank, Punjab National Bank, Syndicate Bank and Union Bank of India.

Table 10: Range of Efficiency Scores of Banks

2011-13 2014-17
Efficiency No. of No. of
Range Banks Percentage Banks Percentage
0.8<= E 0.9 1 3.70 1 3.70
0.9<= E <1 21 77.80 22 81.50
E ==1 5 18.50 4 14.80

A deeper analysis into the distribution of efficiency scores obtained by banks during 2011-13

and 2014-17 shows that the State Bank group was sole bank to have an efficiency score

below 0.9 in both periods while there is no major difference in the number of banks with an

efficiency score between 0.9 and 1 in the two periods (Table 10 and Appendix I). These

40
results indicates that the level of efficiency of bank performance on performing loans has

seen no major difference between the two periods under consideration.

IX.1 Super Efficiency

A decomposition of the efficiency scores of the efficient DMUs has been undertaken using

the super efficiency approach of Anderson and Peterson (1993). Table 11 presents the scores

of efficiency and super efficiency for only the efficient DMUs (or banks that have an

efficiency score of 1). An analysis of super efficiency scores showed that all banks had feasible

solutions and revealed that the efficiency score of 1 actually ranged between the maximum of 1.0375

and minimum of 1 during 2011-13 and between 1.0549 and 1 during [Link], the banks

which were the most efficient (an efficiency score of 1) on their performing loans were the same in

2011-13 and 2014-17 with the exception of IDBI Bank. IDBI Bank had an efficiency score of 1 in

2011-13 and 0.9876 in 2014-17 (see Appendix I).

Table 11: Decomposition of Super Efficiency Scores

2011-13 2014-17

Super Super
Efficiency Efficiency Efficiency Efficiency
Bank Score Score Score Score
Corporation Bank 1 1.0026 1 1.0182
IDBI Bank 1 1.0063 -- --
Punjab & Sind Bank 1 1.0375 1 1.0148
Vijaya Bank 1 1.0057 1 1.0083
YES Bank 1 1.0201 1 1.0549

IX.2 Input Slacks

Slacks in the DEA models represent the additional increase in outputs and/or a decrease in

inputs needed for a unit to become [Link], therefore, exist only for those DMUs

which are identified as inefficient and represent only theleftover portions of inefficiencies;

41
implying that after proportional reductions in inputs or increases in outputs, a DMU cannot

reach the efficiency frontier. Slacks are needed to push the DMU to the frontier (target).

Input slacks refer to the overutilization of inputs to produce a given level of output. A DMU

(in our case a bank) can reduce its expenditure on a given input by the amount of the slack

without reducing its performance with respect to performing loans. Slacks, thus, are

indicative of the potential areas of improvement and suggest an input-output mix which can

improve efficiency (Ozcan, 2014; Zhu, 2014).

Table 12: Mean Input Slacks Values and the Number of Banks
Using Slacks
Slacks 2011-13 2014-17
Salary Expenditure 0.01 0
Operating Expenditure 0.08 0.11
Expenditure – Fixed Assets 0.21 0.22
Number of Banks that used
zero slack 7 banks 9 banks
Number of Banks that used at
least one slack 20 bank 18 banks

Table 12 indicates that the mean value of the input slack was the highest on expenditure

incurred on fixed assets and the least on salary expenditure in both time periods. Further, 7

banks did not use their input slacks during 2011-13 while 20 banks used at least one input

slack. Of the 7 banks, 5 banks had an efficiency score of 1 and hence had no slack and the

other two banks(Syndicate Bank and UCO Bank) did not utilise their input slacks (or reduce

expenditure incurred on inputs) to improve efficiency. In the period 2014-17, of the 9 banks

that did not use the input slack, 4 banks had an efficiency score of 1 and the remaining 5

banks (Andhra Bank, Central Bank of India, Punjab National Bank, Syndicate Bank and

United Bank of India) did not utilise their input slacks to move towards efficiency.

42
X. Determinants of Efficiency

The factors that impact bank efficiency can be studied by looking at the correlations between

the efficiency scores and each of these factors. Correlation analysis, however, would be a

simplistic representation and a univariate approach to understanding the relationship between

efficiency levels of banks and the factors that can determine efficiency. Another way to

model the factors that can impact efficiency is by examining a multivariate Tobit model. This

study has adopted both methods to understand the factors which influence efficiency.

Consequently, the correlation analysis has been undertaken between efficiency scores

obtained by a bank from the DEA model and the level of GNPAs, the exposure of a bank to

sensitive lending (lending exposure of a bank to capital markets, real estate and

commodities), financial soundness captured by the CRAR and the asset size. The results of

the correlation analysis are in Table 13 whereas Table 14 contains the results of the Tobit

model.

There is some sync between the results obtained from the correlation analysis and the Tobit

model. The results of the correlation analysis indicate an inverse and significant relationship

between the efficiency scores and the level of GNPAs. The coefficient of the GNPA variable

is also negative and significant in both time periods in the Tobit analysis. This clearly

indicates that higher the GNPAs lower the efficiency of a bank. A similar inverse and

statistically significant correlation is observed, for both time periods, between the efficiency

scores and the extent of sensitive lending undertaken by a bank. In the Tobit analysis the

relationship is inverse and significant in 2014-17. This implies that higher the exposure of a

bank to sensitive lending lower the efficiency. Asset size of a bank also matters is an

important determinant of efficiency. Larger the asset size lower the efficiency. The

relationship between efficiency scores and CRAR, a measure of financial soundness, is not

43
statistically significant in the correlation and Tobit analysis implying that efficiency of a bank

in managing its performing loans is independent of the capital adequacy ratio. Ownership

was a dummy variable which took the value of 1 for public sector banks and 0 for the six new

private sector banks to understand if ownership is an important determinant of efficiency. The

co-efficient of the ownership variable was not statistically significant in 2011-13 and is

marginally significant in 2014-17. Das [Link] (2000) observed that ownership had a positive

and significant impact on profit and revenue efficiency. The management of a bank (defined

as the ratio of operating expenses to total assets) is also an important determinant of

efficiency.

Table 13: Spearman’s Correlation for the Major Determinants of Efficiency

2011-13 2014-17
GNPAs -0.71*** -0.61***
&Efficiency (-4.99) (-3.89)
Sensitive Lending & -0.67 -0.67***
Efficiency (-4.56) (-4.46)
GNPAs& Sensitive 0.77*** 0.63***
Lending (6.01) (4.03)
Asset Size &Efficiency
Less than 1000 0.16 Less than 2000 -0.38
billion (7 banks) (0.36) billion (9 banks) (-1.07)
1000 – 2000 billion 0.07 2000 – 5000 -0.09
(8 banks) (0.18) billion ( 11 banks) (-0.27)
Greater than 2000 -0.05 Above 5000 -0.46
billion (12 banks) (-0.15) billion (7 banks) (-1.17)

44
 Table 14: Results of the Tobit Model
Variable 2011-13 2014-17
CONSTANT 1.48*** 1.54***
OWNERSHIP -0.008 -0.02*
CRAR 0.00006 -0.001
LSIZE -0.0220** -0.03***
MANAGEMENT -5.76*** -6.89***
LSENSITIVE 0.002 -0.02**
LGNPA -0.013* -0.02***
*** - 1% level of significance ** - 5% level of significance * - 10% level
of significance

XI. Asset Quality and Solvency

This section examines the relationship between asset quality, which is (measured by the

GNPAs as a ratio of gross advances) and solvency of banks (measured by the CRAR).

GNPAs as a ratio of gross advances reflect ex-post the inability to judge credit risk and

examining its relationship with the capital adequacy ratio will indicate whether this mis-

judgement has had an impact on solvency. The objective being to study whether the increase

in GNPAs witnessed has had any impact of the solvency of a bank. The analysis of this

section follows the framework adopted by Rajaraman [Link] (2002) which indicates that there

is no prior expectation that solvency would be a correlate of bank performance with respect

to NPAs. The reason for this could be that banks are mandated to maintain the CRAR and

that the level of GNPAs should not be high as to affect solvency of a bank.

A correlation analysis was conducted between GNPAs as a ratio of gross advances of a bank

and CRAR, this would provide insights on whether asset quality has an impact on bank

solvency. The issue of whether the efficiency of a bank on managing its performing assets is

correlated with the CRAR of a bank was also examined using the efficiency score obtained

by a bank from the DEA model estimated above. The results of the correlation analysis are in

45
Table 15 and indicate that there is a negative and statistically significant relationship between

asset quality and solvency indicating that banks which have high GNPAs as a ratio of gross

advances would have lower CRAR.

Table 15: Results of Spearman’s Rank Correlation -

Asset Quality and Solvency
2011-13 2014-17
GNPAs as a ratio of gross advances & CRAR -0.50*** -0.73***
(-2.90) (-5.30)
CRAR& Efficiency -0.01 0.08
(-0.56) (0.41)

However, the correlation between CRAR and the efficiency achieved by a bank in managing

its performing loans is not statistically significant implying that efficiency scores have no

bearing on the CRAR.

The relationship between asset quality and solvency has also been studied using a fixed

effects panel model for all the 27 banks (21 public sector banks and 6 new private sector

banks).The fixed effects model assumes that something within the bank may impact the

outcome (here GNPAs as a percent of gross advances) variable and which needs to be

controlled. A fixed effects panel model was estimated as the focus was on capturing the

variations over time of the impact of bank specific changes in GNPA levels, after controlling

for an underlying uniform impact of NPAs, on bank specific solvency. Table 16 contains the

results of the fixed effects panel estimation. A baseline model for the entire period 2011-17.

Time effects were also considered by introducing a dummy variable (D2) in the baseline

model. The dummy variable took a value of 0 in 2011-3 and a value of 1 during 2014-17.

46
 Table 16: Results of Panel Data Estimates
Dependent Variable: GNPAs as a ratio of Gross Advances
All Banks – 2011-17
Constant CRAR D2 Rho F No. of
Observations
25.61*** -1.58*** -- 0.24 34.34*** 189
(7.31) (-5.86) (1, 161)
6.87* -0.33 4.40*** 0.23 43.02*** 189
(1.62) (-1.06) (6.54) (2,160)
All Banks – 2011-13
5.17*** -0.20** -- 0.58 4.04** 81
(3.73) (-2.01) (1,53)
All Banks – 2014-17
-0.35 0.61 -- 0.52 0.89 108
(-0.44) (0.94) (1,80)
*** indicates significance at 1% of significance
** indicates significance at 5% of significance
* indicates significance at 10% level of significance

The results indicate an inverse and statistically significant between CRAR and GNPAs as a

percent of gross advances in the baseline model implying that as GNPAs decline there will be

an improvement in the CRAR. However, when the time effects are introduced in the baseline

model, the dummy variable (D2) is statistically significant but the co-efficient of CRAR is

not significant. The results of the panel model estimated for 2011-13 point to an inverse and

statistically significant relationship between GNPAs as a percent of gross advances and

CRAR while for the period 2014-17 the relationship is not statistically significant. A plot of

the average values of the two variables for all the banks in the sample supports the results of

the panel estimates.

Figure 8 plots the average value of GNPAs as a percent of gross advances and the CRAR of

the 27 banks considered in the sample and evident from the graph is the inverse relationship

between these GNPAs and CRAR. As GNPAs as a percent of gross advances was increasing

the CRAR was declining especially in the period 2011-13. After 2014, the CRAR has

stabilised around 12 percent while the GNPAs of banks have shown a marked increase.

47
 Figure 8: GNPAs as a percent of Gross Advances and CRAR:
2011-17
16 14.21 13.67 13.22
14 12.20 12.21 12.30 12.64
12
Percent 10 8.73 11.74
8
6 4.64
3.82
4 2.36 2.88
1.93
2
0
2011 2012 2013 2014 2015 2016 2017

GNPAs as a percent of gross advances CRAR

XII. Prompt Corrective Action Framework

Prompt Corrective Action (PCA) refers to the action initiated by the Reserve Bank of India

when the financial condition of a bank worsens and is below certain limits (trigger points) on

three major financial indicators viz. the net NPA ratio, the CRAR, Return on Assets and

leverage ratio. Policy guidelines for such corrective actions were initiated by the Reserve

Bank of India in 2002 and later revised in 2014 and 2017. The provisions of the revised PCA

framework of 2017 were effective from April 1, 2017 based on the financial health of a bank

on the above specified indicators and was applicable to all banks operating in India. The

breach of any risk threshold would result in the activation of the PCA Framework for a bank.

Although the PCA framework is applicable to all banks operating in the country, it is worth noting

that the 11 banks under this framework are public sector banks, namely, Allahabad Bank, Bank of

India, Bank of Maharashtra, Central Bank of India, Corporation Bank, Dena Bank, IDBI Bank Ltd.,

Indian Overseas Bank, Oriental Bank of Commerce, UCO Bank and United Bank of India under the

PCA framework in 2017. This section concentrates on the performance of banks with respect to two

asset quality indicators - GNPAs as a percent of Gross Advances (this indicator has been represented

in terms of times increase/decrease between 2011 and 2017) and the NNPA ratio ( as of end March

48
2017), and CRAR(as of end March 2017) for banks under the PCA framework. Table 17 presents the

performance on each of these indicators. It can be observed that each of these banks has seen a more

than 10 times increase in GNPAs as a percent of Gross Advances between 2011 and 2017. All the

banks under this framework were at varying Risk Threshold levels, 2 banks at Risk Threshold 1, 6

banks at Risk Threshold 2 and 3 banks at Risk Threshold [Link] Net NPA ratio for 9 of the 11 banks

under this framework was at Risk Threshold 2 as of end March 2017.

All banks considered in the sample, not just those under the PCA framework, had witnessed a decline

in CRAR between 2011 and 2017 but the decline in CRAR for most banks under the PCA framework

was greater than 2 percentage [Link] is worth noting that no bank (under the PCA framework or

otherwise) had a CRAR below the trigger level either in 2014 (below 9 percent and above 6 percent)

or 2017 (below 10.25 percent and above 7.75 percent). Appendix III contains the details of these

indicators for all banks.

Table 17: Asset Quality Indicators, CRAR and Efficiency Scores of Banks under the
PCA Framework
GNPA as a percent
Percentage Efficiency Score
of Gross Advances NNPA
Bank Change in (DEA Model)
(Times Ratio
CRAR 2014-17
Increase/Decrease)
Allahabad Bank 11.96 9.19 -1.51 0.9523
Bank of India 11.94 7.13 -0.03 0.9505
Bank of Maharashtra 15.49 12.26 -2.17 0.9605
Central Bank of India 17.70 10.68 -0.70 0.9213
Corporation Bank 11.23 8.54 -2.79 1.0000
Dena Bank 15.51 11.03 -2.02 0.9748
IDBI Bank Ltd. 21.68 14.08 -2.94 0.9876
Indian Overseas Bank 22.23 14.79 -4.06 0.9400
Oriental Bank of Commerce 12.49 9.33 -2.59 0.9725
UCO Bank 15.65 9.29 -2.78 0.9806
United Bank of India 14.02 10.28 -1.91 0.9609
Average Efficiency Score 0.9620
Source: Author Calculations

Given that the NPA profile of these banks is high, it would be of interest to examine if they

also display lower than average efficiency in managing their performing loans? So the

efficiency scores obtained from the DEA model were juxtaposed with the asset quality

49
indicators and a glance at Table 17 indicates that 6 of the 11 under the PCA framework in

2017 had an efficiency score lower than the mean efficiency score obtained for the 27 banks

in the sample. Interestingly, Corporation Bank (Risk Threshold 1) had an efficiency score of

1 and IDBI Bank Ltd. (Risk Threshold 3) had an efficiency score of 0.9876. Of the 6 banks at

Risk threshold 2, 3 banks had higher than average efficiency and three had lower than

average efficiency. Hence, one can infer that there is no clear relationship between the

efficiency score obtained by the bank from a model which estimates how well it manages its

performing assets and the GNPA profile of a bank under the PCA framework.

IX. Conclusions

The current NPA crises facing banks in India is much larger and more widespread than the

previous one and the total size of the banking sector GNPAs as of end March 2018 stood at

Rs. 12.1 million. The significant increase observed after 2015 could be on account of the

implementation of the Asset Quality Review which made banks recognize and report stressed

assets.

The public sector banks accounted for nearly 88.17 percent of the banking sector GNPAs and

GNPAs as a percentage of gross loans of PSBs have shown an increase whereas that of

private sector and foreign banks showed a decline in 2011-13 followed by an increase during

2014-17. A sector wise decomposition of GNPAs shows an increase in the profile of non

priority sector for both public and private sector banks since 2013. Further, write-offs as a

percent of total GNPAs have been increasing and have been higher for the private banks as

compared to the public sector banks. A bank level quartile analysis of GNPAs as a ratio of

gross advances shows that the new private sector banks considered in the study with the

exception of ICICI Bank were among the best performing banks whereas public sector banks

featured in the worst performing quartile. Banks that had the highest percentage of GNPAs to

50
gross advances and common to both periods of analysis viz. 2011-13 and 2014-17 are Punjab

National Bank, Indian Overseas Bank, United Bank of India and UCO Bank.

The mean efficiency score of a bank in managing its performing loans (obtained from an

input oriented DEA model) remained constant around 0.96 (0.9627 in 2011-13 and 0.9605

2014-17) and scores compare well with efficiency scores observed in literature. The

efficiency analysis also helped identify the twelve banks (all in the public sector) that had an

efficiency score lower than the mean efficiency in both time periods viz.2011-13 and 2014-

17. Also, slack was the highest on expenditure incurred on fixed assets and the least on salary

expenditure. Further, efficiency of banks was impacted by the level of GNPAs, extent of

sensitive lending, asset size and ownership. Besides, a significant relationship exists between

asset quality and solvency but one cannot establish a clear relationship between efficiency

score obtained by a bank and solvency.

51
References
Ahluwalia, M.S. (2000), Reforming India's Financial Sector: An Overview, July 6,
[Link]
m
Ali, A.I., Seiford L.M. (1993), The Mathematical Programming Approach to Efficiency
Analysis in Fried, H.O. and Schmidt, S.S. (eds.) “The Measurement of Productive
Efficiency: Techniques and Applications”, Oxford U.K., pp. 120-159.
Barr, R S, K A Killgo, T F Siems and S Zimmel (2000), Evaluating the Productive Efficiency
and Performance of US Commercial Banks, “Managerial Finance”, vol. 28, no. 8, pp.
3-25.
Benston, G.J. (1965), Branch Banking and Economies of Scale, “Journal of Finance”, vol. 20,
no.2, pp. 312-331.
Berger, A. N., Hanweck, G. A., & Humphrey, D. B. (1987). Competitive viability in banking:
Scale, Scope, and Product Mix Economies, “Journal of Monetary Economics”, vol.
20, pp.501–520.
Berger, A.N., Humphrey, D.B. (1997), Efficiency of Financial Institutions: International
Survey and Directions for Future Research, “European Journal of Operational
Research”, vol. 98, no.2, pp. 175-212.
Bhatia, A. and Mahendru, M. (2015), Assessment of Technical Efficiency of Public Sector
Banks in India Using Data Envelopment Analysis, “Eurasian Journal of Business and
Economics”, vol.8, no.15, pp.115-140.
Bhattacharyya, A., Lovell, C.A.K., Sahay, P. (1997), The Impact of Liberalization on the
Productive Efficiency of Indian Commercial Banks, “European Journal of Operational
Research”, vol. 98, no.2, pp. 332-345.
Coelli, T.J., Perelman, S. (1999), A Comparison of Parametric and Non-Parametric Distance
Functions: With Application to European Railways, “European Journal of Operational
Research”, vol. 117, no.2, pp. 326-339.
Coelli, T.J., Rao, D.S.P., O’Donell, C.J., Battese, G.E. (2005), “An Introduction to Efficiency
and Productivity Analysis”, Second Edition, Springer, New York.
Das, A., Ghosh, S. (2006), Financial Deregulation and Efficiency: An Empirical Analysis of
Indian banks During Post-Reforms Period, “Review of Financial Economics”, vol. 15,
pp. 193-221.
Das, A., Nag, A., Ray, S.C. (2005), Liberalization, ownership and Efficiency in Indian
Banking: A Nonparametric Analysis, “Economic and Political Weekly”, vol. 40,
no.12, pp. 1190-1197.
Frexias, X., and Rochet, J. C. (1997), “Microeconomics of Banking”, Cambridge: MIT Press.
Fukuyama, H. (2000), Returns to Scale and Scale Elasticity in Data Envelopment Analysis,
“European Journal of Operational Research”, vol. 125, pp.93-112.
Galagedera DUA, and Edirisuriya P. (2005), Performance of Indian Commercial Banks
(1995–2002), South Asia Journal of Management, vol.12, no.4, pp. 52-74.
Hancock, D., (1985), Bank Profitability, Interest Rates and Monetary Policy, “Journal of
Money, Credit and Banking”, vol.17, pp.19-202.

52
Kumar, S., Gulati, R. (2010), Dynamics of Cost Efficiency in Indian Public Sector Banks: A
Post-deregulation Experience, Paper Presented at Twelfth Annual Conference on
Money and Finance in the Indian Economy, March 11-12.
Kumbhakar, S.C., Sarkar, S. (2003), Deregulation, Ownership and Productivity Growth in the
Banking Industry: Evidence from India, “Journal of Money, Credit and Banking”, vol.
35, no.3, pp. 403-424.
Leightner, E. J., and Lovell, C. A. K. (1998), The Impact of Financial Liberalization on the
Performance of Thai banks, “Journal of Economics and Business”, vol.50, pp.115 -
132.
Mahesh, H.P., Bhide, S. (2008), Do Financial Sector Reforms make Commercial Banks More
Efficient? A parametric Exploration of the Indian Case, “Margin: The Journal of
Applied Economic Research”, vol. 2, no. 4, pp. 415-441.
Meeusen, W., van Den Broeck, J. (1977), Efficiency Estimation from Cobb-Douglas
Production Functions with Composed Error, “International Economic Review”,
vol.18, no.2, pp.435-444.
Navajas, M.C. and A. Thegeya (2013): “Financial Soundness Indicators and Banking Crises”,
IMF Working Paper, WP/13/263, December 2013.
Ozcan, Y.A. (2014), Healthcare Benchmarking and Performance Evaluation: An Assessment
Using Data Envelopment Analysis, Springer. [Link]
Rajaraman, I. and G. Vashistha (2002): Non Performing Loans of PSU Banks: Some Panel
Results, “Economic and Political Weekly”, vol.37, no.5, pp. 429-433.
Rajaraman, I., S. Bhaumik and N. Bhatia (1999): NPA Variations across Indian Commercial
Banks: Some Findings, “Economic and Political Weekly”, vol.34, no.3/4, pp. 161-
168.
Ram Mohan, T.T. and Ray, S.C. (2004), Comparing Performance of Public and Private
Sector Banks: A Revenue Maximization Efficiency Approach. Economic and Political
Weekly, vol. 39, no.12, pp.1271–1275.
Ramanathan, R. (2003), “An Introduction to Data Envelopment Analysis: A Tool for
Performance Measurement”, Sage Publishers, New Delhi.
Ray, S.C.(2007), Are Some Indian Banks Too Large? An Examination of Size Efficiency in
Indian Banking, “Journal of Productivity Analysis”, vol.27, no.1, pp.41–56.
Ray, S.C. (2014), “Branching Efficiency in Indian Banking: An Analysis of a Demand-
Constrained Network”, University of Connecticut working Paper 2014-34, November
2014.
Reserve Bank of India (2005), Master Circular – Prudential norms on Income Recognition,
Asset Classification and Provisioning pertaining to Advances dated July 1, 2005.
Reserve Bank of India (2014), Master Circular – Prudential norms on Income Recognition,
Asset Classification and Provisioning pertaining to Advances dated July 1, 2014.
Sathye, M. (2003), Efficiency of Banks in a Developing Economy: The Case of India,
“European Journal of Operational Research”, vol. 148, no.3, pp. 662-671.
Sealey, C.W., Lindley, J.T. (1977), Inputs, Outputs and a Theory of Production and Cost at
Depository Financial Institutions, “Journal of Finance”, vol.32, no.4, pp.1251-1266.

53
Seiford, L.M., Thrall, R.M. (1990), Recent Developments in DEA: The Mathematical
Programming Approach to Frontier Analysis, “Journal of Econometrics”, vol. 4, nos.
1-2, pp. 7-38.
Sengupta, R. and Vardhan, H. (2017): Non Performing Assets in Indian Banks: This Time its
Different, Economic and Political Weekly, vol.52, no.12, pp. 85-95, March 25.
Sensarma, R. (2005), Cost and Profit Efficiency of Indian Banks during 1986-2003: A
Stochastic Frontier Analysis, “Economic and Political Weekly”, vol. 40, no.12, pp.
1198-1208.
Sensarma, R. (2006), Are Foreign Banks Always the Best? Comparison of State-Owned,
Private and Foreign Banks in India, “Economic Modelling”, vol.23, no.4, pp.717–735
Sensarma, R. (2008), Deregulation, Ownership, and Profitability Performance of Banks:
Evidence from India, “Applied Financial Economics”, vol. 18, no.19, pp. 1581–1595.
Shanmugam, K.R., Das, A. (2004), Efficiency of Indian Commercial Banks During the
Reform Period, “Applied Financial Economics”, vol. 14, no.9, pp. 681–686.
Sherman, H.D., Gold, F. (1985), Bank Branch Operating Efficiency, “Journal of Banking and
Finance”, vol.9, no.2, pp. 297-315.
Singh, C. and J.S. Brar (2016): “Stressed Assets and Banking in India”, Indian Institute of
Management Working Paper, No. 507.
Tabak, B.M. andTecles, P.L. (2010), Estimating a Bayesian Stochastic Frontier for the Indian
Banking System, “International Journal of Production Economics”, vol. 125, no.1,
pp.96–110.
Yeh, Q. (1996), The Application of Data Envelopment Analysis in Conjunction with
Financial Ratios for Bank Performance Evaluation, “Journal of Operational Research
Society”, vol.47, no.8, pp.980-988.
Zhao T, Casu B, Ferrari A. (2008), Deregulation and Productivity Growth: A Study of Indian
Commercial Banking, International Journal of Business Performance Management,
vol. 10, no.4, pp. 318-343.
Zhu, J. (2014), quantitative Models for Performance Evaluation and Benchmarking: Data
Envelopment Analysis with Spreadsheets, Springer, [Link]
319-06646-2

54
 Appendix I
Bank wise GNPAs as a percent of Gross Advances: 2011 -2017

Average Average
Bank 2011 2012 2013 2014 2015 2016 2017 2011-13 2014-17
State Bank
Group 3.06 4.19 4.55 5.10 4.39 6.60 9.52 3.93 6.40
Allahabad Bank 1.76 1.85 3.97 5.85 5.58 10.10 13.72 2.53 8.81
Andhra Bank 1.39 2.16 3.78 5.44 5.46 8.75 12.91 2.44 8.14
Bank of Baroda 1.38 1.55 2.43 2.99 3.80 10.56 11.15 1.79 7.12
Bank of India 2.26 2.60 3.22 3.47 5.52 13.89 14.20 2.69 9.27
Bank of
Maharashtra 2.50 2.31 1.51 3.22 6.49 9.66 18.00 2.11 9.34
Canara Bank 1.49 1.73 2.58 2.51 3.95 9.74 10.00 1.93 6.55
Central Bank of
India 1.85 4.93 4.92 6.49 6.30 12.62 19.55 3.90 11.24
Corporation
Bank 0.91 1.27 1.73 3.46 4.90 10.36 12.14 1.30 7.72
Dena Bank 1.88 1.69 2.21 3.37 5.57 10.40 17.39 1.92 9.18
IDBI Bank 1.77 2.52 3.29 5.04 6.09 11.52 23.45 2.53 11.52
Indian Bank 0.98 2.05 3.38 3.73 4.51 6.84 7.73 2.14 5.70
Indian Overseas
Bank 2.76 2.79 4.12 5.13 8.69 18.68 24.99 3.22 14.37
Oriental Bank of
Commerce 2.00 3.20 3.24 4.04 5.28 9.87 14.49 2.81 8.42
Punjab & Sind
Bank 1.00 1.65 2.99 4.46 4.83 6.62 10.80 1.88 6.67
Punjab National
Bank 1.81 2.97 4.36 5.41 6.75 13.54 13.20 3.05 9.72
Syndicate Bank 2.43 2.57 2.02 2.65 3.18 6.87 8.82 2.34 5.38
UCO Bank 3.18 3.54 5.56 4.43 6.97 16.61 18.83 4.09 11.71
Union Bank of
India 2.40 3.06 3.03 4.17 5.10 9.04 11.77 2.83 7.52
United Bank of
India 2.53 3.45 4.30 10.82 9.82 13.92 16.56 3.43 12.78
Vijaya Bank 2.58 2.97 2.20 2.44 2.82 6.77 6.75 2.58 4.69
Axis Bank 1.12 1.06 1.22 1.37 1.46 1.80 5.70 1.13 2.58
HDFC Bank 1.06 1.02 0.97 0.99 0.94 0.95 1.06 1.02 0.98
ICICI Bank 4.64 3.73 3.31 3.10 3.90 6.02 9.08 3.89 5.53
IndusInd Bank 1.02 0.99 1.03 1.13 0.82 0.88 0.93 1.01 0.94
Kotak Mahindra
Bank 2.06 1.57 1.56 2.00 1.87 2.39 2.63 1.73 2.22
YES Bank 0.23 0.22 0.20 0.31 0.41 0.76 1.53 0.22 0.75
Average 1.93 2.36 2.88 3.82 4.64 8.73 11.74 2.39 7.23

55
 Appendix II
Efficiency Scores of Individual Banks

Bank Name 2011-13 2014-17 Bank Name 2011-13 2014-17

Punjab & Sind
State Bank Group 0.8747 0.8854 1 1
Bank
Punjab National
Allahabad Bank 0.9512 0.9523 0.9217 0.9328
Bank
Andhra Bank 0.9894 0.9727 Syndicate Bank 0.9587 0.97

Bank of Baroda 0.9465 0.9496 UCO Bank 0.9931 0.9806

Union Bank Of
Bank of India 0.9505 0.9505 0.9464 0.9528
India
Bank of United Bank Of
0.9559 0.9605 0.9838 0.9609
Maharashtra India

Canara Bank 0.9401 0.9434 Vijaya Bank 1 1

Central Bank of
0.9327 0.9213 Axis Bank 0.9508 0.9653
India

Corporation Bank 1 1 HDFC Bank 0.9147 0.9445

Dena Bank 0.9989 0.9748 ICICI Bank 0.9264 0.9439

IDBI Bank Ltd. 1 0.9876 IndusInd Bank 0.9827 0.984

Kotak Mahindra
Indian Bank 0.9524 0.9638 0.9289 0.9657
Bank
Indian Overseas
0.9519 0.94 YES Bank 1 1
Bank
Oriental Bank of
0.9696 0.9725 Average 0.9600 0.9620
Commerce

56
 Appendix III
Change in GNPAs as a percent of Gross Advances and CRAR
(All banks from 2011 to 2017)

GNPA as
a percent
of Gross
Bank Advances CRAR
State Bank Group 6.46 -1.34
Allahabad Bank pca 11.96 -1.51
Andhra Bank 11.52 -2.00
Bank of Baroda 9.77 -2.28
Bank of India pca 11.94 -0.03
Bank of Maharashtra pca 15.49 -2.17
Canara Bank 8.52 -2.52
Central Bank of India pca 17.70 -0.70
Corporation Bank pca 11.23 -2.79
Dena Bank pca 15.51 -2.02
IDBI pca 21.68 -2.94
Indian Bank 6.74 0.08
Indian Overseas Bank pca 22.23 -4.06
Oriental Bank of Commerce pca 12.49 -2.59
Punjab & Sind Bank 9.80 -1.89
Punjab National Bank 11.39 -0.76
Syndicate Bank 6.39 -1.01
UCO Bank pca 15.65 -2.78
Union Bank of India 9.37 -1.16
United Bank of India pca 14.02 -1.91
Vijaya Bank 4.17 -1.15
Axis Bank 4.58 2.30
HDFC Bank 0.00 -1.67
ICICI Bank 4.44 -2.15
IndusInd Bank -0.08 -0.58
Kotak Mahindra Bank 0.57 -3.15
YES Bank 1.29 0.50
Average 9.81 -1.57
pca indicates bank under PCA Framework

57
 Appendix IV

Prompt Corrective Action Framework

The PCA Framework: Areas, Indicators and Triggers, 2014
The PCA framework is applicable only to commercial banks and not extended to co-
operative banks, non-banking financial companies (NBFCs) and [Link] trigger points
along with structured and discretionary actions that could be taken by the Reserve Bank are
described below:

Area & Indicator Risk Threshold Action to be Initiated by RBI

Capital less than 9%, but bank to submit capital restoration plan;
equal or more than restrictions on RWA expansion, entering into
CRAR 6% new lines of business, accessing/renewing costly
deposits and CDs, and making dividend
payments; order recapitalisation; restrictions on
borrowing from inter-bank market, reduction of
stake in subsidiaries, reducing its exposure to
sensitive sectors like capital market, real estate
or investment in non-SLR securities, etc.
CRAR less than in addition to actions in hitting the first trigger
6%, but equal or point, RBI could take steps to bring in new
more than 3% Management/ Board, appoint consultants for
business/ organizational restructuring, take steps
to change ownership, and also take steps to
merge the bank if it fails to submit
recapitalization plan.
CRAR less than in addition to actions in hitting the first and
3% second trigger points, more close monitoring;
steps to merge/amalgamate/liquidate the bank or
impose moratorium on the bank if its CRAR
does not improve beyond 3% within one year or
within such extended period as agreed to
Asset Quality over 10% but less special drive to reduce NPAs and contain
than 15% generation of fresh NPAs; review loan policy
Net NPAs and take steps to strengthen credit appraisal
skills, follow-up of advances and suit-
filed/decreed debts, put in place proper credit-
risk management policies; reduce loan
concentration; restrictions in entering new lines
of business, making dividend payments and
increasing its stake in subsidiaries.
15% and above In addition to actions on hitting the above trigger
point, bank’s Board is called for discussion on
corrective plan of action.
Profitability less than 0.25% restrictions on accessing/renewing costly
deposits and CDs, entering into new lines of
Return on Assets business, bank’s borrowings from inter-bank

58
 market, making dividend payments and
expanding its staff; steps to increase fee-based
income; contain administrative expenses; special
drive to reduce NPAs and contain generation of
fresh NPAs; and restrictions on incurring any
capital expenditure other than for technological
upgradation and for some emergency situations.
Source: Reserve Bank of India, Circular, 2014.

PCA Matrix: Areas, Indicators and Risk Thresholds Indicator, 2017

59
The PCA framework would apply without exception to all banks operating in India including
small banks and foreign banks operating through branches or subsidiaries based on breach of
risk thresholds of identified indicators.

60
1
CET 1 ratio – the percentage of core equity capital, net of regulatory adjustments, to total risk weighted assets
as defined in RBI Basel III guidelines
2
NNPA ratio – the percentage of net NPAs to net advances
3
ROA – the percentage of profit after tax to average total assets
4 Tier 1 Leverage ratio – the percentage of the capital measure to the exposure measure as defined in RBI
guidelines on leverage ratio.
Source: Reserve Bank of India Circular, 2017.

61
62