Asset Quality Based Ranking of Indian Commercial Banks: A Superefficiency Approach

Transcription

1 Research Note Artha Vijnana Vol. L, No. 1, March 2008, pp Asset Quality Based Ranking of Indian Commercial Banks: A Superefficiency Approach The present paper tries to make an asset quality based ranking of select (28) Indian commercial banks for four years period to using the super efficiency approach - a non-parametric tool. The super efficiency approach due to Andersen and Petersen (1993), allows rank order technically efficient firms. In this approach, a firm which is under evaluation is not included in the reference set of the relative envelopment model. A firm is called super efficient if its observed output exceeds what is necessary for the firm to be considered efficient relative to other firms in the sample. Out of the 28 observed commercial banks, the number found to be super efficient was only in the range between 5 to 8 during observed years. The rest were found to be inefficient. Further, the results obtained from the exercise indicate gradual improvement in mean technical efficiency scores (excepting ) over the observed years. The observed private sector banks exhibit higher mean technical efficiency relative to the observed public sector banks for all the observed years. Introduction In the pre-1991 period, the commercial banking sector in India operated under a regime of financial repression where the allocation of resources and pricing of deposits and loans were, to a large extent, administered. In the nineties, however, the banking environment experienced radical transformation consequent on the dismantling of entry barriers, rate deregulation, introduction of prudential accounting norm and the implementation of Basel I capital adequacy norms. The changed competition scenario compelled the commercial banks to provide unprecedented attention towards reduction in operating cost. Inter alia, improvement of asset quality has become an important agenda for the commercial banks in the reform years. This is likely to gain further momentum in the coming years in view of the introduction of Basel II. In the context of the changed operational scenario in the Indian banking sector during the reform period, the present paper tries to make an asset quality based ranking of select (28) Indian commercial banks (20 public sector and eight private sector commercial banks) for the four year period to Ram Pratap Sinha, Assistant Professor and Head, Department of Economics, ABN Seal (Government) College, Coochbehar , West Bengal, rp1153@rediffmail.com, and Biswajit Chatterjee, Professor of Economics and Dean, Faculty of Arts, Jadavpur University, Kolkata , West Bengal, chatterjeeb@vsnl.net

2 using super efficiency approach - a non-parametric tool. For inter bank comparison purposes Performing Asset has been taken as the output indicator. The paper is divided into five sections. Section I provides a comparative picture of the public, private bank groups in respect of asset quality. Section II provides a brief review of the literature on efficiency of Indian Commercial banks. Section III discusses the methodology. Section IV states and analyses the results obtained from the exercise. Finally, Section V provides the concluding observations I Commercial Banks Operating In India: Asset Classification Guidelines and Trends in Asset Quality, to Indian Asset Classification System The incidence of credit risk, for practical purposes, is usually measured in terms of the proportion of non-performing assets in a banks total asset portfolio. In developed countries an asset is considered non-performing if payment of interest/principal remain unpaid for a period of 90 days or more. The details and classification standards of non-performance however many from one country to another depending on the circumstances prevailing. In India, a comprehensive and uniform asset classification system was introduced by the RBI in with the objective of appropriate credit monitoring. The Health Code classification of 1985 classified the assets into eight categories (satisfactory, irregular, sick (viable), sick (non-viable/sticky), advances recalled, suit filed accounts, decreed debts and bad and doubtful debts) depending on the quality of credit subsequently an income recognition norm was introduced in 1989 which prohibited the banks to recognize income in respect of categories six to eight (later category five was also included) on the accrual basis. This Health Code system was replaced by a new one in following the recommendation of the Committee on Financial System (1991). The Health Code System has been continuously upgraded during the past one decade. As per the present Health Code System (1992), assets have been classified into four categories Standard, Substandard, Doubtful and Loss. The last three categories are considered non-performing. The total assets falling under the categories 1-3 constitute the Gross Non-Performing of the bank. Criteria for Asset Classification Form the year ending March 31, 2004, banks have adopted 90 day norm for recognizing an asset to be substandard. With effect from March 2005, an asset is classified as doubtful if it remained in the substandard category for 12 months. An account is considered as loss, without reference to any waiting period, if the dues are considered uncollectable/marginally collectable while comparing NPA figures, value of collateral is not deducted from the balance outstanding.

3 Ranking of Indian Commercial Banks 81 In India, we have another concept of NPA -the Net NPA. Net NPAs are obtained from Gross NPAs after deduction of the following: i) Balances in interest suspense account i.e., interest due but not received. ii) Claims received on credit guarantors and kept in suspense accounts pending final settlement. iii) Part payments received and kept in suspense account. iv) Total provisions held. As per the present income recognition norm, income should not be recognised on non-performing assets on accrual basis but should be booked only when it is actually realized in respect of such accounts. Finally the aspect of provisioning. The provisioning norms have become tighter and tighter over the years. The banks are required to make provision against the NPAs at 100 per cent for loss assets. In respect of doubtful assets the requirement is as follows: Hundred per cent provisioning is to be made in respect of the unsecured portion. In respect of the secured portion, the provisioning requirement is 20 to 50 per cent depending on the period for which the account has remained in the doubtful category. Twenty per cent upto one year, 30 per cent for 1-3 years and 50 per cent for more than three years. The Committee On Banking Sector Reform (1998) recommended one per cent general provisioning norm on standard assets. With effect from March 31, 2000 banks have been advised to provide for standard assets at a minimum of 0.25 per cent. The provisioning requirement for assets belonging to the substandard categories is 10 per cent. In respect of small advances of Rs /- and below, with effect from 31 st March 1998, banks are required to make the four way asset classification. In case they are not able to do so, they are required to provide at the rate of 15 per cent of aggregate outstanding including performing loans. Asset Quality Profile of Indian Commercial Banks During the period to , there has been a secular decline in the proportion of non-performing asset in the portfolio of the Indian commercial banks. Tables 1 and 2 show how the relative composition has changed during the period. The tables show that not only the asset quality in general has improved, but there also has been a significant decline in the quantum of doubtful and loss assets.

4 82 Table 1: Asset Composition of Public Sector Banks ( ) (figures in Rs. Crores) Year Standard Per cent of Total Asset Substandard Doubtful Loss Total Non- Performing Per cent of Total Asset Source: R.B.I. (2005), Statistical Tables Relating to Banks in India, , Table 2: Asset Composition of Private Sector Banks ( ) (figures in Rs. Crores) Year Standard Per cent of Total Asset Substandard Doubtful Loss Total Non- Performing Per cent of Total Asset Source: R.B.I. (2005), Statistical Tables Relating to Banks in India, II Previous Studies on Bank Productivity and Efficiency Bhattacharyya, Lovell and Sahay (1997) measured technical efficiency of 70 commercial banks operating in India for the period using data envelopment analysis. They used a two stage approach: in the first stage they calculated the radial technical efficiency scores using DEA. In the second stage they used stochastic frontier analysis to attribute variation in efficiency scores to three sources: temporal, ownership and noise component. In their study, Bhattacharyya, Lovell and Sahay took advances, investment and deposits as outputs while interest expense and operating expense were taken as the inputs. They found that the public sector banks had much higher efficiency as compared to the private and foreign banks. Saha and Ravisankar (2000) examined the efficiency of the Indian public sector banks in two phases during the period In the first phase, certain key ratios like deposit to establishment expenses and advances to establishment expenses; and deposits to staff and advances to staff were considered. The banks were plotted in a two dimensional graph to identify the better performing banks. Saha and Ravisankar considered four input variables-

5 Ranking of Indian Commercial Banks 83 interest expenditure, establishment expenditure, non-establishment expenditure and six output variables: deposits, advances, investments, non interest income, interest spread and total income. The results obtained by them show that the performance of the public sector banks (with the exception of a few) had improved over the years of study. Rammohan and Ray (2004) compared performances of 58 public, private sector and foreign banks using a revenue maximisation efficiency approach for the period Loans, investments and other incomes were taken as bank outputs. Rammohan and Ray took the following inputs for their study- deposits and operating costs. They argued that during the period, Indian banks did not have much freedom in trimming costs especially the cost of labour. Under the circumstances, revenue maximisation best describes the objective that banks have been focusing during the period. The results obtained by Rammohan and Ray relating to revenue maximization efficiency, technical efficiency and allocative efficiency reveal the following: a) Public sector banks are significantly better placed than private sector banks on revenue maximization efficiency but there is no difference between public sector banks & foreign banks. b) Public sector banks are significantly better than private banks in respect of technical efficiency but not in respect of allocative efficiency. Das, Nag and Ray (2005) examined the output oriented technical efficiency, cost efficiency, revenue maximizing efficiency and profit efficiency of Indian (public, private and foreign) banks for the period They considered four inputs for their study-borrowed funds (deposits and other borrowings), number of employees, fixed assets and equity. They included in their study only those banks which had at least three branches during the entire study period. The results obtained by Das, Nag and Ray show that the Indian banks are still not much differentiated in terms of input or output oriented technical efficiency or cost efficiency. However, they differ sharply in respect of revenue and profit efficiencies. Sinha (2007) made a comparative study of the asset quality profile of the Indian commercial banks for the reform period using the stochastic frontier approach. Further, he analysed the impact of factors like bank operating efficiency, capital adequacy ownership and bank size on the asset quality of the observed commercial banks. From the technical efficiency scores relating to asset quality, it appeared that the public sector commercial banks have outcompeted their private sector counterparts. The econometric results showed that operating profit ratio and capital adequacy are two important determinants of asset quality. However, the effect of size and ownership were found to be statistically insignificant. This is probably suggestive of the fact that the quality of bank management (which gets reflected in operating efficiency, capital

6 84 adequacy and asset quality) is the key variable in separating good banks from bad banks. Sinha (2008) compared select public and private sector commercial banks in respect of priority sector lending for the period to For comparison purposes the following indicators have been used: (i) technical efficiency, (ii) scale efficiency and (iii) Malmquist Total Factor Productivity Index. The results obtained from the exercise indicate substantial fluctuations in mean efficiency scores for the observed years. The mean technical efficiency scores of the observed public sector commercial banks is, however, marginally higher than the observed private sector banks. In so far as total factor productivity growth is concerned, the observed private sector commercial banks exhibited marginally higher Malmquist TFP Index than the observed public sector banks. All the observed commercial banks registered positive total factor productivity growth during the period. III Measurement of Technical Efficiency Concept of Technical Efficiency The performance of productive units is often described in terms of efficiency which implies a comparison of actual output/input to optimal output/input. Productive efficiency has two components: technical and allocative. The purely technical component refers to the ability to avoid waste by producing as much output as input usage allows, or by using as little input as output production allows. The allocative or price component refers to the ability to combine inputs and output in optimal proportions in light of prevailing prices. Koopmans (1951) gave a formal definition of technical efficiency: a producer is technically efficient if an increase in any output requires a reduction in at least one other output or a increase in atleast one input, and if a reduction in any input requires an increase in at least one other input or a reduction in at least one output. A popular measure of technical efficiency is by Debreu (1951) and Farrell (1957). In case of output/revenue maximisation, their measure is defined as one minus the maximum equiproportionate expansion in all outputs with given input. In case of cost minimisation, it is one minus the maximum equi-proportionate reduction in all inputs that still allows continued production of given output a score less than unity implies that the firm is technically inefficient. The Data Envelopment Approach Data Envelopment Analysis is a non-parametric mathematical programming technique used for assessing/evaluating and comparing the relative performances of economic units with minimal prior assumption on input-output relation. The DEA method is a generalisation of Farrell s Single input single output technical

7 Ranking of Indian Commercial Banks 85 efficiency measure to the multiple output-multiple input case. The methodology was originally developed by Charnes, Cooper and Rhodes (1978) and was later further extended by Banker, Charnes and Cooper (1984). The DEA approach forms the efficiency frontier out of piecewise linear stretches thereby forming a convex production possibility set. In DEA frontier, efficient observations are those for which no other decision making unit or linear combination of units has as much or more of every output (given inputs) or as little or less of every input (given outputs). Computation of Technical Efficiency The production possibility set of any productive organisation can be represented by two alternative but equivalent ways in terms of the input and output set. For any output bundle y 0 the input requirement set is V(y 0 ) = { x: (x, y 0 ) Ps} Similarly for any input bundle x 0, the producible output set is P (x 0 ) = { y : (x 0, y) Ps } There are accordingly three alternative ways of computing technical efficiency: the revenue maximization approach, the output maximisation approach and the cost minimisation approach. In this paper the output maximization approach has been followed. In the output maximisation approach, the firm seeks to maximise output given the input bundle. As per the Charnes, Cooper,Rhodes Approach (1978) orientation (under the assumption of variable returns to scale) the problem is, Max φ 0 0 s. t. φy λy, x λx, λ j 0 The Concept of Super Efficiency One problem with the Charnes Cooper Rhodes (1978) model for construction of production frontier is that the technically efficient firms are all ranked equally. This is so because any firm which is being evaluated is included in the reference set for the purpose of construction of the frontier. Andersen and Petersen (1993), however, suggested a criterion which allows rank order technically efficient firms. In this approach, a firm which is under evaluation is not included in the reference set of the relative envelopment model. This allows us to identify super efficient firms. A firm is called super efficient if its observed output exceeds what is necessary for the firm to be considered efficient relative to other firms in the sample. Between two super efficient firms, the one with more elbow room for reducing output without being categorised as inefficient is more super efficient than the other one.

8 86 IV Framework of the Present Study The present study follows the radial super efficiency approach under constant returns to scale. The output oriented model has been followed. The mathematical program for the computation of technical efficiency is as under: sup er er * Maxφ sup φ = s. t. φ x 0 sup er y 0 λjx λjy, j 0 sup er φ, λj 0, j 0 Technical Efficiency =1/Ф super It is now clear that the efficiency scores from this model are obtained by eliminating the data on the firm to be evaluated. Note that since the firm under evaluation is not included in the reference set for the computation of technical efficiency, the resultant technical efficiency can be greater than one. Implication of Exclusion of Observation for the Computation of Technical Efficiency The frontier of the production possibility set in any DEA application is formed by a subset of the observed input-output bundles. Deletion of any one of these observations from the data set results in a revision of the frontier causing the measured efficiency level of some of the other observations in the data set to change. It may be noted in this context, however, that if a firm is not technically efficient, it can never play a role in defining the benchmark input-output bundle for evaluating the efficiency of any other firm. For a formal proof of this, see Ray (2004). Output- Input Relationship Since we are interested to compare the observed commercial banks in respect of asset quality, Performing Asset (=Total Asset- Net Non Performing Asset) has been taken as the output indicator. Net worth, borrowed capital and branch have been taken as the inputs. Commercial banks depend significantly on the branch network for the growth of their asset base. The importance of borrowed capital (i.e, deposits and borrowings) can not be undermined. Empirical evidences suggest that the commercial banks with high net worth have better asset quality. Sinha (2007) found that asset quality of Indian commercial banks is positively related to their operating profit ratio and Tier I CRAR. Therefore, we have Output (Performing Asset) = f(net Worth, Borrowed Capital, No. of Branches)

9 Ranking of Indian Commercial Banks 87 Descriptive Statistics of Input/Output Variables Tables 3-6 present the descriptive statistics of the output/input variables. Table 3: Descriptive Statistics ( ) (Figures for Net Worth, Borrowed Capital and Performing Asset are in Rs. Crores) Particulars Net Worth Branches Borrowed Capital Performing Asset Max Min Average S.D Table 4: Descriptive Statistics ( ) (Figures for Net Worth, Borrowed Capital and Performing Asset are in Rs. Crores) Particulars Net Worth Branches Borrowed Capital Performing Asset Max Min Average S.D Table 5: Descriptive Statistics ( ) (Figures for Net Worth, Borrowed Capital and Performing Asset are in Rs. Crores) Particulars Net Worth Branches Borrowed Capital Performing Asset Max Min Average SD Table 6: Descriptive Statistics ( ) (Figures for Net Worth, Borrowed Capital and Performing Asset are in Rs. Crores) Particulars Net Worth Branches Borrowed Capital Performing Asset Max Min Average SD Estimates have been made under constant returns to scale.

10 88 Sample Size and Data Source In the present study, we have considered a sample of 28 Indian commercial banks. The banks were selected on the basis of size (total asset). The banks included in the study contribute to per cent of the banking sector asset. Thus the sample chosen is fairly representative of the Indian commercial banking sector. Data used in the study have been obtained from the Statistical Tables Relating to Banks in India (R.B.I. publication) of the respective years. V The Results Bank Wise Technical Efficiency Scores and Ranking The bank wise technical efficiency scores and associated ranks are provided in Tables 7 and 8. The technical efficiency scores>1 are marked with asterisks. Table 7: Bank Wise Technical Efficiency Scores Bank Mean State Bank of India 1.061* 1.043* 1.019* 1.069* Allahabad Bank Andhra Bank * Bank of Baroda Bank of India Bank of Maharashtra * Canara Bank Central Bank Corporation Bank Dena Bank Indian Bank * Indian Overseas Bank 1.084* 1.034* Oriental Bank of Commerce Punjab & Sind Bank * 1.261* 1.317* Punjab National Bank * Syndicate Bank UCO Bank * * Union Bank United Bank of India Vijaya Bank Federal Bank HDFC Bank * ICICI Bank 1.398* 1.501* 1.232* 1.910* Indusind Bank 1.036* * 1.123* ING Vysya Bank J&K Bank Karnatak Bank Axis Bank 1.215* 1.261* * 1.087

11 Ranking of Indian Commercial Banks 89 Table 8: Ranking of Observed Commercial Banks Based on Technical Efficiency Scores Bank State Bank of India Allahabad Bank Andhra Bank Bank of Baroda Bank of India Bank of Maharashtra Canara Bank Central Bank Corporation Bank Dena Bank Indian Bank Indian Overseas Bank Oriental Bank of Commerce Punjab & Sind Bank Punjab National Bank Syndicate Bank UCO Bank Union Bank United Bank of India Vijaya Bank Federal Bank HDFC Bank ICICI Bank Indusind Bank ING Vysya Bank J&K Bank Karnatak Bank Axis Bank Descriptive Statistics of Efficiency Scores Table 9 presents the descriptive statistics of the efficiency scores of the observed commercial banks. The results obtained from the exercise indicate gradual improvement in mean technical efficiency scores (excepting ) over the observed years.

12 90 Table 9: Descriptive Statistics of Efficiency Scores Descriptive Statistics Average S.D Maximum Minimum Efficiency Based Classification of Banks We now turn to the technical efficiency score based classification of the observed commercial banks. Of the 28 observed commercial banks, the number of commercial banks found to be super efficient were 5, 7, 6 and 8 for the observed years (refer Table 10). The rest were found to be inefficient. Table 10: Efficiency Based Classification of Observed Banks Bank Group Inefficient Super Ineffi- Super Ineffi- Super Efficient cient Efficient cient Efficient Banks Banks Banks Banks Banks Banks Super Efficient Banks Inefficient Banks Public Sector Banks Private Banks All Table 11 presents mean technical efficiency scores across ownership categories. The results reveal that the observed private sector banks exhibited higher mean technical efficiency relative to the observed public sector banks for all the observed years. Table 11: Mean Technical Efficiencies by Ownership Category Bank Group Public Sector Banks Private Banks All Consistency in Performance Table 12 shows how the efficiency score based ranks of a particular year are correlated to that of another year. The results suggest a high degree of correlation.

13 Ranking of Indian Commercial Banks 91 Table 12: Rank Correlation of the Ranks Obtained by the Observed Commercial Banks Rank Correlation Rank ( ) Rank ( ) Rank ( ) Rank ( ) Rank ( ) Rank ( ) Rank ( ) Rank ( ) Ranking of Observed Commercial Banks According to Mean Technical Efficiency Scores We now rank the observed commercial banks according to the mean of the efficiency scores obtained for the years under study (Table 13). The commercial banks with mean technical efficiency below 0.9 are marked with asterisk. Table 13: Ranking of Observed Commercial Banks According to Mean Technical Efficiency Scores Bank Mean Technical Efficiency Rank ICICI Bank Punjab & Sind Bank Axis Bank Indusind Bank State Bank of India Indian Overseas Bank UCO Bank Indian Bank Bank of Maharashtra HDFC Bank Syndicate Bank Central Bank Dena Bank Punjab National Bank Andhra Bank Bank of India Corporation Bank Allahabad Bank Vijaya Bank Canara Bank Union Bank Oriental Bank of Commerce United Bank of India Federal Bank* ING Vysya Bank* J&K Bank* Bank of Baroda* Karnatak Bank*

14 92 VI Concluding Remarks The present empirical study on asset quality shows that the private sector commercial banks still have slight edge over their public sector counterparts in the matter of technical efficiency, although they are gradually converging. Among the commercial banks under observation, State Bank of India and the ICICI Bank exhibited super-efficiency for all the years under observation. IndusInd bank and Axis Bank and Punjab and Sind bank have also performed well in terms of technical efficiency scores. Reference Andersen, P. and N.C. Petersen (1993), A Procedure for Ranking Efficient Units in Data Envelopment Analysis, Management Science, Vol. 39, pp Banker, R., A. Charnes and W.W. Cooper (1984), Some Models for Estimating Technical and Scale Efficiency, Management Science, Vol. 30, pp Bhattacharyya, A., C.A.K. Lovell and P. Sahay (1997), The Impact of Liberalisation on the Productive Efficiency of Indian Commercial Banks, European Journal of Operational Research, Vol. 98, pp Charnes A., W.W. Cooper and E. Rhodes (1978), Measuring Efficiency of Decision Making Unit, European Journal of Operational Research, pp Das Abhiman, Ashok Nag and Subhash C. Ray (2005), Liberalisation, Ownership and Efficiency in Indian Banking A Non-parametric Analysis, Economic and Political Weekly, March 19. Farrell, M.J. (1957), The Measurement of Productive Efficiency, Journal of The Royal Statistical Society, Series A, General, Vol. 120, No. 3, pp Koopmans, T.C. (1951), An Analysis of Production as an Efficient Combination of Activities, in T.C. Koopmans (Ed.), Activity Analysis of Production and Allocation, New York Cowles Commission for Research in Economics, Monograph No. 13, John Wiley and Sons. Ram Mohan, T.T. and S.C. Ray (2004), Comparing Performances of Public and Private Sector Banks-A Revenue Maximising Approach, Economic and Political Weekly, 20 March, pp Ray, S.C. (2004), Data Envelopment Analysis: Theory and Techniques for Economics and Operations Research, Cambridge University Press, U.K. RBI: Statistical Tables Relating To Banks In India, to , RBI: Trends and Progress of Banking In India, to , Sinha, R.P. (2007), Asset Quality Profile of Indian Commercial Banks: A Stochastic Frontier Approach, ICFAI Journal of Financial Risk Management, September, ICFAI University Press, Hyderabad (2008), Priority Sector Lending of Indian Commercial Banks: Some Empirical Results, Prajnan (Forthcoming), National Institute of Bank Management, Pune, January-March.