Efficiency Performance in Indian Banking—Use of Data Envelopment Analysis Sathya Swaroop Debasish

Performance evaluation of the banking sector in India has assumed primal importance due to intense competition, greater customer demands and changing banking reforms. This study attempts to measure the relative performance of Indian banks over the period 1997–2004 using the output-oriented CRR DEA model. The analysis uses nine input variables and seven output variables. Segmentation of the banking sector in India was done along the following basis: bank assets size, ownership status and years of operation. Overall, the analysis supports the conclusion that foreign owned banks were on average most efficient and that new banks are more efficient that old ones, which are often burdened with old debts. In terms of size, the smaller banks are globally efficient, but large banks are locally efficient. Moreover, this study finds evidence of concentration of efficiency parameters among peer bank groups.

Introduction Data Envelopment Analysis (DEA) has become increasingly popular in measuring efficiency in different national banking industries for it allows comparison of relative efficiency of individual banks and also peer group performance. The most traditional method to benchmark efficiency in the banking sector is the ratio analysis of different financial parameters (like ROA or ROI). However, these ratios give a one dimensional, incomplete picture of the process and

fail to account for the interaction and tradeoff between the various parameters. Apart from the traditional analysis of financial statements of banks, a most common way to tackle the issue is to use an econometric approach to measure various aspects of bank efficiency in a multi-bank environment. DEA has been widely used to measure efficiency performance of different financial institutions like banks, insurance and mutual funds. Particularly in the banking sector, it has been applied to benchmark the performance of different banks or to study the

Sathya Swaroop Debasish is Lecturer (Finance Area), PG Department of Business Management, Fakir Mohan University (UGC-State University), Vyasa Vihar, Balasore-756 019, Orissa. E-mail: [email protected] GLOBAL BUSINESS REVIEW, 7:2 (2006) Sage Publications New Delhi/Thousand Oaks/London DOI: 10.1177/097215090600700209

326 l Sathya Swaroop Debasish

efficiency estimates of different branches of a particular bank. The post-liberalization era in Indian banking has witnessed a host of financial reforms leading to stiff competition among banking units. In recent times, the question of relative comparison of banks by size, type of ownership or date of appearance has been a pertinent issue to reckon with. Performance evaluation becomes more significant to give a true and fair picture of the financial health of an organization. There are various parametric and non-parametric approaches to measure performance. Performance ratios are widely used in all sectors of business.

Objectives of the Study The major objectives of this study are: i) To present a conceptual understanding of DEA. ii) To measure the relative efficiency of banks segmented on the basis of: bank size, ownership structure and new economy/old economy banks. iii) To analyse the trend of concentration among segmented peer groups over the study period and provide reasons thereof. The article is divided into five sections. After an introduction of performance measurement and objectives of the study, the next section includes a brief review of literature on the use of DEA in measuring organizational performance. The DEA model used for the present study is presented in the following section. The sources of data and methodology for the study are addressed

next followed by a discussion of the results and findings. The last section presents concluding remarks and the limitations of the study.

Literature Review DEA is an approach that compares the efficiency of organizational units such as local authority departments, school, hospitals, shops, bank branches and similar instances where there is a relatively homogenous set of units. One of the earliest studies on DEA was conducted by Farell (1957) who attempted to measure the efficiency of production in the single input and output case. Charnes, Cooper and Rhodes (1978) proposed a model that generalizes the single-input, single-output measure of a decision-making unit (DMU) to a multiple-input, multiple-output setting. A DMU is an entity that uses inputs to produce outputs. Since the mid-1980s, DEA has become increasingly popular in measuring efficiency in different national banking industries, for example Ferrier and Lovell 1990; Rangan et al. 1988; Sherman and Gold 1985. Leibenstine and Maital (1992) have argued that DEA is a superior method for measuring overall technical inefficiency. Kraft and Tirtirgolu (1998) used the stochastic frontier function to estimate efficiency of Croatian banks in 1994–95. A comprehensive list of applications of different parametric and nonparametric approaches to study the efficiency of financial institutions has been summarized by Berger and Humphrey (1997). Vassiloglou and Giokas (1990) in their empirical study on operating efficiency of banks

Efficiency Performance in Indian Banking

attempted to draw comparisons between DEA techniques and the log linear model in its utility in analysing the bank branches in India. The study, however, did not compare the performance of various sectors in banking viz. private, public and foreign banks. Robert (2000) used data envelopment analysis for performance measurement using enterprises from 12 different industries operating in European nations. The study highlighted the superiority of the DEA model in measuring overall performance. In the Indian context, Bhattacharya et. al. (1997) examined the impact of liberalization on the efficiency performance in banks using 70 Indian commercial banks during the early stages (1986–1991) of liberalization. This study used the CRR DEA model to calculate the radial efficiency scores and stochastic frontier analysis to attribute variation in the calculated efficiency scores to three sources: a temporal component, an ownership component and random noise component. Very few previous studies have elaborately studied Indian banking having segregated them among peer groups of public sector banks, private banks and foreign banks. Moreover, the present study is more relevant keeping in view the host of banking sector reforms initiated in India during the postliberalization era. This study covering the period 1997–2004 (post-liberalization) attempts to measure the relative efficiency of banks segmented on the basis of: bank size, ownership structure and new economy/old economy banks. An exhaustive list of 93 commercial banks, nine input variables and seven output variables make this study more significant and comprehensive for research purposes.

l

327

Data Envelopment Analysis (DEA) In view of the banking activity as a transformation of a particular set of inputs (e.g., capital, labour and deposits) into a particular set of outputs (e.g., loans and securities), the relative efficiency of banks can be analysed by using DEA. DEA is a non-parametric, deterministic methodology for determining the relatively efficient production frontier, based on empirical data on chosen inputs and outputs of a number of entities, called Decision Making Units (DMUs). In banking, a bank constitutes a DMU. DEA is a linear programming-based technique for measuring the relative performance of organizational units, where the presence of multiple inputs and outputs make a comparison difficult. Any resource used by a unit is included as input. A unit will convert resources to produce outputs so that the outputs include the amount of product or services produced by the unit and these products or services may be produced at different levels of quality. Hence, the outputs may include a range of performance and activity measures. The number of factors selected (inputs and outputs) need to be small compared to the total number of DMUs to strengthen the discrimination power of DEA. Usually, the total number of DMUs should be at least twice the number of inputs plus output factors. Our study used 14 factors or variables (inputs and outputs) and 93 DMUs (banks). From the set of available data, DEA identifies reference points (relatively efficient DMUs) that define the efficient frontier (as the best practice production technology) and evaluate the inefficiency of other, interior

328 l Sathya Swaroop Debasish

points (relatively inefficient DMUs) that are below that frontier. Efficiency is equal to ratio of total sum of ‘weighted outputs’ to total sum of ‘weighted’ inputs. Efficiency = Weighted Sum of ‘Outputs’/ Weighted Sum of ‘Inputs’ One of the basic choices in selecting a DEA model is whether to use input-orientation or an output-orientation. This study employs the CRR model (after Charnes, Cooper and Rhodes 1978) which is an output oriented model where DMUs are deemed to produce the highest possible amount of output with the given amount of input. CRR-model: Charnes, Cooper and Rhodes (CRR) introduced a measure of efficiency for each DMU that is obtained as a maximum of a ratio of weighted outputs to weighted inputs. The weights for the ratio are determined by a restriction that similar ratios for every DMU have to be less than or equal to unity, thus reducing multiple inputs and multiple outputs to single ‘virtual’ input and single ‘virtual’ output without requiring preassigned weights. The efficiency measure is then a function of weights of the ‘virtual’ input-output combination. Formally the efficiency measure for the DMU0 can be calculated by solving the following mathematical programming problem: Max h0 (u,v) = {sΣr = 1 ur yr0 } / {mΣi = 1 vi × r0} ...... Equ. (1) {sΣr=1 ur yrj} / {mΣi=1 vi × ij} ≤ 1, j = 1,2,..... jo,..n ... Equ. (2) ur ≥ 0, r = 1,2,.....s vi ≥ 0, i = 1,2,.....m

where xij = observed amount of input of the ith type of the j th DMU (x ij > 0, 1 = 1,2,...n, j = 1,2....n) and yrj = observed amount of the ith type of the jth DMU (yij > 0, r = 1,2,...s, j = 1,2....n). The variables ur and vi are the weights to be determined by the above programming problem. s is the total number of input variables ( in this study, 9 inputs) and m is the total number of output variables (in this study, 7 outputs). s = 9 and m = 7.

Source of Data and Research Design Ninety-three Indian commercial banks (27 in the public sector, 30 in the private sector and 36 foreign banks) were taken for the study over the period 1997–2004 (seven years). As a matter of simplicity, the data for the financial year 1997–98 is stated as year 1998, 1998–99 as 1999 and so on. The study used 08 (eight) input variables and 06 (six) output variables: The input data are: Input 1 Total deposits received (balance sheet items) Input 2 Total liabilities (working funds + other liabilities) Input 3 Labour related administrative costs (Gross wages) Input 4 Capital related administrative cost (amortization, office maintenance and office supplies, etc.) Input 5 Operating Expenses Input 6 Fixed Assets (balance sheet item) Input 7 Total borrowings Input 8 Net worth Input 9 Net NPA (Non-Performing Assets)

Efficiency Performance in Indian Banking

The output data are: Output Output Output Output Output

1 2 3 4 5

Total loans extended Total investments Net profits Interest and related revenues Non-interest income (commissions for provision of services and related revenues) Output 6 Short-term securities issued by official sectors (CNB bills and MOF treasury bills) Output 7 NIM (Net Interest margin) As a statistical basis for input and output data, both end-of-year balance sheets and financial statements of the 93 commercial banks were used. The relevant data for the commercial banks was obtained from the CMIE database. All the 93 banks (as shown in the Appendix) were segregated into various peer groups on the basis of: a) BANK SIZE (TOTAL ASSETS): Large (Avg.Assets > Rs 100 crores) Medium (Avg. assets between Rs 10 cr.–Rs.100 cr.) Small (Avg.Assets < Rs 10 crores) b) OWNERSHIP: Public, Private, Foreign c) YEARS OF OPERATION: Old-Economy (Established in yr. 1990 or before) New-Economy (Established in yr. 1991 or after)

l

329

The CRR output model (output maximation) was utilized to evaluate the data. DEA attempts to find out which of the decisionmaking units (DMUS) determine an envelopment surface, when considering inputs and outputs. Units that determine the envelopment surface are deemed to be efficient and units that do not lie on the surface are defined as inefficient. Banxia’s Frontier Analyst Software was used to compute the relative efficiencies of the DMUs for each year under various segments of the study.

Empirical Findings The Indian commercial banks are classified into peer groups (according to the size of their total assets, ownership status and years of operation) and the results are presented separately for each group. The composition of bank peer groups classified by assets size (Table 1) changed significantly over the study period. The number of smaller banks decreased from 24 in 1998 to 15 in 2004. Large banks which totalled 42 in 1998 increased to 54 in 2004. Table 1 Average Efficiency of Banks Grouped by Size Year 1998 1999 2000 2001 2002 2003 2004

Large 0.65 0.67 0.69 0.73 0.77 0.79 0.86

(42) (44) (44) (48) (49) (52) (54)

Medium 0.56 0.59 0.61 0.62 0.60 0.57 0.52

(27) (28) (27) (25) (25) (24) (24)

Small 0.82 0.77 0.74 0.70 0.66 0.65 0.63

(24) (21) (22) (20) (19) (17) (15)

Source: DEA results obtained from the Frontier Analyst Software. Note: Figures in parenthesis give the number of banks classified under each peer group for the various years.

330 l Sathya Swaroop Debasish

The efficiency of smaller banks declined from 0.82 in 1998 to 0.63 in 2004. Most of the sample banks were classified as large banks under the basis of asset size. Larger banks showed a complete reverse trend with their peak efficiency of 0.86 in later years (2004) and showed a steady increase in efficiency from 1998 to 2004. The efficiency of large banks were higher than smaller banks during 2000–2004.This was perhaps because in the beginning of the study period, the large banks were overstaffed and burdened with huge net NPAs (Non-performing assets) inherited from the previous system. The average efficiency of medium (classified) banks did not show any consistent upward or downward trend and the range was between 0.52–0.62 during 1997–2004, although their number marginally decreased from 27 to 24. Most of the banks showed better performance indicated by efficiency scores in later years which may be due to widespread consolidation, cost cutting and strategic investments in the banking sector which had a positive effect on the bottom line (net interest margin). Public sector banks were the least efficient with values in the range of 0.44 to 0.54. This figure is quite low when compared against the findings of Berger and Humphrey (1997) that the range of efficiency for Indian banks is 0.75 to 0.86. This is also in coherence with previous results (Table 1), since three out of every five public banks are large banks. Table 2 shows the average efficiency figures of banks groups segregated under ownership status. The basis for grouping was the dominant type of ownership, thus classifying the banks with more than 50 per cent of their capital in government hands as public sector banks, with the same principle applied to private domestic and foreign-owned banks.

Table 2 Average Efficiency of Banks Grouped by Ownership Status Year 1998 1999 2000 2001 2002 2003 2004

Public 0.46 0.44 0.51 0.52 0.49 0.53 0.50

(27) (27) (27) (27) (27) (27) (27)

Private 0.69 0.63 0.78 0.65 0.67 0.74 0.77

(30) (30) (30) (30) (30) (30) (30)

Foreign 0.69 0.63 0.65 0.71 0.75 0.77 0.81

(36) (36) (36) (36) (36) (36) (36)

Source: Same as Table 1. Note: Figures in parenthesis give the number of banks classified under each peer group for the various years.

Relatively, the foreign banks were found to be more efficient relative to private banks for most of the years. The highest efficiency for the foreign banks group was 0.81 in 2004 and least in 1999 (0.63). On the other hand, the private banks group had peak efficiency of 0.78 in 2000 and least efficiency of 0.63 in 1999. This can be partly attributed to the spate of merger activities in private domestic banking, the predominant being HDFCTimes Bank, UTI-GTB and ICICI Bank-Bank of Madura. Other factors for this transformation are: technological advancements in banking, increased foreign investment, stable stock markets and failure of public sector banks to provide better services in the class banking customer segment. Table 3 shows the average efficiency of new and old banks over the period 1998–2004. New economy banks were found to be more efficient than the old economy banks. The average efficiency of new banks increased from 0.72 in 1998 to peak efficiency of 0.89 in 2002 and declined to 0.86 in 2004. The range of efficiency for old banks was 0.52 to 0.98, which is much below the average figure for Indian banks (0.75–0.86) shown in the study

Efficiency Performance in Indian Banking

Table 3 Average Efficiency of Banks grouped by Years of Operation Old-economy Banks

1998 1999 2000 2001 2002 2003 2004

0.57 0.56 0.52 0.63 0.65 0.61 0.69

New-economy Banks

(57) (57) (57) (57) (57) (57) (57)

0.72 0.76 0.77 0.85 0.89 0.82 0.86

331

needed only 44.5 per cent of its inputs currently being used (or in terms of average inefficiency, it would have needed 124.6 per cent more inputs to produce the same outputs as an efficient bank. About fourth-fifths of the DMUs were found to stay in the interval (I) of within one σ from the average efficiency in 1998 which declined marginally to 61.29 per cent by 2002. This was partly due to the decreasing value of σ over the years, showing greater stability in performance and a trend of concentration among the segmented peer banks group. The third objective of the study is thus analysed for evidence. The proportion of DMUs in interval (I) increased from 61.29 per cent (2002) to 75.22 per cent in 2004, thus showing a better trend in later years. The number of banks in the public sector becoming efficient increased from two in 1998 to seven in 2004. The private banking segment witnessed phenomenal improvement by increasing its efficient banks from three in 1998 to 19 in 2004. This is a fair indication of the supremacy and financial hegemony maintained by domestic private banks in the Indian scenario when it comes to efficiency and performance measurement. In the foreign banking sector, the proportion of DMUs in the interval (I) increased from

by Berger and Humphrey (1997). Particular problems for old, as well as public banks were high operating costs, legacy of accumulated establishment expenses, non-performing assets and inability to quickly and economically adopt technology banking.

Year

l

(36) (36) (36) (36) (36) (36) (36)

Source: Same as Table 1. Note: Figures in parenthesis give the number of banks classified under each peer group for the various years.

Table 4 shows the summary results for average efficiency in Indian banking. In 1998, only 10 (out of 93) banks were efficient with an average efficiency of only 0.445 which means that the average bank, if producing its outputs on the efficiency frontier instead of at its current virtual location, would have

Table 4 Summary Results of Average Efficiency (CRR Model) Year

No. of DMUs

PUB

1998 1999 2000 2001 2002 2003 2004

93 93 93 93 93 93 93

2 1 2 3 4 5 7

No. of Efficient DMUs PVT FOR ALL 3 6 5 7 11 12 19

5 5 10 10 7 10 6

10 12 17 20 22 27 32

Average Efficiency (M) 0.445 0.513 0.492 0.561 0.673 0.743 0.779

Standard Interval % of DMUs Deviation (σ) I = (M – σ; M + σ) in Interval (I) 0.282 0.269 0.267 0.232 0.189 0.168 0.159

(0.163; (0.244; (0.225; (0.329; (0.484; (0.575; (0.620;

0.727) 0.782) 0.759) 0.793) 0.862) 0.911) 0.938)

80.61 76.32 70.96 66.67 61.29 73.14 75.22

% % % % % % %

Source: Same as Table 1. Note: PUB—Public sector banks; PVT—Private sector banks; FOR—Foreign banks; ALL—All 93 banks together.

332 l Sathya Swaroop Debasish

five banks becoming efficient in 1998 to 10 banks in 2001 and this declined to just six banks in 2004. Thus, the Indian banking sector witnessed large asymmetry between banks regarding average efficiency.

Concluding Remarks This study tried to measure the relative performance of Indian banks over the period 1997–2004 using the output-oriented CRR DEA model. Segmentation of the banking sector in India was done on the following basis: bank assets size, ownership status and years of operation. Foreign-owned banks were on average found to be most efficient. It was also found that the new banks are more efficient than the old ones, which are often burdened with old debts. Further, both new economy and old economy banks showed an increasing trend in average efficiency although relatively the former ones are more efficient. Again large-sized and

small-sized banks were relatively more efficient than the medium-sized banks over the study period. Since about four-fifths of the banks (DMUs) were found to lie in interval (I), this evidences the existence of ‘concentration’ among the peer banks group. The smallest banks are often niche banks. Being a small bank, however, does not guarantee relative efficiency, as the coefficient of variation of efficiency scores in that group is the highest and as a number of banks from that group have failed in the recent past. The current study does not include a few of the important banking efficiency parameters like non-performing asset, capital adequacy figures, customer satisfaction index and other service quality variables. Although DEA has fewer limitations than other performance measurements approached in the choice of input and output variables, the efficiency measure obtained by DEA is sensitive to the combination of inputs and outputs.

REFERENCES Berger, A.N. and D.D. Humphrey. 1997. ‘Efficiency of Financial Institutions: International Survey and Directions for Future Research’, European Journal of Operational Research, 98: 175–212. Bhattacharya, A., C.A.K. Lovell and P. Sahay. 1997. ‘The Impact of Liberalization on the Productive Efficiency of Indian Commercial Banks,’ European Journal of Operational Research, 98: 332–45. Charnes, A., W.W. Cooper and E. Rhodes. 1978. ‘Measuring the Efficiency of Decision Making Units’, European Journal of Operation Research, 2: 427–44. Farell, P. 1957. ‘DEA in Production Center: An Inputoutput Model’, Journal of Econometrics, 3: 23–49. Ferrier, G.D. and C.A.K. Lovell. 1999. ‘Measuring Cost Efficiency in Banking: Econometric and Linear

Programming Evidence’, Journal of Econometrics, 46 (1): 229–45. Kraft E. and D. Tirtigolu. 1980. ‘Bank Efficiency in Croatia: A Stochastic Frontier Analysis’, Journal of Compatarive Economics, 26: 282–300. Leibenstine, K. and R. Maital. 1992. ‘Allocative Efficiency vs. X-efficiency’, American Economic Review, 56 (2): 392–415. Rangan, N., R. Grabowski, H.Y. Aly and C. Paruska. 1988. ‘The Technical Efficiency of US Banks’, Economic Letters, 28: 169–75. Robert, Dyson. 2000. ‘Performance Measurement and Data Envelopment Analysis’, Operational Research Insight, 13 (4): 3–8.

Efficiency Performance in Indian Banking Sherman, H.D. and F. Gold. 1985. ‘Bank Branch Operating Efficiency—Evaluation with Data Envelopment Analysis’, Journal of Banking and Finance, 9: 297–315.

l

333

Vassiloglou, M. and D. Giokas. 1990. ‘A Study of the Relative Efficiency of Bank Branches: An Application of Data Envelopment Analysis’, Journal of the Operational Research Society, 41 (7): 14–26.

APPENDIX List of Sample Commercial Banks (93 Banks) Public Sector Bank (27) 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

Allahabad Bank Andhra Bank Bank of Baroda Bank of India Bank of Maharashtra Canara Bank Central Bank of India 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 of India United Bank of India Vijaya Bank State Bank of India State Bank of Hyderabad State Bank of Indore State Bank of Mysore State Bank of Patiala State Bank of Saurashtra State Bank of Trancore State Bank of Bikaner & Jaipur

Private Sector Banks (30) 28. Bharat Overseas Bank Ltd 29. City Union Bank Ltd 30. Development Credit Bank Ltd 31. Lord Krishna Bank Ltd 32. SBI Coml. and Intl. Bank Ltd 33. Tamiland Mercantile Bank Ltd 34. The Bank of Rajasthan Ltd 35. The Catholic Syrian Bank Ltd 36. The Dhanalakhsmi Bank Ltd 37. The Federal Bank Ltd 38. The Ganesh Bank of Kurundwad Ltd 39. The Jammu & Kashmir Bank Ltd 40. The Karnataka Bank Ltd 41. The Karur Vysya Bank Ltd 42. The Lakshmi Vilas Bank Ltd 43. The Nainital Bank Ltd 44. The Nedungadi Bank Ltd 45. The Ratankar Bank Ltd 46. The Sangli Bank Ltd 47. The South Indian Bank Ltd 48. The United Western Bank Ltd 49. The Vysysa Bank Ltd 50. Bank of Punjab Ltd 51. Centurion Bank Ltd 52. Global Trust Bank Ltd 53. HDFC Bank Ltd 54. ICICI Bank Ltd 55. IDBI Bank Ltd 56. IndusInd Bank Ltd 57. UTI Bank Ltd

Foreign Banks (36) 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86. 87. 88. 89. 90. 91. 92. 93.

ABN-Amero Bank N.V. Bank Internasional Indonesia Bank Muscat SAOG Ban of America NA Bank of Bahrain and Kuwait BSC Bank of Ceylon Barclays Bank PLC BNP Paribas Chinatrust Commercial Bank Chohung Bank Citibank N.A. Commerzbank AG Credit Agricole Indosuez Credit Lyonnais Deustsche Bank AG Dresdner Bank AG Ing Bank JP Morgan Chase Bank KBC Bank N.V. Krung Thai Bank Public Company Ltd MashreqBank psb MIZUHO Corporate Bank Ltd Oman International Bank Ltd Oversea-Chinese Banking Corporation Ltd Science Generale Sonali Bank Standard Chartered Bank Standard Chartered Grindlays Bank Ltd State Bank of Mauritius Ltd Sumitomo Mitsui Banking Corporation The Bank of Nova Scotia The Bank of Tokyo The Development Bank of Singapore Ltd HSBC The Saim Commercial bank PCL The Toronto Dominion Bank