Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 Comparative Technical Efficiency of the Bangladesh Banking Industry Abdus Samad This paper estimates the technical efficiency of the banking industry in Bangladesh during 2010. Applying the time invariant stochastic frontier function the paper finds that the mean technical efficiency of the Bangladesh banking industry was 69.5 percent. The paper also finds that 63 percent of the total banks operate above the mean efficiency level. The estimates of comparative efficiencies for the five groups of nationalized banks, private domestic banks, foreign banks, specialized banks, and nonscheduled banks find that the mean efficiency of private domestic banks is 72 percent and it ranks the highest in efficiency level. The nationalized banks’ mean is 70 percent and it ranks second in efficiency. Foreign banks have the lowest efficiency in loans and advances. JEL Classification: G20, G21 and C33 Keywords: Bangladesh, Efficiency, Banking Industry, Stochastic Frontier. I. Introduction Banks operate in asymmetric information which leads to adverse selection and moral hazard. Given asymmetric information, some banks perform well above the average efficiency level while other performs below the average efficiency. Determining bank efficiency and relative efficiency is a key element in a competitive market for improving the bank efficiency level. The allocation of bank scarce resources for the improvement of its efficiency is worthwhile only when its current efficiency level is known. So, the study of the estimates of efficiency level is important. The study of efficiency is also important for bank customers, bank managements, and bank regulators in determining a bank position relative to its peer group. Since Bangladesh opened its liberalization policy in 1982, there are now fifty-one banks operating in Bangladesh. Among the fifty-one banks, there are thirty domestic banks, eleven foreign banks, four specialized banks, and four non-scheduled banks, in addition to four nationalized banks. It is important to know the average efficiency level of the Bangladesh banking industry as well as the five groups’ relative efficiency position before making any resource allocation for their improvement. The Bangladesh government laid great emphasis on efficient performance. In its effort, the government formed a national commission on money, banking, and credit in 1986 for the efficient operation and management of the banking system. ____________________________________________________________________________ Abdus Samad, Ph.D., Associate Professor, Department of Finance and Economics, Utah Valley University, 800 W. University PKY, Orem, UT 84058, USA, Phone: 801-863-8368, Fax: 801-863-7218, Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 A current survey of literature shows no record of the comparative efficiency study for the Bangladesh banking industry since the last study of Samad (2007) for the period of 2000. The importance of the estimates of efficiency and the absence of the comparative efficiency study in the Bangladesh banking industry constitutes the key motivating factor for this study. The paper is organized as: Section II provides a brief survey of literature. Data, methodology and the description of mode are discussed in Section III. Empirical results and conclusions are provided in Section IV. II. Survey of Literature Bank efficiency studies are large; however, a large volume of these studies focused on the banks of developed countries, USA and Europe in particular. Berger and Humphrey (1997), Berger and Mester (1997) and Bar et al (1999).There are few bank efficiency studies for less developed countries. Samad (2009) estimated the technical efficiencies of the Bangladesh banking industries during the year 2000 and found that the average efficiency of banks was 69.6 and the average efficiency lies between 12.7 and 94.7 during 2000. This study is different from the previous studies with respect to outputs and input. Compared to loans as output, deposits and employee expenses as inputs in my previous studies, this study uses labor, capital, and deposits as inputs for loans and advances as outputs. Samad (2010) estimated the technical efficiency of just one bank, the Grameen Bank, the symbol of micro-financing founded by Nobel Laurate, Professor Younus. There are several studies in India. Some of these studies used bank financial indicators for measuring bank efficiency. Rammohan and Roy (2004) and Sarkar et al (1998) used financial indicators for measuring bank efficiency. Rammohan and Roy found that public sector banks are more efficient than the private sector banks. In Kumbhakar and Sarkar’s (2003) cost efficiency approach for measuring bank efficiency, they found that private sector banks improved their performance compared to public sector banks. There are other studies in India that used different approaches for measuring bank efficiency. Saha and Ravishankar (2000), Bhattacharyya et al (1997), and Sanjeev (2006) measured the efficiency Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 of Indian Banks using the DEA approach. Bhattacharyya et al (1997) found that the public sector banks were the best performing banks in India during the late 1980s and early 1990s. Shanmugam and Das (2004) used stochastic frontier analysis (SFA) for measuring technical efficiencies of Indian commercial banks. They found that the state bank group and foreign banks were more efficient than their counterparts during 1992-1999. Andries and Cocris (2010) analyzed the comparative efficiency of the main banks in Romania, the Czech Republic, and Hungary during the 2000-2006 period using both data envelopment analysis (DEA) and stochastic frontier approach (SFA). They found that the banks in these three countries operate at a low level of technical efficiency. III Methodology This paper employs the time invariant model of stochastic frontier production function for the measuring technical efficiencies (TE) of banks. The actual production function of a bank can be written as: Qit = f(Xit, β) exp(-uit); uit ; i= 1, 2, ----n, t = 1, 2, -------T (1) Where Qit = actual output of sample bank i in period t; Xit is a vector of inputs used by bank, and β is a vector of parameters Uit is one sided (non-negative) residual. If the performance of a bank is inefficient, its actual output is less than its potential output. (In other words, if a bank is efficient, its output is equal to its potential output). Therefore, TE is the ratio of the actual output Qit to potential output of a bank in period t as: TEi = exp( xiB ď€ ui Qit = = exp(-ui). exp( xiB ) exp( xiB ) (2) When the residual term (uit) is zero, the bank produces the potential output and the bank is fully TE. When (uit) >0, the bank produces less than its potential output. If the bank is less than full TE, it operates below the production frontier. Thus, uit is a bank’s TE and it is inversely (negatively) related. Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 For capturing the effects of omitted variables (i.e. measurement errors), a random error, it is included in (1) which gives: Qit = f(Xit, β) exp( it-uit) Where I (3) is a random error and is assumed to be iid (independent and identically distributed) as N(0,σv2) and independent of ui which represents technical efficiency/inefficiency. Battese and Coelli (1992), proposed a model in which the parameters of technical inefficiency (uith) are estimated with the method of maximum likelihood and can be written as: Uith = {exp[ - (t – T)]}ui ui ‘s (4) are non-negative random variable and ui ( , ), independently and identically distributed as truncated normal with mean Where = ). The value of = it N( ). and variance, That is, ui is . . It is the estimate of the ratio of standard deviation of inefficiency to the standard deviation of random error. It explains the variation of inefficiency as a percentage of error term. = explains inefficiency as a percentage of total deviation of output. Whether there is no technical inefficiency component in the model, it is tested by the null hypothesis H0: = 0 against the alternative hypothesis, Ha: > 0. It is tested by standard LR. Data and Model The measure of bank output (production) is controversial in banking literature mainly because a bank provides a variety of services such as issue of deposits, loans, and discounts. Hence, there is no consensus over what banks produce (i.e. output) and what are banks’ inputs (i.e. resources) used to produce outputs. There are two approaches which most researchers use in banking studies. They are Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 either the production approach or the intermediate approach. Berger, et al (1992) provides detailed discussions about it. According to the intermediate approach, a bank collects deposits with labor and capital and intermediate these resources into loans, other assets, and incomes. Deposit is, thus, considered as input and loans and advances, and investments are defined as output. According to the production approach, banks use labor and capital including other financial inputs to produce outputs. The outputs are services that a bank provides to the account holders of deposits and advances (Ferrier and Lovell, 1990). Thus, production approach measures outputs by the number of accounts (deposits and advances) that a bank generates with cost. Since the numbers of accounts are hardly available, intermediate approach is widely used. This paper uses the intermediate approach in determining bank inputs and outputs. That is, loans and advances are used as a banks’ output (Q). These outputs are generated by applying three inputs such as deposits, labor, and capital. The total value of a bank’s capital is estimated as total assets minus total liabilities and is denoted as capital (K). The total full-time and part-time number of employees is considered as labor (L). Demand deposits and fixed deposits are considered as deposits (D). Data of all variables such as loan and advances (Q), deposits (D), total employee (L), and capital (K) are obtained from Banks and Financial Institutions Activities 2010, Finance Department, Finance Ministry, and Peoples’ Republic of Bangaldesh, This paper estimates the following Cobb-Douglas production function as: ln(Qit) = β0 +β1ln(Kit) + β2ln(L it) + Vit –Uit (5) Where Q is the output represented by total loans and investments (LNADV), K is capital, L is labor, and ln is a natural log of all variables. Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 IV Empirical Results Result of the ML estimation for equation (7) is presented in Table 1. Stochastic frontier model Number of obs = Wald chi2(3) Log likelihood = -26.272368 = 51 285.43 Prob > chi2 = 0.0000 -----------------------------------------------------------------------------LogLNADV | Coef. Std. Err. z P>|z| [95% Conf. Interval] -------------+---------------------------------------------------------------LogDep | .040128 9.73 0.000 .3118576 .4691564 LogK | .0938308 .0330951 2.84 0.005 .0289655 .158696 4.64 0.000 .1513291 .3725844 8.81 0.000 3.200311 5.031349 LogL .390507 | .2619567 .0564437 _cons | 4.11583 .4671101 -------------+---------------------------------------------------------------/lnsig2v | -2.657581 .5317582 -5.00 0.000 -3.699808 -1.615354 /lnsig2u | -1.292396 -2.63 0.009 -2.256629 -.3281619 .491965 -------------+---------------------------------------------------------------- λ .2647974 .0704041 .1572523 .4458928 .5240345 .1289033 .3235781 .8486733 .3447298 .1127478 .1237483 .5657113 1.979002 .1845487 1.617293 2.34071 -----------------------------------------------------------------------------Likelihood-ratio test of sigma_u=0: chibar2(01) = 1.41 Prob>=chibar2 = 0.108 All inputs—deposits, capital and labor—have a positive impact on production as shown in Table 1. The low p value, 0.000, 0.005, and 0.000 for the coefficients of deposits, capital, and labor suggest that they are significant factors for bank production. Among the inputs, deposits and labor are dominant Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 factors for the production of loans and advances. The positive coefficient for deposit, capital, and labor indicates they affect production of loans and advances positively. The variance of inefficiency, σ2=0.52, in Table 1, indicates that 52 percent of the variation of bank output is due to technical inefficiency and is statistically significant. The variance of random error measured by σ2v= 0.26 indicates that 26 percent of variation of output is due to model’s measurement error and is statistically insignificant. A test on the significance of technical inefficiency has been done. A test on the significance of technical inefficiency measured by random variable, ut, is obtained from LR ratio of σu. The LR value has approximately đťś’2 distribution with parameter shown in Table 1. LR = -26.27 and is significant. The significant is provided by the probability of Wald đťś’2> is 0.0000. This means that the inefficiency effect as measured by uit is significantly different from zero. Thus, H0: σ 2ui=0 is rejected which suggests Ha: σ 2ui>0. Table Descriptive Statistics of Estimated Efficiency of Bangladesh Banking Industry Observation Mean Median Maximum Minimum S.D. 51 0.720 0.925 0.268 0.14 0.695 Jarque Bera 15.40 2 Probility 0.000 Table 2 shows that the average technical efficiency of the Bangladesh banking industry is 69.5 percent with a median efficiency of 72 percent. The result is consistent with the result of the previous study (Samad, 2009. The range of estimated efficiency—the difference between the maximum and minimum efficiencies—is 0.657 and is very high. The 0.000 probability of Jarque-Bera suggests that the distribution of efficiency is normally distributed. Table Frequency Distribution of Estimated Efficiencies of Banking Industry Range efficiencies 0.0 – 0.3 0.3 – 0.5 of Frequency 1 5 Relative Frequency Cumulative Relative Frequency 0.019 0.098 0.01 0.117 3 Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 0.5 – 0.7 0.7 – 0.9 0.9 and above Total 13 30 2 51 0.255 0.588 0.04 1.00 0.372 0.96 100 Table 3 shows that 37.2 percent of the total banks of Bangladesh operated below the industry mean efficiency of 70 percent and 63 percent of banks in Bangladesh operated above the banking industry average during 2010. The results are very much in consistence with the findings of the previous study (Samad, 2009) Table Comparative Technical Efficiency of Bangladesh Banks Bank Group Nationalized Banks Private Banks Foreign Banks Specialized Banks Non-scheduled Banks Mean 0.70 0.72 0.59 0.60 0.62 Median 0.71 0.72 0.60 0.62 0.62 4 Maximum 0.80 0.83 0.82 0.75 0.91 Minimum 0.60 0.39 0.26 0.55 0.35 S.D. 0.08 0.09 0.20 0.09 0.28 Table 4 shows that among the four groups of banks, the mean efficiency of private banks is 72 percent and it is the highest level of efficiency in Bangladesh. The nationalized banks rank second with an average efficiency level of 70 percent. The foreign banks have the lowest level of efficiency in loans and advances production. The mean efficiency of foreign banks is 59 percent. Conclusions: The technical efficiency of the Bangladesh banking industry is estimated during 2010 using the stochastic frontier production function. The estimates find that the mean efficiency of the banking industry is 69.5 percent. About 37 percent of the total banks operate below the mean efficiency level and 63 percent of Bangladesh banks operate above the mean efficiency level. Results are consistant with the findings of the previous study (Samad, 2009) Among the four groups of banks, the efficiency level of private commercial banks is the highest. Private bank mean efficiency is 72 percent. Nationalized banks stand second. Their efficiency level is 70 percent. The lowest efficiency level is observed for foreign banks. The foreign bank efficiency level is 59 percent. References Berg, S.A., Forsund, F., Hjalmarsson, L. and Suominen, M. 1993. ―Banking efficiency in Nordic countries‖, Journal of Banking and Finance, 17, 371-388. Proceedings of 11th Asian Business Research Conference 26-27 December, 2014, BIAM Foundation, Dhaka, Bangladesh, ISBN: 978-1-922069-68-9 Samad, Abdus. 2009. ―Measurement of Inefficiencies in Bangladesh Banking Industry Using Stochastic Frontier Production Function‖ Global Journal of Business and Research 3 (1,) 41-48. Samad, Abdus (2010). “Estimate of Production efficiency: Evidence from Grameen Bank Microfinancing‖ Global Review of Business and Economic Research, 6 (2), 183-189 Ram Mohan, T.T., Roy, S.C. 2004. ―Comparing performance of Public and Private Sector banks: A Revenue Maximization Efficiency Approach‖, Economic and Political Weekly, 39,(12), 37-48?. Sarkar, J., Sarkar, S. and Bhaumik, S.K. 1998. ―Does ownership always matter? Evidence from Indian banking Industry‖, Journal of comparative economics, 26, 213-27?. Kumbhakar, S.C., and Sarkar, S. 2003. ―Deregulation, Ownership, and Productivity Growth in the banking Industry: Evidence from India‖, Journal of Money, Credit and banking,35 (3), 403-424. Saha, S., and Ravishankar,T.S. 2000. ―Rating of Indian commercial Banks: A DEA approach‖, European Journal of Operations Research, 124, 312 -320?. Bhattacharyya, A., Lovell, C.A.K., Sahay, P. 1997. ―The Impact of Liberalization on Productive Efficiency of Indian Commercial Banks‖, European Journal of Operations Research, 98, 113 -127?. Sanjeev, G. M. 2006. ―Data Envelopment Analysis (DEA) for measuring Technical Efficiency of Banks‖, Journal of Business Perspective, 10 (1), 97- 113?. Shanmugam, K.R., and Das, A. 2004. ―Efficiency of Indian Commercial Banks during the reform period‖, Applied Financial Economics, 14 (June), 681-686.